We entered the past year with a treasury diversification and formalized operations that represented these ambitions. Initially, we focused on building out Neighborhood Zero as an embassy for DAOs. We accomplished an incredible breadth of divergent explorations, including retreats, residencies, build weeks, conferences, three in-house software products, and some of the best content about DAOs on the internet. 

As the market corrected over the course of the year, we focused on what was most important: building coliving neighborhoods to form the backbone of our network city. We completed top-notch coliving experiences at potential Cabin neighborhoods, which has given us confidence to focus on building coliving for deeply rooted communities in beautiful natural locations.

We came together to build our city from the ground up. We grew Neighborhood Zero from a single structure to a compound featuring community-built amenities like a sauna, plunge pool, bath house, outdoor gym, pergola, fire pit, park, and tiny cabin. We gathered in California and Puerto Rico to build out new cabin neighborhoods and planted the seeds of a strong culture of people who find joy in creating a built environment designed for communities living in nature.

And, of course, we got together and vibed. From DAO Camp to dinner parties, we grew our culture through meaningful IRL time together.

Through conversations across the community, we’ve identified several opportunities going into 2023. Our community wants clearer ways to engage IRL and URL, including:

1. more experiences at neighborhoods

2. clearer roles within the community

3. better ways to earn and use ₡ABIN

4. opportunities to start and build neighborhoods

We’ve synthesized this community input into an ambitious new roadmap for Cabin’s next year. Our focus and execution has put us in a position to dream big for 2023, and we are thinking about some wild ideas that feel closer at hand than expected. These are themes that feel alive for 2023:

10x cheaper and better housing

New cities will need to solve a problem current cities are failing at: providing affordable housing. At Cabin, we believe that by building coliving communities in nature, we can provide 10x cheaper and better housing options for people. We have a long way to go to accomplish this mission, but we’re taking big steps in the right direction.

First, we are now offering free housing via work-stay exchange programs for each of our roles. If you’re a builder, naturalist, host, or creator who wants your housing expenses covered: check out our programs here. We are excited to continue exploring these types of definancialized transactions, which have potent intrinsic value that remains largely unrecognized in modern economies.

Second, we completed our first low-cost housing prototype, the Tiny Cabin. For $25k we have a beautifully designed studio unit by our Builder in Residence, Charlie Frise. Partially prefabbed and finished on site, it can be installed easily anywhere in the United States. In 2023, we will continue developing prototypes and building more housing. 

Third, we will continue to bring the community together for Build Weeks, where we deepen our culture by creating permanent improvements at neighborhoods. We flex our build skills, learn from working together, find satisfaction in our shared results, and build deep relationships with each other and neighborhoods. Our next Build & Grow Week is in the Portuguese Azores (!) and we’ll have a slate of them throughout the year.

If we keep playing intrinsically motivated games through shared experiences, pushing the limits of housing cost reduction, and exploring definancialized forms of exchange, we can start to build the basis of a new kind of economy. If it works, it will be an economy very different from our current one. It won’t produce the exact same externally fungible value that our current economy is designed for. But we hope and expect immense value will emerge. 

Choose Your Character

Cabin is starting to imagine the first civilization-scale game that is played both online and offline. This IRL RPG (in-real-life role playing game) is a wild experiment in bringing together the best and brightest of the internet in-person to solve some of our biggest problems through novel, emergent bottom-up human coordination. This game starts with you selecting a role. There are six of them to start:

1. Caretaker: Operator of a neighborhood. Jack-of-all-trades responsible for a space.

2. Builder: Maker of physical things that improve neighborhood built environments. 

3. Host: Space-maker in residence, building containers and culture.

4. Naturalist: Grower of plants, animals, and human systems to support nature.

5. Creator: Very-online producer of art, novelty, content, and experiments.

6. Resident: Remote worker living across the neighborhood network.

Our onboarding process will help new members find the right path into each role. Roles will have a guild to connect with like-minded members, levels you can earn, and work-stay exchange programs at neighborhoods. We’re building a directory of member profiles around roles, stamps, and other on-chain data from our community. Here’s an alpha leak on some early design explorations of the Cabin passport:

Cabin passports are the profile you use to play games with us, IRL and URL. We’ve already begun distributing stamps, roles, and levels to community members as NFTs. These tokens of identity, reputation, and participation get added to your passport and serve as a collective vibe check. But creating your profile is just the start of the games. 

In 2023, we’re kicking off our biggest game yet. It’s called the Neighborhood Race, and while it takes place online, its impact is IRL. 

The Neighborhood Race

We believe that network cities will grow by crowdfunding new neighborhoods. 

Crowdfunding is a useful mechanism, because it creates a focused pool of people with skin-in-the-game towards a specific goal. The original creator residency program taught us that every crowdfund completed on-chain has the starting ingredients of a DAO. It’s no surprise that crowdfunding territory is the core action completed by Balaji’s definition of network states—they are potent coordination mechanisms that can be harnessed to create impact in the real world. 

The Neighborhood Race is how the Cabin community will come together to select and support new neighborhoods. Members of the community can pledge support and make contributions. Winners of the race get access to crowdfunding tools, advising, and matching funds from Cabin. 

In 2023, we’ll run our first race. If it goes well, we’d love to turn the race into a recurring event. We’re making a big bet on this game, because we believe it could be a factory for local, self-sovereign, land-centric organizations. These are exactly the kind of organizations we hope will thrive in a regenerative solarpunk future. 

Political polarization has become so entrenched that it is hard to see the path out. While the temperature of political discourse has turned up dramatically in the past few years, it’s a phenomenon that has been growing for decades. Here’s a visualization of polarization in the US House of Representatives, 1949-2011:

Suffice it to say: the picture has not gotten prettier since 2011 and it’s increasingly unclear how we can break this cycle. We need a new paradigm of political philosophy that steps outside of the well-worn troughs we find ourselves stuck in.

Those Right and Left wheels of democratic society are deeply entrenched in ruts caused by two foundational works of political theory: Hobbes and Rousseau. If we backtrack to the origins of Right and Left, we can better understand how modern technologies and ancient social practices offer a more compelling future that throws out the Right/Left divide in favor of a focus on supporting a wider range of small, local, diverse, and decentralized governance structures.

A short refresher on Political Philosophy 101

The grandaddy of the Right is Hobbes, who espoused the need for centralized sovereign leadership in order to avoid “the war of all against all” in an imagined state of human nature. You can skip the fine print and get the basic gist of his perspective by admiring the original cover etching for Leviathan, which Hobbes helped design himself. It features a giant white dude with a crown and a dashing mustache—scepter of governance in one hand, sword of war in the other—a sovereign body, literally made of people, paternally watching over the city of civilization:

In contrast, the Left looks to Rousseau to define a social contract for humanity. Rousseau’s Social Contract provided, at the time, a radical new perspective: sovereignty was not just a top-down phenomenon, but a bottom-up one. People could devise methods of self-governance that replaced a traditional all-powerful leviathan ruler with a system of government that derived legitimacy from the people.

It’s easy to draw analogies along this well-worn spectrum: Hobbes/Rousseau, Right/Left, autocracy/democracy, centralization/decentralization. But this dichotomy blinds us to other ways of thinking about self-governance.

Breaking the dichotomy

While Rousseau’s Social Contract developed the crucial principle of self-governance, it disappoints in the same way the modern Left tends to disappoint: glimmers of a hopeful alternative, ultimately undermined by complicated bureaucratic abstractions.

Rousseau has long been viewed as the grand alternative to Hobbes, but a modern reading of his works leaves the reader feeling like Rousseau is just rearranging furniture in Hobbes’ house. The problem is that both men came from a time and place deeply steeped in the idea of monarchs and large sovereign states. Hobbes justifies this condition as a necessity, and Rousseau suggests that some technocratic administrative changes to the governance structure could promote greater equality (sound familiar?).

Ultimately, both of these threads of modern political philosophy rely on the assumption that governance happens at the scale of millions of people. These philosophies were born and formed as part of the transition from kingdoms to nation states, and the conclusions they come to are constrained by the scale of the problem they are trying to solve.

But Rousseau drops some hints, little sets of breadcrumbs, that point towards different paths outside of the culturally accepted standards of his time. The first breadcrumb comes from the preamble to his Discourse on the Origin of Inequality.

It’s a cheeky bit of subterfuge. While Rousseau takes six opportunities to dedicate the essay to his “most honourable, magnificent and sovereign lords,” he uses the rest of the dedication to describe the type of society he would prefer to live in: one that doesn’t have sovereign lords.

Instead, he describes a society “which had an extent proportionate to the limits of the human faculties […] every person being equal.” In other words, a small, cooperative, democratic city-state:

a free city situated between several nations, none of which should have any interest in attacking it, while each had an interest in preventing it from being attacked by the others; in short, a Republic which should have nothing to tempt the ambition of its neighbours, but might reasonably depend on their assistance in case of need.

The second breadcrumb comes in a preface and footnote in his seminal work, The Social Contract. In the preface, he claims he “will show later how the external power of great people can be combined with the ease of administration and the good order of a small state”. In a later footnote, he acknowledges that while he “would have come to confederations” to address this question, he has “long since abandoned” the pursuit and “the rest of the work no longer exists.”

It’s impossible to know if Rousseau abandoned these paths because he understood the ramifications they could create for the sovereign lords that determined if he stayed alive or if he ran into mental roadblocks trying to imagine a future beyond the sovereigns and nation states of his day. But I believe that these two breadcrumbs—a preference for small, local, cooperative political structures and the idea of autonomous confederations of these structures—point the path towards a better social contract.

The success and failure of early federations

The founding fathers of the United States seem to have understood this in principle when they created a carefully balanced federation of semi-autonomous states. While their creation was a miracle of political progress, it suffered the same two limitations as Rousseau’s philosophy: a culture deeply steeped in existing sovereign monarchies and the need to create administrative systems that could work at the scale of a growing nation-state.

Their goal was to organize 2.5 million former subjects of kings, spread across a quarter million square miles of land, with extremely limited transportation and communication technologies. The options on the table were limited by these realities, and further constrained by the boundaries of the Rousseau-Hobbes spectrum of political possibilities (and, of course, the fact that it wasn’t exactly a diverse decision-making body):

Given these limitations, our current political state should come as no surprise. The grand attempt at a federated structure has ultimately re-centralized, devolving into a low fidelity duopoly of political parties, entrenched in each of the basic social contracts offered by Hobbes and Rousseau. It’s a damning choice between being trapped in submission to a strongman and the bureaucratic administration of increasingly complicated attempts to manage inequality.

Considering alternative paths

What happens if we take a step back and retrace the origins of these political philosophies to look for alternatives? Are there other approaches that make more sense within the context of the coordination tools now at our disposal?

If we want to find alternatives to the trappings of our Western political philosophy, an obvious starting point is the woefully understudied political organizations of non-Western civilizations. David Graeber and David Wenrow’s The Dawn of Everything extensively documents historical examples of political and social structures that are not well understood or accounted for by Western political theory. The WEIRD (western, educated, industrialized, rich, and democratic) society that has been writing history recently has deep cultural blinders to other ways people can self-organize. Our basic cultural story — simple egalitarian tribes in the garden of Eden, the emergence of agriculture and hierarchy, the development of states and economies, the inevitable resulting tradeoffs of leviathans and inequality — is not the only path:

We do not have to choose between an egalitarian or hierarchical start to the human story. Our early ancestors were not just our cognitive equals, but our intellectual peers too. It’s becoming increasingly clear that the earliest known evidence of human social life resembles a carnival parade of political forms.

The society described by Rousseau is not just some abstract ideal. It was the lived reality of some groups of humans for long stretches of history, alongside an incredible diversity of other approaches to political systems and social contracts. The goal here is not to identify some abstract idealist tribal state to return to, but to consider the full set of options available to us and then run lots of local experiments.

Governance experiments

Political science is not, per se, about science. Most of academia got physics envy over the past half century and tried to turn everything into science. But no matter how much statistical analysis you do on nation states, the questions of political science are really philosophical debates about how we should live and govern ourselves. There aren’t many nation states, and they typically don’t let academics go run experiments in governance.

The field of political science is defined by the study of sovereign entities and the historical biases of Hobbes and Rousseau. But some political scientists have broken out of their WEIRD blinders and explored other methods of self-governance. The founding explorer of this political tradition is Elinor Ostrom, who studied questions of common pool resources and collective action problems, with a focus on irrigation networks in traditional agricultural societies.

In one of her excellent essays, “Beyond Markets and States: Polycentric Governance of Complex Economic Systems”, Ostrom describes the ways in which “humans have a more complex motivational structure and more capability to solve social dilemmas than posited in earlier rational-choice theory.” In order to understand and improve the complex adaptive systems in which humans self-govern, she argues, it’s “important to examine the effect of precise combinations of variables in an experimental setting.”

As a starry-eyed fan of Ostrom in college, I designed my senior thesis around this goal. I ran experiments using common-pool resource simulation software created by Marco Janssen (a research scientist alongside Ostrom) with the guidance of my thesis advisor Tun Myint (who completed his PhD with Ostrom). The experiments brought groups of students together to play a computer game managing a shared common pool resource. I studied how successfully they maintained the shared resource under different communication and information limitations.

It was a rudimentary experimental design using meager academic resources. There were no conclusions that could be reasonably abstracted to other settings, and Ostrom’s dreams of an experimental political science felt hopeless in the real world. Political science, as a discipline, remained trapped in its philosophical roots. Political scientists assumed that one did not just go start new governments in the real world to see what happens.

Now we can. Informed by the vast tapestry of human organization across history, we can use blockchain tools to create new models of coordination. We can design new social contracts—not by writing theoretical essays and submitting them to academic contests (as Rousseau did), but by studying history and testing in prod out in the real world.

The blockchain leviathan

The Left/Right dichotomy is an increasingly false choice. It treats the leviathan — whether in the form of a king or a federal bureaucracy — as a basic requirement to organize large societies and enforce the rule of law.

What is fundamentally interesting about blockchains is that they refactor the basic assumptions of Hobbes and Rousseau into a technology that doesn’t require human administrators: a new type of leviathan. By allowing people to self-organize into capture resistant small pods of effective coordination, blockchains rewrite the basic assumptions about the necessary scale of governance.

I’m not suggesting that blockchains are prepared to fully replace the monopoly on violence of nation states. But in the same way that iPhones put a supercomputer in everyone’s pocket, blockchains put the basic building blocks of sovereignty in everyone’s private keys.

Without the need for any additional sovereign entity, anyone can now create an organization that provides immutable rights of governance to members and, if they want, create an independently controlled currency for the organization. Re-read that last sentence, and then consider how Rousseau might have reacted if you told him this was possible.

Social smart contracts

DAOs are social smart contracts. Each one uniquely embeds a set of cultural norms and immutable rules into an entity. They are using this socialware and trustware to actively develop, test, and explore blockchain tools for coordination and self-governance.

Through building and operating these organizations, we can experiment across a broad design space of rules, norms, and behaviors for people working together. Through the chaos, we can develop political systems that reach into the knowledge of the past and apply it to the tools of the future.

Over the past 18 months, a few principles have developed across many projects simultaneously, pointing towards some possible underlying truths of the unique powers of these new tools. One principle that has become increasingly clear for DAOs is the need for small, autonomous pods loosely coordinated through diverse, decentralized governance. Vitalik wrote about this model in his most recent piece on DAOs and Metropolis has pioneered on-chain management tools for pods.

Whether they are called pods, working groups, fellowships, guilds, or subDAOs, most decentralized organizations have come around to the realization that you need small groups to get things done. If you ask people what group sizes they like working with, they will generally tell you numbers between 2 and 12 people. Amazon is well known for popularizing the term two pizza team to describe this concept (though at Cabin, we prefer one sauna teams).

Adding more people usually doesn’t result in better outcomes, because coordination costs increase geometrically. This coordination cost is the flip side to Metcalfe’s Law:

Macroeconomics generated the idea of economies of scale, and similar calculations are made in the context of the macropolitics of nation states. But in what we might call micropolitics — the study of small group collective action — scale comes with significant costs. When there is no need to use economies of scale to subsidize centralized coordination and trust management mechanisms, the default size is small and the topology becomes a network.

David Ehrlichman, who has been studying and growing impact networks for years, has diagrammed the way that this network topology grows over time:

These networks of small self-sovereign entities are starting to look a lot like what Rousseau hinted at in the margins of his greatest works. We are rediscovering in practice what he began to theorize about: small, local, cooperative political structures organized into autonomous confederated networks.

But we no longer need to theorize about how they could work. With the tools of governance and self-sovereignty offered by the blockchain leviathan, we can begin to explore and create these complex network structures. Hopefully they point the path towards a better social contract.

Thanks to Lauren Alexander and Chase Chapman for ideas and feedback for this essay

I’ve been living on the road a lot this summer—jumping around between Texas, Minnesota, California, Colorado, and the East Coast. Dropping into different little pockets of the greater Cabin universe has given me a stronger sense of what it will be like for people to live across the Cabin network. For this update, I’ll share some of the slices of Cabin I was fortunate to be a part of during my recent travels.

My personal highlight of the season was spending time at Montaia, a Cabin neighborhood in the Eastern Sierra. I flew into Bishop, CA and within minutes I was up in the mountains, hearing stories of the epic tubing adventure of the previous day and participating in a group circle where people were opening up and getting to know each other. As soon as I joined the squad of colivers there, I knew I was back home at Cabin: an incredible group of humans had come together in a stunning part of nature and were forming deep relationships. It was a moment of joy for me to realize that this was just happening in a place I’d never been with people I’d never met.

To top it off, I got to join in and help lead a backpacking expedition through the wilderness into the backside of Yosemite. Backpacking has always been one of my favorite activities. I’ve covered many hundreds of miles of trail across some of the best nature in the world—but nothing prepared me for the jutting rock peaks, deep alpine lakes, and wide open meadows of true Sierra wilderness. It was magical:

Which brings me to the most exciting news of this update: a new Caretaker has joined Cabin. Kaela, who created Montaia, embodies what it means to be a Caretaker. The biggest bottleneck to building our network of neighborhoods is great Caretakers: people who build a sense of place and create space for others to join them. Caretakers are leaders and builders of the physical and social spaces that make up Cabin. It’s a deeply challenging and rewarding job, which makes it fun to swap stories and learn from others who also practice the art of deep community building. A huge welcome to Kaela, with whom I look forward to welcoming many more Caretakers to Cabin in the future! (Maybe I can even get her to start writing some of these posts…)

After Montaia, the Cabin train headed to Denver for MCON. As a lead sponsor, we ran core programming spaces for connection and coherence, sold our first merch, talked to dozens of squads interested in starting Cabin neighborhoods, and gave people the opportunity to apply to come colive with us. Our programming in The Room and Phil’s talks on co-buying were packed sessions that we heard from many people was a highlight of the conference:

I’m grateful to the team of Cabin contributors who came out to support our MCON efforts (including Kaela, who—as she often does—was taking the picture and therefore is not pictured):

We also rolled out beautiful posters, which included Cabin’s newly worded principles:

It’s too early in Cabin’s journey to write anything down in stone (or on-chain), but I do think these three words could serve as what Balaji might call our moral innovation or what Toby Shorin might call our flavor of Life After Lifestyle. While Balaji insists network states should have One Commandment, our network city prefers Three Suggestions:

Conserve. Colive. Create.

Conserve is about our relationship with nature and the land we live on. It's about building a sustainable future by living in balance with the earth's natural systems.

Colive is about our relationships with each other. To borrow Radish's "obvious truth": we are happiest and healthiest surrounded by people we love and admire.

Create is about what we can do when we live in this type of environment: imagine and build great things with other people across the world.

No governance process or declaration could make these words into values for Cabin — only we can do it, by living in accordance with them. Conveniently, we think these principles make for a simple set of guidelines for a well-lived life.

In the weeks since MCON, Cabin hosted its first international events. Community members gathered for Build Weeks in Puerto Rico, an island camping retreat in Greece, and a builder residency at Neighborhood Zero. Despite a hurricane in Puerto Rico (everyone is safe!), a hard-working crew came together to build a beautiful new outdoor community space:

I was particularly excited to get back home to Neighborhood Zero after a long set of traveling. While it’s great to bounce across the Cabin network, it’s even better to have a neighborhood you can call home. Charlie, our Builder in Residence for the fall, was already there when I arrived. Building physical spaces is a core motivation that led me to work on Cabin, and I can see that same energy in Charlie. You can see it too in the hilarious, fun, and informative weekly YouTube series he’s been producing:

Spending time with Charlie as he simultaneously builds and creates videos about his building process has been such a clear reminder of one of our core feedback loops as an organization—the one between builders and creators:

Charlie and I spent hours walking around the property, considering and modifying the emerging master plan. We talked through design, material, and placement options for buildings we’ll be completing this fall. And we talked about the path towards providing a 10x cheaper and better lifestyle for people by drastically reducing housing costs and building community-centric living spaces.

On Sunday, the New York Times ran a front page story headlined Key Housing Crisis Factor: The Disappearance of Starter Homes in the US. It makes the case that Americans want more low-cost housing, but developers are not building it for two reasons: land costs and regulation. In urban areas, high land costs and excessive regulation make it economically impossible for developers to build the lower cost housing people want. But while these two factors are crippling housing in urban areas, they are the same two factors that are allowing Cabin to develop lower cost housing for flexible online workers.

We have a long way to go before we can deliver a consistently 10x cheaper and better living experience for people, but we are building a basic proof-of-concept. This week, Dwell announced a partnership with Adobu to sell 540 square foot prefabricated and installed 1-bedroom homes for around $400,000. For comparison, this is roughly the cost of our 1280 square foot 4-bed / 4-bath custom prefab shipping container home. And now, Charlie is building tiny homes using partially prefabbed sheds (which we call shedrooms) for under $25k. If we keep innovating on our building methods, I am confident we can create long-term spaces for communities that offer a better and much more affordable lifestyle for people.

A compelling moral innovation is one that responds to people’s basic needs and offers a different, better path. As I’ve lived in community with Cabin people this season, it’s become abundantly clear to me that we are building this type of innovation. We can conserve nature, colive with people we love and admire, and create spaces together that are simultaneously more affordable and better living environments. We’re not there yet, but the path is becoming clear in front of us. See you out on the trail.

[originally published September 28th, 2022]

Our current world is designed primarily around large centralized structures like national governments, universities, and corporations. But this hasn’t always been the case—cycles of decentralization and centralization define the ebb and flow of major periods of human history.

We seem to be in a late-stage period of the modern era, where centralized institutions are failing. So it’s natural to run in the other direction, to swing the pendulum toward decentralized maximalism. Maximalism is the strategy of a cancer cell. Successful complex systems are diverse, permeable, and inherently pluralistic.

If we’re going to run screaming from centralization, we need to have a sense of where we are running towards as well as what we are running from. Movements focused on abandoning centralized structures without suitable replacements tend to end in apathetic collective action failure, or worse, mob rule. If we are to avoid the proverbial guillotine, we should learn what has worked and failed in other decentralized organizations. We need to ask ourselves:

In what ways can decentralized organizations outcompete more centralized ones?

Ideologies don’t win by claiming superiority—they win by demonstrating that they can solve real problems more effectively than the alternatives. So it’s worth deeply understanding what has actually made decentralized organizations better (and worse) than the available alternatives.

Principles of successful decentralized organizations

Here are four very different kinds of organizations that have something to teach us about building DAOs:

• Valve, the company behind Steam

• the US Marine Corps

• intentional co-living communities

• Amazon

Each of these organizations operates in decentralized ways that result in better outcomes than more centralized peers. They have created flexible, adaptable, creative, and capable evolutionary systems that can outperform more rigid hierarchies in rapidly changing environments.

This wide range of organizational types shows the different ways that decentralization can manifest as a strength. By looking at the commonalities in their principles, we can understand what makes decentralized organizations successful. Here are a some of the principles they all have in common:

1. Recruit the best people

2. Help members self-organize

3. Empower mission-driven leaders

4. Have a bias towards action

5. Play infinite games

Note that none of these principles are about decentralization absolutism. They are about finding the right balance between decentralized self-organization and flexible leadership structures that create Schelling points for collective action. Let’s explore each one in a bit more detail.

1 - Recruit the best people

There’s one principle that every successful organization seems to agree on: great organizations are made up of great people. No amount of organizational design can make up for people who lack the skills, ambition, and passion to accomplish great things. One of the most triumphant passages of Valve’s employee handbook reads:

You have the power to green-light projects. You have the power to ship products. A flat structure removes every organizational barrier between your work and the customer enjoying that work.

If you’re thinking to yourself, “Wow, that sounds like a lot of responsibility,” you’re right. And that’s why hiring is the single most important thing you will ever do at Valve. Any time you interview a potential hire, you need to ask yourself not only if they’re talented or collaborative but also if they’re capable of literally running this company, because they will be.

As far as I can tell, “engage amazing humans” is the core strategy of all great organizations, and especially those that aspire to high levels of decentralization. Organizations are made up of people, so if you want to build great things, it’s natural that the first rule is to work with great people.

Working with great people makes your life easier. Great people are smarter and more competent than you at the things they spend time on. They see around corners you didn’t know existed. They deliver excellence in ambiguous situations without guidance. They have strong vibes and attract each other with these vibes.

Companies like Valve and Amazon, elite military units like the US Marines, and co-living environments like intentional communities all draw boundaries around themselves and carefully evaluate who they let in. By constantly attempting to raise the bar, they can keep scaling without the need for more structure and rules. From Valve again:

We believe that if we’re careful, it will work better and better the larger we get. This might seem counterintuitive, but it’s a direct consequence of hiring great, accomplished, capable people. Getting this to work right is a tricky proposition, though, and depends highly on our continued vigilance in recruiting/hiring. If we start adding people to the company who aren’t as capable as we are at operating as high-powered, self-directed, senior decision makers, then lots of the stuff discussed in this book will stop working.

DAOs often attempt to be permissionless, allowing anyone to join the community. Maintaining openness, permeability, and an inclusive approach to newcomers allows DAOs to grow quickly, increase diversity, and progressively decentralize. But there’s a clear tension between bringing in the best people and maintaining an open and accessible community:

I think the best solution we have to solve for this balance is contributor proof-of-work. Proof-of-work, in this context, refers to the ability for new community members to demonstrate their capabilities and increase the scope of their work as they build trust within the community. By creating systems for self-organizing teams to publicly accomplish the organization’s goals, DAOs can build feedback loops for strong contributors to succeed and positively reinforce the organization's culture, capabilities, and scalability.

2 - Help members self-organize

Once you are surrounded by great people, you can work together to build systems for effective collective action. In order to self-organize, people need:

1. Clearly defined missions

2. Small groups executing at the edges

3. Information accessibility

4. Minimum viable self-governance

Clearly defined missions

Like a slime mold growing towards an available source of food or an oak tree growing towards sunlight, the first thing self-organizing systems need is a goal to grow towards.

Jim Mattis, a US Marine who served as Secretary of Defense, helped define the US military’s counterinsurgency (COIN) doctrine. COIN is an approach to warfare designed to react to highly decentralized terrorist networks. This decentralized military doctrine considers clearly defined missions to be the most important prerequisite for allowing groups to self-organize, because no one can coordinate without a shared understanding of the goal. From Mattis’ memoir, Call Sign Chaos:

The correct exercise of independent action requires a common understanding of both the mission and intent of what the mission is expected to accomplish.

This is why the most successful DAOs so far are organized around strong memes that give members a shared sense of purpose and mission. At Cabin, we are experimenting with a city council that defines objectives for each season and creates budget allocation for those objectives.

Small groups executing at the edges

Once participants have a sense of the mission, they can get together in groups and attempt to accomplish it. Successful groups tend to be small and purposeful. More people does not usually result in better outcomes, because coordination costs increase geometrically. This coordination cost is the flip side to Metcalfe’s Law:

If you ask people what group sizes they like working with, they will generally tell you numbers between 2 and 12 people. Amazon is well known for popularizing the term two pizza team to describe this concept (though at Cabin, we prefer one sauna teams). DAOs can be thought of as constellations of these small teams, sometimes called pods, squads, guilds, or—if you want bad lore—subDAOs.

Information accessibility

Allowing people to self-organize into small teams requires them to have access to information and the ability to execute autonomously. Information access may be the most underestimated benefit of DAOs. Blockchains, by nature, are public ledgers. This means that DAOs, by nature, operate with high degrees of transparency about resource allocation. This is a particularly important breakthrough for DAOs because information transparency and accessibility, especially for newcomers, is a weakness of decentralized organizations like Valve:

Companies end up developing implicit hierarchies of information availability, even if no formal hierarchies exist. Knowledge, as they say, is power. If conversations, transactions, and processes are publicly available, anyone can (at least in theory) enter the organization and contribute positively to it.

Minimum viable self-governance

Autonomy means individuals can make execution decisions without bureaucratic processes that constrain their decision-making authority. The problem with rules or processes is that they can be captured by special interests who use them to retain power instead of for the common good. An example of this is the United States House Rules and Manual, which contains hundreds of pages of drivel like this:

A reappropriation of unexpended balances of appropriations may not be reported in a general appropriation bill, and may not be in order as an amendment thereto…

Rules beget more rules, followed by processes to make sure rules are followed. Rules that determine available actions are inherently political tools. Before long, the meta-game of rule making becomes the primary force for executing the will of power, regardless of how useful the outcomes of this process are. This was one of the primary fears of the US founding fathers in designing a system of governance.

But rules exist for a reason. Resources are constrained, and if anyone can do anything with collective resources, they are quickly diminished in a tragedy of the commons. What is needed to solve this problem is minimum viable self-governance. Elinor Ostrom’s design principles for common pool resource institutions is the leading framework for how humans have solved this problem in a bottom-up way for millennia.

More recent thinking on lorecraft is an interesting reframing of the rules problem. By focusing on memetic tools for building cultural stories, lore is a compelling alternative to rules-based systems. Telling stories about cultural norms is a powerful narrative tool for organizing without adding process.

3 - Empower mission-driven leaders

There’s no such thing as a leaderless organization. When you try to build perfectly flat organizations, you just end up hiding hierarchy in illegible ways, which seems to be an issue at Valve. While transparent and accessible information can help combat this, the most illegible hierarchies form when groups pretend they don’t exist.

When we say mission-driven leaders, we mean it in two ways. First, it means that the best leaders are missionaries, not mercenaries. They believe in and are dedicated to the greater mission of the organization. Second, it means that leadership is a temporary, transient role designed to solve for a specific mission.

DAOs offer a unique solution to this problem that hasn’t been technically feasible before: put direct executive power in the hands of all individuals in an organization. In his anti-capture framework, Spencer Graham identifies direct executive power as the key to successful decentralized organizations:

Separation between decision-making and executive power is uniquely possible within capture-resistant structures. It is why decentralized decision-making is not a strict requirement. It is also why decentralized organizations can be fluid and adaptable. When executive power is decentralized, leadership can truly emerge from anywhere.

Humans want and need leaders to create shelling points for collective action. DAOs should be leaderful organizations — groups that support just-in-time leadership from a wide range of capable individuals. A leader is whoever knows what to do next. Effective DAOs give members the tools to execute when they are the best leader for a shared cause.

While Amazon is well known for their use of two-pizza teams, what’s less well known is that this type of leadership––known internally as single threaded owners––is considered even more important for team cohesion and success. This type of fluid leadership only works if people within the organization have a strong bias towards action. The organization must be designed to support people taking the initiative to step into self-defined leadership roles and making things happen.

4 - Have a bias towards action

The CIA wrote a fascinating manual called the Simple Sabotage Field Manual. In addition to detailed tips and tricks for starting fires, targeting warehouses, destroying tools, and clogging toilets, the manual includes an entire section on undermining the work of organizations. The manual describes the failure state of organizations that have unclear leadership and a lack of bias towards action:

Anyone who has ever been in an organization attempting too much flatness and consensus has probably experienced these problems. This was the downfall of intentional living communes of the 60s and larger experiments in communism through the 1990s. When everyone has an equal voice in everything all of the time, groups end up talking a lot and accomplishing nothing.

One antidote to this problem comes from co-living communities, which are often organized around a principle called do-ocracy. This principle states that, generally, whoever actually does the thing makes the rules. Phil Levin, one of Cabin’s city council members and founder of several successful co-living communities, put together a fantastic presentation called What DAOs can learn from co-living communities. The communities Phil has started thrive on the principle of self-appointed do-ocratic leadership:

The goal is to minimize governance and consensus decision-making by empowering individuals to make autonomous actions as long as they are reversible and inexpensive. Another good example of this type of decentralized leadership comes from Cabin’s Build Weeks.

Build Weeks are self-organized by a primary build leader, and also support room for other builders to go on do-ocratic side quests that contribute to the built environment of Cabin’s Neighborhood Zero. Here’s how participants in a recent Build Week described effective do-ocracy:

Despite a lack of stringent roles and duties, everything got done, we were able to use our unique skills and interests without any conflict, and most importantly, everyone got to drive the skidsteer.

Valve also emphasizes a do-ocratic approach to product development, where anyone can “come up with a bright idea, tell a coworker about it, work on it together, and ship it”:

To make this work, you need to design your organizational processes to explicitly reward people trying new things. As Jim Mattis explains:

Instillation of personal initiative and risk taking doesn't spring forward spontaneously. It must be cultivated and rewarded in an organization's culture […] Be tolerant of mistakes. If risk takers are punished, then you will retain only the risk averse.

When things don’t go well, ask yourself: is it because someone took strong initiative and the experiment failed, or because they failed to take the initiative in the first place?

5 - Play infinite games

Rewarding leaders with a bias towards action is key to successfully executing as a decentralized organization. But execution is a finite game—a convergent action to accomplish a specific goal. The final principle of effective decentralized organizations is inherently divergent: play infinite games.

Vibes, reciprocity, and trust are infinite games. They are not played to win, but to keep playing. Infinite games are played for intrinsic rewards:

The co-living communities Phil has helped start revolve around the principle of intrinsic rewards. Phil’s partner Kristen, a behavioral scientist, knew that intrinsic rewards are a much more powerful and fulfilling form of motivation than extrinsic ones. As a result, their co-living communities do not offer direct compensation for essentially any work within the community. This policy has ultimately made the community stronger and more productive than if everyone played finite games of resource allocation.

Intrinsic community building means valuing community members beyond the work they do for the organization and allowing people to contribute to the community without expecting extrinsic rewards. While the work of a DAO should reward excellent executors with increased responsibility and financial upside, the whole community does not need to be held to the same standard because the motivations for participation are different. Communities are about vibes and relationships — both of which are driven by intrinsic motivation and infinite games.

Infinite games will take DAOs to interesting places. Amazon is playing an infinite game. Notice what’s missing from Amazon’s famous flywheel—any arrows leaving the diagram:

Amazon is one of the world’s most powerful self-perpetuating economic feedback loops. It is a mechanism for feeding back on its own growth, which creates exponential growth in revenue without capturing profits:

The most potent DAOs will follow a similar path—they will create intrinsically motivated communities that self-perpetuate memes and culture while simultaneously developing protocol hyperstructures that live indefinitely and create value without necessarily extracting profits.

This is the end state of sufficiently decentralized organizations that points us towards a future that companies are not designed to produce. Organizations of the future can be collectively owned and organized around a token that provides utility to those who hold or stake it and that never attempts to produce dividends or profits that are extracted from the system—all while still providing massive value to network participants. Like ecosystems, economies, and other forms of complex adaptive systems, DAOs can be structures for playing infinite self–fulfilling games that reward participants with meaning and purpose.

The DAO playbook

No two DAOs will be able to follow the exact same playbook, because their purposes, communities, and resources will look different. But, like companies, a diversity of products, business models, and cultures will develop in DAOs—and there will be repeatable patterns of success in the ways organizations treat DAO as a verb.

Ultimately, by integrating the ways that centralized hierarchies create efficiency and decentralized networks create resilience, we can chart a path towards more effective organizations. Charting this path will require a deep understanding of historical precedent and a penchant for constant experimentation. No matter what these organizations of the future end up looking like, the ones that succeed will probably do these things:

1. Recruit the best people

2. Help members self-organize

3. Empower mission-driven leaders

4. Have a bias towards action

5. Play infinite games

thanks to Phil Levin, Zakk Fleischmann, Chase Chapman, Commodore, Erik Trautman, Allen Taylor, Spencer Graham, and Rafa Fernandez for ideas and edits reflected in this essay

[originally published June 6th, 2022]

Lauren and I spend the summers living in a cabin on a lake in Minnesota, something we’ve done together for at least part of every summer for a decade. In addition to her love of children, having flexible summers to come to the lake has always been a key fringe benefit of her career as a school counselor. We have developed a deep and abiding sense of place here. It is where we fell in love, where we got married, and where we tend to feel most relaxed and happy together.

Deep in the early pandemic summer of 2020, I worked remotely from this cabin for the first time. It was the proof-of-concept for Cabin. That summer—living in a cabin surrounded by nature and still deeply connected to my work—taught me what was going to becoming possible for tech workers like myself.

It was the summer I took Write of Passage, started the Creator Coop, wrote and published essays consistently for the first time, began actively using twitter, started angel investing in creators and communities, and committed to building my own cabin in the woods. The seeds of Cabin were planted here, and returning each summer feels like a return to some of Cabin’s earliest roots.

We hosted Cabin’s core contributors and city council out here at the lake over the past few weeks as we planned for Cabin’s next season. It felt like the DAO had reached a new cross roads. Two years ago, we were a loose set of ideas, not yet woven together. A year ago, we were a scrappy crowdfunded creator residency project. Now, we were steering a large community in the vague but exciting direction of founding a new city.

In the past 8 months, we accomplished a tremendous amount. We sold out seasons of residencies and retreats, ran deeply meaningful and impactful programs and camps for DAO leaders, brought the DAO together to build impressive physical projects, launched innovative products, published top-notch content, and built a community of the best people in web3. In short, we accomplished the bulk of our 2022 Roadmap to build an Embassy for DAOs six months early.

But, to be honest, we were beginning to feel rudderless. We had not clearly articulated our vision beyond bringing together DAO people for gatherings. Our organization was growing large, unwieldy, and unhelpfully decentralized. In our success, we were starting to lose sight of the mission and get bogged down in open-ended divergence and introspective governance.

Then, the floor fell out from under the market. Suddenly nothing felt certain or guaranteed. Cabin’s treasury diversification set us up to weather a bear market and our physical infrastructure makes our community more resilient than most of modern society. But a bear market—even if not existential—is a good time to re-examine your approach, focus on what’s important, and prepare for the winter ahead.

The theme of our Cabin retreat, hosted by Zakk and Chalice, was “In Frame”. What should be in frame for Cabin? What should be out of frame? How could we ensure that we were focused on the most important goal for the season? How could we rise to a moment that demanded convergence, practicality, and a unity of purpose?

There’s nothing like a market meltdown to bring a sober-headed realness to such discussions. Over the course of the retreat (and a much appreciated jaunt out to Minnesota from newly minted father Phil), the focus became abundantly clear:

Cabin is a network of neighborhoods

When we talked about “the DAO”, we realized we were falling into a mental trap. We were putting ourselves at the center of a growing, monolithic organization. We knew Cabin needed to grow by breaking itself apart in a constellation of smaller organizations, but we were focused on what that meant for us as contributors. We spent too much time working on the work and lost sight of the goal itself: building a network of neighborhoods.

We realized that we, as contributors, are not the core of the DAO we were building. We are merely a service provider to the DAO. And our online community, while an important part of the organization, is also not—in and of itself—the DAO.

The DAO is the network of neighborhoods that make up Cabin. The physical neighborhoods and the communities built around them, are at the center of Cabin’s story and mission. Those neighborhoods are connected together via Cabin’s online community and the tools we build as contributors.

Anything we do that does not directly contribute to neighborhoods and their development is probably not “in frame”. This is why Cabin’s city council voted to allocate our entire Summer 22 budget to a single Objective: Neighborhood Development. And it’s why all of the core contributors decided to band together into a single Fellowship, a One Sauna Team focused on Neighborhood Development.

With this clarity of purpose, organizational structure, and focus, I am feeling more excited than I have since our very first core team retreat. It feels like we have come through an organizational looking glass, and the nature of our network topology is clear in front of us.

Cabin is a network of independent neighborhoods developing shared culture, economy, and governance structures. We will do everything we can to help these neighborhoods develop, grow, and thrive as communities. Small fellowships of service providers will go on quests to build the incentive, trust, and growth mechanisms to help neighborhoods realize the communities they want to create.

The pace layers of society are bending towards us. If we band together into small groups, focus relentlessly on the mission, and work hard to make it a reality, we have the opportunity to give millions of people a new way of living and working together. It’s time to build.

Financial flash mobs

Financial flash mobs turn potent memes into money very quickly. Constitution DAO was a breakout moment for this type of DAO: over 17,000 people came together to pool $47M in less than a week. Typically, it takes longer than that just to set up a corporation and open a bank account.

Less savory examples of financial flash mobs are token pump-and-dump schemes. By having aligned financial incentives, internet strangers can coordinate to drive up the price of a token and sell to a greater fool. The ponzinomics of greater fools is so powerful that even things with no intention of becoming a meme coin sometimes end up as one. After internet supervillain Ken Griffin revenge trolled Constitution DAO, the plan was for everyone who participated to claim refunds for their contributions. But, much to the surprise of the core team, the token got pumped into a meme coin and is still trading inexplicably above its actual redemption value:

This meme premium (memeium?) to redemption value, like much of crypto-economic theory, is clearly intuitive and also makes no rational economic sense. Beyond the basic ponzinomics of the greater fool theory, there are some people who just love the meme and will never sell. Owning the token (and, for economic purposes, staking it) has become a part of the identity attached to the wallet and its owner.

Another reason to hold on to tokens is that they can become seed lists for other airdrops. From the perspective of a token holder, it’s economically rational to hold on to something you don’t care about if you believe it may generate additional financial assets in the future. From the perspective of an airdropper, using another token as a seed list solves several practical and theoretical problems.

Theoretically, it solves one of the hardest questions of economic systems: how to initially distribute the money. Adam Smith’s theory of the invisible hand explains how money gets efficiently distributed in a market, but notably fails to provide any guidance on the initial distribution of money at the start of the system. Practically, seed lists allow someone to create a new token that forks another existing tokenized community by airdropping a token to their members.

When Constitution DAO lost the bid and decided to wind down operations, dozens of other organizations like People DAO popped up, offering token holders the chance to put their stake into a new organization and continue the journey. The deep irony here is that, while most financial flash mobs burn out with the same rapidity that they caught fire, the ones with the deepest meme potential may cross the chasm into the other end of the spectrum and live on forever.

Hyperstructures

On the other end of the spectrum are DAOs trying to build hyperstructures that will last indefinitely. Hyperstructures, coined by Jacob Horne at Zora, are crypto protocols that can run for free and forever, without maintenance, interruption or intermediaries.

Hyperstructures are a platonic ideal for DAOs—unstoppable, free, valuable, permissionless protocols that can serve as infrastructure for a wide range of use-cases and applications. This platonic ideal is essentially how DAOs were originally conceived, but not how the term is commonly used today.

What we commonly call a DAO today is not exactly what Vitalik imagined when he speculated about them in the Ethereum whitepaper. DAOs were conceived as purely encoded, on-chain entities with immutable properties of decentralization. Most DAOs today, which are structured as multi-sig wallets of delegates implementing the will of off-chain votes, don’t meet the original definition.

This difference is more than a technicality. As Spencer Graham has argued via the anticapture framework, decentralized execution capabilities are a good strategy for limiting the exploitation of common pool resources. Furthermore, having organizations that are designed to run autonomously (i.e. without centralized leadership) is the only way to ensure that a hyperstructure will continue to operate indefinitely.

While there are few examples of truly functional hyperstructures out in the wild, bitcoin provides a benchmark for decentralization and autonomy. You can make the case that the original DAO was bitcoin. If you’re a real DAO purist, you could even make the case that bitcoin is the only DAO (but that’s only something converts to the religion of bitcoin maximalism would be likely to do). Additional examples of emerging hyperstructures include other blockchains like Ethereum, decentralized exchanges like Uniswap and Sushiswap, and auction protocols like Zora.

Most human organizations, like companies, get much less efficient as they grow. What enables these hyperstructures to survive indefinitely without centralized leadership is that they get better with scale. Geoffrey West has looked across natural and human systems at structures that can achieve superlinear scaling, where they get better as they get bigger instead of getting worse. While nature builds ecosystems like this naturally, humans have built very few examples of things that will last indefinitely and get better with scale. The only human structures that have this property are emergent systems like economies and cities:

Most current hyperstructures are essentially economies—what would a hyperstructure city look like? While cities have many potential use cases for immutable public records, the most fundamental could be what counts as the city itself. The idea of a decentralized city may be one of the best use cases for an oft-anticipated type of hyperstructure: the token curated registry. TCRs were a popular idea in 2017-2018, which is long enough ago in crypto years that they have been largely forgotten.

The idea is simple: tokenholders vote on a list. Anyone can apply to be on the list and it can represent anything. Let’s imagine that you wanted to implement a decentralized city as a TCR. The city would be represented as a list of places controlled by the token holders. Without any centralized decision-making body, the tokenholders could vote to determine what is and isn’t part of the city. Because a decentralized city isn’t all in one place, a hyperstructure like this could become the official record of what counts as “inside city limits”.

TCRs, much like DAOs until last year, have been wallowing in theoretical obscurity. They are a brainworm that has gripped thoughtful minds because they are simple rule sets that are ripe to grow into complex outcomes. I haven’t spent enough time with them to understand the pitfalls that may emerge from small rule changes in their structures, but we may have a strong use case for them at Cabin.

From memes to hyperstructures

Decentralized cities are interesting examples that bridge the gap from meme to hyperstructure. Hyperstructures are an incredible form of perpetual public goods infrastructure, but are very hard to bootstrap. Financial flash mobs are relatively easy to spin up, but usually burn out quickly. Could these two DAO-native structures be combined to produce something even more potent?

I think so. Ultimately, the best hyperstructures are very potent memes with thoughtful technical implementations. This is, of course, best explained via a meme:

Bitcoin is only digital gold if everyone believes it is. Token curated registries only matter if people care about the curation. DAO contributors spend so much time talking about vibes because the protocol only matters if there’s a community to use it.

That’s where the horseshoe wraps back around on the financial flash mob / hyperstructure dichotomy. Most memes burn out quickly, and even the best ones eventually exhaust the available hosts like a virus. But blockchain hyperstructures can turn a meme into reality by accruing on-chain resources to it.

Perhaps you start with an idea, and you crowdfund a small treasury to accomplish it. The crowdfund provides the initial capital and the token distribution list. But, instead of exhausting the fuel of the meme on a financial flash mob, you thoughtfully design a hyperstructure protocol to support the meme. Then, you use the flash mob funding to bootstrap from a tokenized community into a DAO-based hyperstructure by embedding the core idea of the meme into a hyperstructure protocol. In the case of Cabin, we are doing this by turning the meme of a decentralized city into a token curated registry that represents the city.

The inevitable endgame is that the best memes will be turned into hyperstructures. Building on the original memetic hyperstructures of Bitcoin (self-sovereign currency) and Ethereum (global shared computer), successful DAOs will figure out how to distill a potent meme, use it as a financial flash mob to gain initial resources like capital and contributors, and then use those resources to embed the meme into a protocol that can live on as an on-chain hyperstructure.

[originally published March 25th, 2022]

Out here at Neighborhood Zero, the last vestiges of the winter grasses are dying and the first sprouts of spring life are blooming forth. As if on queue with Cabin's transition of seasons, the bluebonnets, paintbrushes, buttercups, and coneflowers are bursting forth from the brown detritus of winter and declaring the arrival of spring. The timing is so perfect and the metaphor so clear that it feels like we should retire our seasonal numbering system and consider calling this next three months Spring instead of Season 3.

In my role as Caretaker, I spend my time dipping in and out of conversations and decisions across guilds within Cabin and across DAOs within the ecosystem. This gives me a broad aperture view of the landscape, and I intend to spend more time in the coming seasons distilling and sharing that view in ways that help contributors within Cabin and other DAOs with their work. Seasonal transitions provide a natural opportunity to look back and look forward, which we cover in this piece.

To get more content from us on how to level up as a DAO leader and operator, make sure to subscribe to our weekly newsletter.

Season 2 Lookback

When we proposed a theme for Season 2, we called it Multiplayer Mode: the transition from our proto-DAO residency program into an intentional set of programs for bringing groups together. Early in the season, we expanded this theme with our 2022 Roadmap, which charted a course towards our role as an Embassy for DAOs.

We've made tremendously fast and effective progress in achieving these goals — if anything, we have underestimated what we are collectively capable of. It's hard to believe it's been only six months since we became a DAO. Since then, we have:

• Developed our brand as one of the top social DAOs operating in web3

• Progressively decentralized into five guilds, each with strong leadership and contributors

• Run a full slate of impactful retreats, residencies, and build weeks as part of our transition to Multiplayer Mode

• Built out physical infrastructure at Node Zero and began the process of intentional node expansion

• Published dozens of high-quality essays on creators.mirror.xyz and launched a podcast

• Grown the breadth and depth of our virtual community via intentional onboarding, social clubs, and Guild Games

• Built two software products, Passports and Blaze

• Been featured in the New Yorker as an example of the future of DAOs

• Completed a token migration and treasury diversification to set ourselves up for long-term success

Looking back on our Season 2 goals and 2022 Roadmap, we seem to have underestimated ourselves in two ways:

1. The breadth of community activity that can be developed by a loosely coordinated group with a shared mission

2. The extent of interest in rapidly expanding the Cabin network to new nodes around the world

These strengths represent opportunities for how we think about the next season.

Scaling the breadth of community activity

The thing we have most consistently underestimated is the extent to which our online community shows up and manifests ideas in unexpected ways. This seems to be the superpower of successful DAOs: breadth of exploration around the edges.

When we launched the original residency program, we completely underestimated how many contributors would show up in the discord interested to work on things other than the residency program. Before long, we had writers creating content, engineers building software products, designers making graphics and UX components, placemakers developing and hosting new IRL programs, architects building new physical structures, community builders organizing social clubs, and thousands of people showing up in our discord and manifesting their own roles in the community.

Nurturing and growing this ecosystem is how we will continue to thrive. It will become incredibly tempting, at this point, to start to form more hierarchical structures. We are becoming too big for the loose informality of a small group. The classic solution to this problem is having more process, rigidity, hierarchy, and standardization.

We should not be opposed to these solutions dogmatically, but because they will not help us accomplish our long-term goals. We need to learn to scale superlinearly by finding ways to get better with growth instead of worse. Centralization, in the abstract, is not a root evil — and decentralization is not an unequivocal good. We are carving trails into uncharted territory, and we must use the full set of tools in our knapsack to discern the ways in which we can build organizations that have the scaling principles of cities, ecosystems, and economies.

The powerful emerging coordination tools at our disposal have opened new doors in the idea maze of how to build resilient, complex organizations. But we need the humility of history to bring these new powers to fruition. We can find inspiration in unexpected places. Zakk gave me a copy of Jim Mattis' Call Sign Chaos recently, and reading it has opened my eyes to the ways in which the US military is much less hierarchical than often portrayed. Mattis explains that decentralized leadership is the key to winning battles in rapidly evolving circumstances. While our work is much more cooperative than combative, Mattis' leadership advice resonates with us: "aligned independence is based on a shared understanding of the why for the mission. This is the key to unleashing audacity."

He explains his principles of decentralized leadership:

1. The correct exercise of independent action requires a common understanding of both the mission and intent of what the mission is expected to accomplish.

2. Trust must be the coin of the realm. Guard rigorously against overcontrol, which is compounded by the seduction of immediate communication.

3. Instillation of personal initiative and risk taking doesn't spring forward spontaneously. It must be cultivated and rewarded in an organization's culture.

4. Be tolerant of mistakes. If risk takers are punished, then you will retain only the risk averse.

5. Bottom line: imbue a strong bias for action.

If we are to become the decentralized city we strive for, we need to collectively create a culture that allows us to get better as we scale by fractally embodying (and adapting) these principles. Building this level of common understanding, trust, risk taking, and bias for action, will require deep thinking and experimentation in our norms, governance, organizational structures, metrics, and reward mechanisms — all within the constraints of limited time and resources.

Growing the decentralized city

In our 2022 Roadmap, we outlined a plan to grow our role as the Embassy for DAOs, both online and at Node Zero. Simply put, we are ahead of schedule in accomplishing these goals:

We knew that the next priority beyond this roadmap would be expanding to more nodes, but we didn't expect to get there until 2023. Now, with clear progress towards our Embassy for DAOs, node expansion is on the horizon:

1. Other DAOs focused on developing IRL communities are starting to pop up and want to partner with us

2. Our rapid progress at Node Zero (including a fully booked Season 2) suggests a need for expansion

3. We are developing theoretical frameworks for how a decentralized city could function

4. The community has expressed a clear interest in tangible steps towards expanding the decentralized city

Developing these ideas and frameworks into practice will take time. Keeping with the principles of bias for action, risk taking, and a shared understanding of the mission, we should lean into the uncertainties and spend the next season building a clearer sense of how to turn node expansion into a protocol for our decentralized city. The good news is that we already embody these principles as a DAO, and the community has started running in the right direction. Planning exercises like this are a lagging indicator of progress across the edges:

• We are already partnering with a little neighborhood of three distinct legal entities at Node Zero: Creator Cabins, Montanoso, and Girraween

• We've begun the process of partnering with Montaia Basecamp on a new location in the Eastern Sierras

• The new node pod within Placemakers has been tracking, interviewing, and building relationships with dozens of other organizations that could be future nodes

• We've developed a framework for how token curated registries could form the basis of the decentralized city

• We've shipped physical passport cards and NFTs that could form the basis for membership and identification across the network

Our next season should see continued progress in these directions, as well as other directions we haven't yet explored that contribute to the growth of the decentralized city.

Risks and roadblocks

Just as important as understanding our strengths and opportunities is understanding the risks and roadblocks that lie ahead of us. We should each think critically and independently about these, because risks can emerge in unexpected ways and are best identified at the edges of the network. Here are a few that keep me up at night:

1. Recruiting and retaining top talent

   • Proactively identifying and supporting strong contributors

   • Improving our continuous onboarding process

   • Evolving our compensation structures

2. Progressively decentralizing guilds

   • Empowering decentralized leaders

   • Improving our governance processes

   • Developing budget and metric protocols

   • Documenting to maintain legibility to newcomers

   • Designing on-chain organizational structures

3. Maintaining quality outcomes for end users

   • Balancing risk taking with rewarding success

   • Making outcomes more legible and celebrated

   • Maintaining high executional standards with scale

Onwards and upwards

It is with a deep sense of joy and gratitude that I look back and forwards with you all. I have found no greater joy in life than discovering shared meaning with other humans, and this community has been a tour de force in bringing that joy. In just over a year, decentralized cities have gone from an abstract idea to a rolling snowball, gathering incredible momentum and direction. Now feels like a crucial moment for us to collectively shepherd this vision from its current state into a more tangible reality. Let's build.

[originally published April 22nd, 2022]

There’s only two ways to advance civilization: bundling and unbundling

Jim Barksdale, more-or-less

Once in a while, in the long arc of civilization, a new set of coordination technologies come along and change everything. By allowing small groups of humans to better cooperate in the collective management of resources, these technologies redefine power structures and lay the bedrock of a new civilizational era.

Humans are social creatures—we got to where we are by bootstrapping coordination tools into complex organizational structures. That bootstrapping process follows a repeating pattern of decentralization and centralization over the course of the rise and fall of civilizations:

1. Technologies for coordination and communication rapidly bootstrap themselves into usefulness by civilization

2. These new coordination technologies allow humans to form effective local, decentralized governance structures (eg cities)

3. Ultimately, the federated network of decentralized governance is overpowered by a more efficient centralized structure of sovereignty

4. The centralized sovereign structure eventually collapses under its own weight, restarting the cycle by creating a governance vacuum

We can follow this story line through the epochs of human history. When historians classify that largest arcs of Western civilization, they usually map out four periods:

1. Ancient (10000 - 1000 BCE): emergence of agriculture, Sumerians, Egyptians, etc

2. Classical (1000 BCE - 500 CE): Greek city-states, Roman empire, etc

3. Medieval (500 - 1500 CE): knights, kings, castles, churches, etc

4. Modern (1500 - 2000 CE): Renaissance, Age of Enlightenment, industrialization, etc

This western-centric periodization obviously leaves out vast swaths of the human experience and cultures—it’s a timeline focused on how a particular slice of the world evolved over the course of human history. But, as a continuous story, it’s a helpful shorthand for considering how civilizations evolve. By comparing these periods, a consistent pattern jumps out that holds relevance for where we find ourselves today:

As with any attempt as psychohistory, this framework shouldn’t be taken too literally. I am an armchair historian, an amateur painting broad strokes and telling a story. But in the tradition of Carlota Perez, James Dale Davidson, Lewis Mumford, and David Hackett Fischer, I believe there is value in telling grand stories about the long arcs of human progress. I have drawn inspiration from each of their works here.

Ancient Era: the first cities and sovereignties

Irrigation provides a path for large-scale cooperation

Society started out in dank river valleys, where people figured out how to do agriculture and not die in floods. While humans had already started gathering in small bands, tribes, and villages before this, flooding river valleys required the development of cooperative structures that birthed cities:

The construction of [irrigation networks] demanded a degree of social intercourse, cooperation, and long-range planning that the old self-contained village culture, complacently accepting its limitations, did not need or encourage. The very conditions that made large urban settlements a physical possibility also made them a social necessity.

Lewis Mumford, The City in History

The not-dying-in-floods was nice, but it turned out that people had stumbled onto something that would prove even more valuable to the arc of civilization: cooperative development of public goods.

Individual humans rarely have a clear incentive to work together, leading to the tragedy of the commons. Localized irrigation projects are typically too big for any one person, but manageable by a small group of people working together. Furthermore, they can be collectively managed as a shared resource via systems of rules and norms for use.

In other words, this type of cooperative common-pool irrigation led to bottom-up governance systems (aka the original DAOs). These systems are so effective that they are still in use today and often perform better at managing irrigation than top-down government programs.

The first cities emerge

These newly developed cooperation capabilities led to a rapidly accelerating feedback loop of more people making more food for more people that made more things. Cities became the ultimate manifestation of “the enormous mobilization of vitality, power, and wealth” that “escapes rural limitations” and allows for “no mere change of size and scale”, but a “change of direction and purpose, manifested in a new type of organization” (Mumford).

The first cities became gravitational forces for people, which resulted in a higher density of ideas that led to rapidly escalating technologies: business, writing, money, laws, culture. These primitives of coordination and communication would become the seeds of future developments of civilization-altering technologies.

Fractal federation

Mesopotamia, where much of this action was happening around 10,000 years ago, translates to “land between rivers”. The fractal nature of the merging tributaries of the Tigris and Euphrates river basins created the conditions for an interdependent network of emerging cities:

Rivers are naturally occurring fractal networks of tributaries, joining together into larger basins. The fact that rivers flow in one direction, joining into bigger catchment basins, provides increasing leverage and the opportunity for concentrated power.

The proto-democratic irrigation cooperatives became a victim of their own success. Where there was technological leverage, there was a physical manifestation of power. As more people showed up, it became harder to coordinate them. At some point, a very cunning and charismatic leader, wearing the Mesopotamian equivalent of a Patagonia sweater vest asked: “Does it scale?”

Centralization scales

No one had a very good answer, so the charismatic leader seized the means of production. He called himself a king, got together with his buddies, and started creating propaganda that claimed it had always been this way. Similar stories played out in five other river valleys, with slight variations on the clothing and titles.

These new centralized governance structures were highly efficient. When you claim that God put you in charge and said you could turn most people into slaves, you can rapidly build some big things, like this:

But despite the scalability of centralized governance structures, they had a fatal flaw: unchecked growth without local self-limiting feedback loops eventually leads to collapse. The first civilizations “suffered from the vice that now threatens to overwhelm our own civilization in the very midst of technological advancement: purposeless materialism” (Mumford).

Sumerian Kings and Egyptian Pharaohs eventually learned the hard way that while you can technically build big great things via centralized control of power, it eventually hollows out civilization, overstretches physical resources, and collapses in under its own top-heavy weight. And then things get really bad for a while.

Classical Era: Greek city-states and the Roman Empire

Alphabets are interchangeable parts for the mind

Eventually, someone decides to take another crack at rebuilding civilization. Given that rebuilding civilization is a pretty daunting task that involves lots of people, it gets kicked off by the development of a new communication and coordination technology.

In this case, some sea-fairing traders called the Phoenicians invented the new technology: the phonetic alphabet. By using interchangeable letters to form higher order words and ideas, they massively expanded the capabilities of written language.

It was so effective at simplifying the transmission of ideas that we still use a version of their alphabet adapted by the Greeks and Romans. Instead of an exhausting and cumbersome list of hieroglyphic characters, alphabets simplified all words into a standard set of recombinable phonemes:

The Greeks picked up this technology, threw in a few vowels, and wrote some epic poetry. Armed with these new communication tools and leisure time derived from the comparative advantage of nautical trading, the Greeks decided to get together and play some games.

What Ancient Greek athletes and poets do on the weekend is what everyone else will do during the week in two hundred years.

Chris Dixon, more-or-less

From poetry to democracy

Before too long, the coordination advanced from poetry and games to self-governance. People spent a lot of time talking and thinking and decided that maybe this whole king thing was a crock of shit.

Eventually someone decided to actually do something about it and Greek city-states were born. While Greek city-states were far from perfect (read: lots of slaves), those who could claim citizenship enjoyed an impressively effective form of local democracy, particularly given the limited tools available to them.

Plato and Aristotle both argued that maintaining this pure form of self-sovereignty relied on limiting the size of cities. Plato rather prescriptively set the limit at 5,040 citizens (essentially the number of people who could be physically gathered in the same space to hear a speaker). Local democracy was effective, but it didn’t scale.

The art of reproducing cities

As popular emerging city-state destinations like Delphi, Olympia, and Kos outgrew their physical space, they made a radical choice: organized dispersal to new locations. This splintering allowed city-states to stay small and locally governed, avoiding the problems of large-scale centralization:

The good life, as they understood it and practiced it, depended on intimacy and small numbers… When the polis sent out a colony, they made no effort, it would seem, to extend either its territorial or its economic dominion: it sought only to reproduce conditions similar to those of the mother city. They had mastered the art of reproducing cities.

Lewis Mumford, The City in History

To better coordinate, these organizationally separate but culturally similar cities started creating federations. Ultimately, there were at least twenty different federalist structures across groups of city-states, ranging from loosely shared festivals to representative councils, treasuries, and courts. Some cities even airdropped dual citizenship to members of other friendly city-states.

The Romans show up

This was all fine and good until the neighbors started accruing centralized power and sacking the comparatively powerless individual city-states. While the Romans had started out with their own city-state federation, they eventually succumbed to the siren song of dictatorship and the apparent joy of looting and pillaging:

They say absolute power corrupts absolutely, and Romans made quick haste of trying to prove this. The Roman Republic Empire (thanks, Julius Caesar) followed up the sacking of Greece with a raucous 200 years of universal basic bread, free-love bath houses, and state sanctioned mass murder for sport and pleasure. Historical accounts of the populism of this period are absolutely bananas. In the words of Roman poet Juvenal:

Already long ago, from when we sold our vote to no man, the People have abdicated our duties; for the People who once upon a time handed out military command, high civil office, legions — everything, now restrains itself and anxiously hopes for just two things: bread and circuses.

But all good things must come to an end. Much like the collapse of the ancient centralized civilizations, an insatiable lust for physical comforts, violence, and growth ultimately overextends available resources, hollows out the base, and implodes. In a historical precedent that should shoot up some warning lights if you’re an American, the collapse of infrastructure and security started long before many people realized or acknowledged the end of the Empire.

Medieval Era: market towns and Christian kingdoms

Markets return power to the people

After the sacking of Rome by various barbarians and warlords came a predictable period of local fiefdoms, petty dictatorships, and limited progress. While many historians now reject the term Dark Age to describe the early Middle Ages, it probably wasn’t a great time to be alive for the vast majority of the population experiencing the violence inherent in the system.

In the High Middle Ages, starting around the 11th century, the self-sovereignty of newly emerging cities was reborn. It came not from an act of declarative political will but from the emergence of coordination technologies and the reality of markets. There was a Cambrian explosion of progress: new agricultural techniques unlocked the ability for peasants to form medieval communes, trade and crusades led to an increased exchange of knowledge, and technologies like paper and printing developed. Merchants and the towns they lived in suddenly had the coordination and communication technologies to regain an upper hand.

If you can’t beat them, tax them

Thousands of market towns obtained the crucial economic right to hold a weekly gathering for the exchange of goods and services, generally protected by a “market peace”. Some lords and priests predictably opposed the shift in power back to local, decentralized sovereignty of cities. But others embraced the trend, realizing that they had more to gain from extracting rents, tolls, and taxes from the increased commerce.

Unlike ancient Greece, which rooted its self-sovereignty in an intentional declaration of democratic principles, the emergence of local decentralized governance in the Middle Ages came organically, through the power of comparative advantage of the mutual benefits of trade:

Freedom, corporate equality, democratic participation, autonomy, were never fully achieved in any medieval town; but there was perhaps a greater measure of these qualities than had ever been exhibited before, even in Greece.

Lewis Mumford, The City in History

A new league of city-states

In the now familiar historical pattern, these merchant towns and guilds—freshly emboldened with money and power—began federating. In present-day Germany, hundreds of them banded together into the Hanseatic League, a defensive and commercial confederation of merchant guilds. Others gained relative independence as Free Imperial Cities and participated in representative governance via the Imperial Diet. At least a dozen other Leagues of Cities formed, but many were too weak to gain long-term stability.

As with the Greeks and Romans, the city-states of the Middle Ages met their match against the surviving centralized kingdoms and monarchical states. Some of the local powers of nobility that had avoided succumbing to self-sovereign market forces were now large empires of their own.

One pandemic and those pesky kings make a comeback

With the civilizational shocks of the Black Death, climatic changes, and famine (further compounded for the merchant class by a collapse of metal currencies), the federations proved too weak to survive. The remaining great kingdoms benefitted: “in the social disorganization that followed, power came into the hands of those who controlled armies, trade routes, and the great accumulations of capital” (CIH 346).

The Kingdoms of England, France, and Spain, the Ottoman Empire, the Grand Duchy of Moscow, and other proto-nation states emerged. Before long, centralized forces of military despotism were vying for power, in bloody wars that spanned generations. The hollowed out Catholic Church, burning the last of its divine legitimacy, started selling everything from the priesthood to permission slips to sin. It was understandably hard to keep the peasants happy. The whole place reeked of late-stage centralization.

Modern Era: American villages and nation states

Printing press: interchangeable parts for alphabets

And then a guy with a lovely beard came along and made it much easier to print things. Gutenberg’s printing press was the communication and coordination technology that unlocked another Cambrian explosion of historical greatest hits like the Reformation and the Renaissance.

Before long, everyone had books and maps and a yearning for adventure. For a peasant-cum-explorer, those pesky kings started looking less like subjugators and more like venture capitalists ready to fund a new wave of crazy startups: hopping onto a sailboat with a compass and seeing where you ended up.

New land, new decentralized cities

These sailboats went to many places around the world, where they rapidly claimed and stole everything in sight. Some of them, hailing from England, did their claiming and stealing in a particularly well endowed natural environment that they cleverly called New England. And there, on land freshly cleared of its native inhabitants, they repeated the cycle once again by establishing some towns, setting up local governance, and then copy-pasting them across the country side:

A democratic polity—and the most healthy and comely kind of environment, as long as it remained on a small scale. In every fiber this renewed medieval form contrasted not merely with its own authoritarian past, but with all the anti-democratic assumptions of the new baroque order. The continued growth of New England town by division of the central social nucleus into new cells, having an independent life of their own, recalled the earlier pattern of Greece.

Lewis Mumford, The City in History

From federalist papers to a national identity

As anyone vaguely familiar with American history knows, the cycle continued: local polities banded together into a federalist structure, it was generally too weak for longevity, and over time a more centralized form of government grew stronger and largely superseded it.

The core problem that the founding fathers of the United States grappled with was creating sustainable structures for federalism. While the Articles of Confederation and the US Constitution represent a brilliant balancing of powers to achieve this goal, they are failing under the weight of modern centralization. We appear to have once again reached peak centralization and late-stage capitalism.

America is now exhibiting the insecure self-glorification of late-Ancient Egypt, the crumbling infrastructure and mass appeasement of late-Classical Rome, and the shedding of moral authority of the late-Medieval Catholic Church.

The Information Age: new technologies for coordination and self-governance

On queue: computers, the internet, and blockchains have come along. Powerful new tools for coordination and communication that point a path towards reclaiming autonomy, self-governance, and decentralized cooperation.

The internet has given us an unprecedented depth of historical knowledge to learn from past pitfalls and the communication tools to work together across geographies. Computers provide technological leverage that allow individuals to scale impact to unprecedented levels. And now, blockchains solve the Byzantine Generals problem, allowing us to prove ownership and trust strangers online.

DAOs are the organizational output of combining these powerful new coordination tools. The historical precedents point clearly to what in happening next: small, localized experiments in self-governance are starting to emerge. We are carving a path towards reproducible local towns that can begin to form the mesh network of federated bottom-up governance.

But, as we’ve seen time-and-again, these federated structures are usually short-lived, weak, and fragile. From the irrigation cooperatives of ancient river valleys to Greek city-states to Medieval market towns to New England settlements, the promise of decentralized self-sovereignty has flourished and ultimately failed.

It’s unclear if this cycle can be broken. But we have an opportunity that comes only once in many generations: to use the latest technologies of coordination to take another crack at the problem. Only by deeply understanding the ways decentralized organizations evolve, federate, and centralize can we attempt to build versions that are more resilient to change.

Even if history is destined to repeat itself, the cycles should not be considered failures. Each ebb and flow of civilization brings new and improved forms of technology, governance, and prosperity that creates the foundation for the next wave. The cycles rhyme, but they also manifest in more complex and powerful iterations than their predecessors.

At the end of a cycle, civilization faces two choices: devolve into chaos, or plow forward with new tools for coordination to rebuild decentralized organizations from the bottom up. With eyes wide open to the historical precedents behind us, we are a generation that has been handed the tools to build the next iteration of local self-governance and the federated structures of decentralized cities that will form the basis of the next era of human prosperity. Let’s get to work.

Thanks to Cabin members Lauren Alexander, Étienne Fortier-Dubois, Timothy Roy, and Brett Pucino for edits and feedback on this essay.

[originally published January 5th, 2022]

• Economic value

• Governance rights

• Membership & access

Should all three of these roles be represented by a single token? Which of them are a better match for fungible and non-fungible tokens?

Economic value is the simplest to reason about. While NFTs can store tremendous economic value, they are inherently illiquid. Barter economies based on non-fungible assets inevitably get replaced by much more liquid economies denominated in a common fungible token. You don’t trade a house (illiquid, non-fungible), you sell it in your local currency (liquid, fungible) before buying a different one. Fungible tokens used by communities as a medium of exchange for goods and services represents an incredible opportunity to build new economies.

Governance rights face similar challenges with both fungible and non-fungible tokens: ownership naturally tends to have a power-law distribution so the question becomes: what do you do about whales? This recent essay by Vitalik paints a clear picture for why this will become an increasingly problematic issue for DAOs, both in terms of incentive misalignment and vulnerability to attacks.

There are essentially three ways to deal with the whale problem:

1. Accept them: treat the skin-in-the-game of asset ownership as a feature instead of a bug

2. Minimize them: use quadratic voting to reduce their power in votes

3. Eliminate them: implement some sort of proof-of-humanity mechanism to limit people to one-person-one-vote

If you go down the “accept” or “minimize” routes, fungible tokens are better suited for governance. If you want to implement tighter controls to minimize or eliminate whales, non-fungible tokens (particularly with sybil-resistant proof-of-humanity) can be better.

Membership & access can be managed with fungible and non-fungible assets, or any combination of them. Ultimately though, access becomes increasingly non-fungible or loses meaning. To take an extreme example, if a token is infinitely divisible and any quantity grants you access, a single token can grant access to everyone in the world. To get around this, ERC-20 gated communities like FWB have a minimum required number of tokens you have to hold.

In addition to being non-fungible by definition, NFTs have several other advantages for access and membership. PFP projects have demonstrated the advantages of an internal and external social signaling layers. Setting your profile picture to an NFT publicly shows your community affiliation and advertises the community to the world. The traits or rarity of your NFT may affiliate you with a sub-tribe of the community. Art with a greater sense of direct ownership can support complex social dynamics. As smart contracts, they also have the technical advantage of being programmed with more than just a numeric and/or social value.

So, what types of tokens should be used for which community purposes? As with most things in life, the answer is: it’s complicated! Each DAO needs to evaluate what type of community they are trying to build and determine the right combination of assets for economics, governance, and membership. But, generally speaking, we think it makes more sense for economies to operate on fungible tokens and access to operate on non-fungible tokens:

Printing money 🤝 Printing passports

This model isn’t novel—it’s how most countries already work. The two most important pieces of paper governments produce are money and passports. Money is the economic medium of exchange, passports represent rights of access and citizenship. Governance, in practice, ends up being dependent on both.

Cabin DAO currently operates a token-gated community accessed with $CABIN tokens. While $CABIN is a great internal medium of exchange, we want to explore membership, governance, and access privileges through non-fungible passport NFTs.

In the metaverse, nobody knows you’re a dog. But in reality, it’s helpful to have some basic systems of trust to protect the safety of strangers sharing a space. As a DAO operating a physical space, we are interested in exploring ways of granting verified access to that space using web3 technologies. It’s also fun to LARP as a nation state.

Here are some explorations of passports our artist guild has been working on:

Where passports could go next

NFT Passports offer a great way to combine aspects of membership, access, and identity verification. At the moment, the primary use-case for NFT smart contracts is as a transferable pointer to a jpeg on IPFS. But the possibilities of NFT smart contracts go way beyond that.

NFTs can hold roles, durations, access levels, and other editable data. For example, when someone is voted into a Cabin DAO residency cohort, we could allow them to mint a passport with the role of “Resident”, active for the duration of the Cohort. This passport could provide permissions both online and IRL.

Beyond tying roles and permissions to a member’s passport, we would also like to explore verifying ownership in-person. We imagine we could do this by adding a QR code to the passport. When scanned, we can send the owner a transaction that they have to sign, therefore verifying ownership in the metaverse and the meat space.

We would also like to add the ability for passport holders to set a passport image. Rather than a headshot, you could set it to your favorite PFP or another NFT of your choice. We think there’s value in being able to show membership in a standard format (like a passport) while being able to show individuality through customization.

We will continue to explore new use-cases for NFT passports and would love to hear from other DAOs interested in their own pseudo-state LARPs! Reach out to us @creatorcabins on Twitter or join our Discord server to continue the conversation

[originally published September 10th, 2021]

Inspired by Balaji's essay on The Location Stack for technological progressives and conversations at Creator Cabins this past week, we've been thinking through a rough sketch of the tech stack for decentralized cities:

Like a computer, a decentralized city has hardware and software. Hardware is the physical space: the land, the infrastructure, and the buildings. Software is the virtual layer: incorporation, governance, and social interaction.

This separation can allow decentralized cities to operate with compatible and interchangeable hardware and software architectures. If a local node in the network doesn't agree with decisions being made, they can spin up a new virtual governance layer. And if a physical location becomes an untenable place to continue living, the community can migrate to a new location without losing its governance and community.

We've learned a lot about these layers while building Creator Cabins, a community of remote cabins for independent online creators in the Texas Hill Country. And we're hopeful that we can be a node in a future decentralized city. Here's our hardware and software stack:

Hardware for a decentralized city

There are three layers of the physical hardware stack:

1. Land

2. Infrastructure

3. Buildings

Land: Develop multiple exurban locations

While the metaverse will increase our time spent in virtual realities, we still need to live in good old-fashioned physical reality. And if you want to be able to control your physical space, you need to own land.

Decentralized cities need land, but a key difference that separates them from conventional cities is that the land doesn't all need to be in the same place. Multiple physical nodes in the network allow flexibility and a plausible path to exit from any specific location.

Where should these nodes be located? The best living locations are usually determined through quality-of-life factors like:

• Cost: housing, food, taxes...

• Environment: climate, access to nature...

• Infrastructure: internet speed, airports, healthcare...

• Culture: access to dining, museums, concerts...

• Regulation: the ability to build, experiment...

We think that exurban environments provide the best trade-off among these factors for the first decentralized cities. They have relatively low costs and regulations and provide access to both nature and urban areas.

Unincorporated areas approximately 30 minutes - 1 hour outside of core urban areas seem to be ideal. For example, we've located the first Creator Cabins location in an exurban sweet spot, out in nature but only 45 minutes from downtown Austin:

Infrastructure: Go off-grid

To build a robust decentralized city node, off-the-grid technologies make up the core of our infrastructure tech stack. Solar panels, fixed wireless or satellite internet, a well, and a septic system can provide for an entirely independent retreat. Here are some pictures of our infra set-up:

There are probably several whole posts to be written about lessons learned putting in our infrastructure. But generally speaking, these technologies have gotten good enough that you can create a functioning outpost on your own infrastructure nearly anywhere in the world. With the proliferation of 5G and Starlink, internet access will soon be very fast in most places.

Buildings: design for community and flexibility

The physical buildings should be suited to the target citizens of your decentralized city. Architects have done a lot of thinking about this, and I'm trying to learn from their interrogation of the problem-space. I've personally found A Pattern Language to be a helpful guide, featuring hundreds of patterns (or principles) for the spatial design of integrated physical communities.

In our case, we designed the new cabin from the ground up for creator co-living and working. It features private bedroom suites with built-in desks and large connected public areas for social interaction and collaboration:

We partnered with Forte Shelter to pre-fabricate this cabin out of shipping containers. It's an inherently modular and extensible design. While pre-fab construction is not a panacea, it allows the flexibility to install similar cabins anywhere in the world and move existing cabins to new locations if needed:

Other flexible living options like Kibbo could allow for even easier migration of living spaces. Chenoe Hart has written about the ways in which self-driving technology may blur the lines between vehicles and buildings.

Software for a decentralized city

Decentralized cities run on three primary layers of the software stack:

1. Incorporation

2. Governance

3. Social

Remote incorporation

In Balaji's Location Stack post, he asks for people to help research options for jurisdictions friendly to tech progressives. Posts like this one walk-through options for living tailored to local governance preferences. But the underlying assumption is that the place you live is also the place you will be governed. This may always be true, but it doesn't have to be.

Traditionally, cities are in a single place and are incorporated locally. But much like startups now incorporate in Delaware regardless of where they do business, cities may start to incorporate in one place and then do their work in many places. Wyoming creating a legal structure for DAOs could make it a compelling option for incorporating a city.

I am not a lawyer, and state laws aren't going anywhere. But decentralized cities with remote incorporation could actually work elegantly within the existing structures of American federalism. The idea of powers left to local governing bodies is core to the structure of the US Constitution.

If local governance isn't tied to a single location, where does it happen? On-chain.

On-chain governance

Blockchains are a natural fit for decentralized city governance. They allow for open access, transportability, and a permanent and inalterable historical record.

One of my favorite characters of Texas history is Angelina Eberly, an innkeeper who protected Texas' government records with a cannon when Sam Houston (then the president of Texas) tried to forcibly relocate the state capital from Austin to his namesake town of Houston. There's a statue to honor her near the capitol in downtown Austin:

The Texas Archive War, as it was called, isn't possible when the governance records are located on a blockchain. Anyone can look at the records at any time from anywhere, and no one can change them.

We are experimenting with this at Creator Cabins by launching CabinDAO, a token-based governance structure for funding and awarding creator residencies at our cabins. Stay tuned for our $CABIN crowdfund on Mirror later this week!

Off-chain social communication & connection

While Web3 provides an ideal governance layer for decentralized cities, we can borrow lessons from Web2 in developing the social layer to tie decentralized communities together.

In the 20th century, people gathered in churches that represented shared cultural, social, and religious values. We need new ways of doing this in the 21st century:

The internet era has opened us to the possibility of curating and creating our own tailored practices and to looking to our peers for guidance as much as any teacher or authority figure

The Power of Ritual, Casper ter Kuile

As traditional religion falls by the wayside, people are developing new structures of accountability and community connection. The social layer of decentralized cities can provide these services.

Tools for social connection, community onboarding, reputation management, and member recruiting can all draw from existing Web2 models. However, because these tools will be owned and operated collectively by the community, they can be designed to exclusively serve the users instead of advertisers or other attention economy intermediaries.

A rough sketch

This post provides a rough outline of a tech stack, but take the implementation details with a grain of salt. They are our best guess from an initial exploration of the idea maze, not an ideal solution. They are tailored to the needs of independent online creators and may not be suited to other types of communities.

We'd love to hear from other people thinking about building innovative living spaces and organizing social and governance structures for them. After all, only through many experiments with hardware and software stacks will we start to evolve a network of nodes in the first decentralized cities.

Acknowledgments: thanks to Jon Gold, Zachary Fleischmann, Lauren Alexander for feedback, edits, and ideas

[originally published May 23rd, 2021]

This week, a group of internet friends tried an experiment: what happens when you meet online and then come together IRL to build something? What ideas emerge when you create the open-ended space to explore them collectively?

I've spent the past nine months building remote co-living cabins with high-speed internet outside of Austin, TX. Now that construction is complete, a group of indie hackers, online course creators, solopreneurs, and crypto-hippies came together to live and work in the cabins for a week and see what happened.

The result: great conversations, deepened friendships, and exciting new ideas. We've spent the past week dreaming about decentralized cities, crypto-creator funding mechanisms, and internet-enabled collaboration.

After a year of pandemic, we had forgotten some of the magical energy of open-ended living and working with other people. You don't need to schedule a Zoom call with an agenda and a hard-stop after 30 minutes. As you're drinking your morning coffee, you can end up down a rabbit hole, swapping book recommendations, productivity tools, and podcast episodes. Before long, we were asking each other questions:

• What would it look like to find like-minded collaborators online and then come together IRL to build?

• How could we support each other with the time and space to try new things?

• How do we make physical space and creative resources accessible to anyone with a good idea?

Our answer to these questions is crowdfunding collaborative creator residencies. The idea is to allow anyone to propose a project, recruit collaborators, crowdfund a stay at Creator Cabins, and get together in-person to build it.

We have a sketch of what this could look like, and would love your feedback.

$CABIN

$CABIN is a token tied to a one-night stay in a room at Creator Cabins. If you have an idea, you can raise $CABIN, find collaborators, and then get together IRL to build it. Here's how it could work:

1. Submit a proposal

2. Recruit collaborators (optional)

3. Get funded

4. Come build it

Submit a proposal

The first step is to describe what you want to build. It could be a project you're already working on, a new idea, or an open-ended R&D plan. It could be a piece of digital art, or a physical installation you'd like to build on the Creator Cabins property.

Recruit collaborators (optional)

Who else do you need to bring this idea into reality?

Maybe you're an engineer and you want to recruit a designer and a product manager to build software. You're an online course creator who wants to collaborate with a course manager. You have a podcast and want to bring out a few guests to do a deep-dive series. Or you just want to build something independently.

As part of your proposal, you can write role descriptions for each collaborator you want to recruit, evaluate applications, and add them to your project.

Get funded

Let's say you want to bring together a team of four people for a one week stay to build your project. You set your crowdfunding goal at 28 nights worth of $CABIN. Once you raise enough for the project, you and your collaborators can schedule a stay at Creator Cabins.

Come build it

The cabins are turn-key spaces designed from the ground up for co-working and co-living in a beautiful natural environment. You show up, meet the team, and start building.

Like other crowdfunding platforms, you can offer rewards to the people who made your stay possible. Perhaps you're producing a piece of open source software people are interested in funding. Or you are going to mint an NFT at the end of your stay and share a split of the proceeds with your backers.

As a proof of concept for a collaborative residency, we are using Creator Cabins this week to work on this idea. We're building a roadmap and would love your thoughts, feedback, ideas, and involvement. If you're reading this, you're part of the community we want to meet and collaborate with. Reach out online, and hopefully we'll see you out here in the Texas Hill Country.

Find us on twitter @CreatorCabins or reach out to us individually: @JonathanHillis, @JonGold, and @ZFleischmann

[originally published May 14, 2021]

This is the 1st essay in a 5-part series on funding models for the creator economy

1. Funding the creator economy

2. Avoiding the venture trap

3. Cash rules everything around me

4. Shared income and bespoke finance

5. NFTs, $GME, and the crowdfunded Cambrian explosion

The idea of the American Dream emerged in 1930s, as the United States was reborn from the ashes of the Great Depression. The unemployment rate was over 15%. Most Americans didn't own their homes. It was a time ripe for hopeful aspirations of a better tomorrow.

Within a decade the USA was back on its feet, starting to realize the dream born 10 years earlier. Because nothing embodied that dream better than owning a home, homeownership rates exploded over the next 30 years.

But it was more than just a dream that led to this outcome. Middle class home ownership was possible because of a financial innovation: the 30-year fixed rate mortgage. Before standard long-term/low-rate home mortgages, only wealthy people could afford to put down the upfront cash to buy a house. While accessible mortgages have a mixed history, they undeniably democratized access to home ownership for millions of Americans.

Now home ownership levels have peaked and are starting to drop again. Homes have become unaffordable, Millennials want more lifestyle flexibility, and more of us have student debt that precludes taking on more loans to buy a house.

So there's a new American Dream: building a profitable, sustainable, remote business that can be run from anywhere, scales nicely, and prints money. This is the dream at the core of the creator economy. It is a dream shared by indie developers, TikTok influencers, online course creators, OnlyFans stars, NFT artists, YouTubers, freelance designers, newsletter writers, and Twitch streamers. This dream could offer a path that is more democratized, accessible, flexible, and financially beneficial than home ownership ever was. And it's no longer just an American Dream, it's a global phenomenon.

Li Jin has made a compelling case that this new American Dream can also help create a new middle class. But in order to make this dream accessible, we need new financing structures to democratize access to the capital for people to start their own businesses. What democratizes access to capital for online creators the way that 30-year fixed rate mortgages did for home owners?

This originally started as a single essay cataloging innovative funding models to answer that question. It blossomed into a five-part series because I kept uncovering new pieces of the puzzle that suggested these alternatives to venture capital might be much more important in the future. We are entering a Cambrian explosion of funding models that blur the lines between previously clear categories like customers and investors.

Now, after completing this series, I feel more confident and optimistic than ever about where we are heading and how the internet will enable the creator middle class and the new American Dream. The remaining four essays deep-dive in to the current and future state of creator funding:

https://www.meatspacealgorithms.com/avoiding-the-venture-trap/

https://www.meatspacealgorithms.com/revenue/

https://www.meatspacealgorithms.com/shared-income/

https://www.meatspacealgorithms.com/nft-gme-crowdfunding/

[originally published Mar 17, 2021]

This is the 2nd essay in a 5-part series on funding models for the creator economy

1. Funding the creator economy

2. Avoiding the venture trap

3. Cash rules everything around me

4. Shared income and bespoke finance

5. NFTs, $GME, and the crowdfunded Cambrian explosion

Venture capital is the shiniest of the capitals. You, the founder, get a big check at an eye-popping valuation. TechCrunch writes an article saying you are the coolest kid on the block. Your mom is proud. And if the whole thing goes belly-up, you just walk away debt-free.

I am not here to tell you not to take that deal. In fact, I spent six years working at a venture-backed company and I have invested in over 30 companies on traditional venture-style terms. I am not anti-VC! But I do want to share why you might not want that deal and why alternatives might be a better fit for you, particularly if you want to build independently online.

Why venture capital might not be for you

Much virtual ink has been spilled on why not to raise VC, so I'll stick to a brief summary of the highlights:

1. With the incredible leverage individuals can get from cheap online software (what Tyler Tringas calls the Peace Dividends of the SaaS Wars), you probably don't need a large chunk of capital to get started.

2. VCs want your business to go to the moon or die quickly so they can move on to other bets. If you want  something a little less bimodal, your values aren't strongly aligned.

3. The venture model takes permanent partial ownership of your company in exchange for an investment. People compare it to getting married. If you'd prefer to maintain long-term ownership and control of your company, consider alternatives. This is particularly problematic for individual creators with diverse revenue streams. What does it even mean to own 10% of a creator? Doesn't sound good.

4. Most companies in most industries are not going to be unicorns. The early days of tech were defined by huge, winner-takes-all global markets: search, social media, smartphones, silicon chips. Now, we are entering the deployment phase of the information revolution, where software is replacing existing systems in essentially every industry. Everything from bingo card creation to niche newsletters can be profitable businesses—just not venture backed ones.

The astute reader will notice that while people have been talking about these reasons for years, all of them are becoming much more true. There will always be bloody, monopolized multi-billion dollar markets that make sense for VC. But increasingly, new markets will be niches. Venture capital is a particularly bad solution for the online creators, solopreneurs, and indie developers that play well in these niches.

Misaligned incentives

If this is whole creator economy thing to be a big deal, why doesn't venture capital change to fit the needs of these groups? It can't. While venture capital will undoubtedly help create some of the biggest platforms of the creator economy, it has no business funding indie companies. I mean that literally.

Indie.vc, the leading innovative model for startup funding, called it quits last week. While it's a sad set-back for alternative funding models, there was something peculiar about the announcement. These two sentences stood out to me:

I have no doubt that in 4–5 years we’ll see our Indie companies posting comparable results as our previous funds that have generated 5x+ net multiples for our LPs.

[indie.vc] cost us 80% of our LP base. Unfortunately, as we’ve sought to lean more aggressively into scaling our investments and ideas behind an “Indie Economy” we’ve not found that same level of enthusiasm from the institutional LP market.

Wait, a second. If Indie has helped great companies generate solid returns, why did they lose 80% of their investors? The answer to that question gets at the heart of venture capital: it doesn't exist because startups need it, it exists because large investors (LPs) need high-risk portfolio allocations.

A fundamental premise of portfolio management is risk allocation. You want most of your assets in relatively safe investments and a small percentage of your assets in high risk investments. If the high-risk investments tank, you lose a couple percentage points of returns. But if they succeed, they can return more money than all of the other asset types. Low downside risk, high upside potential.

This chart shows why large pension funds and private equity firms need to have VC allocations to stay competitive:

Venture capital has been a tremendously useful tool for fulfilling this need (and creating a huge consumer surplus in the form of hundreds of large tech companies). It's a perfect asset class for solving a particular problem in portfolios & creating some incredible things that wouldn't otherwise exist. But it is not a one-size-fits-all funding model. It is designed for outliers—in other words, not most businesses!

Alternatives

The greatest trick venture capital ever pulled was convincing the world other funding models didn't exist. When people talk about fundraising in the context of startups, they almost always mean selling equity to venture capitalists.

But when I say "funding model" I am generally not talking about who you raise money from. Venture capitalists, angel investors, incubators, syndicates, friends & family rounds — ultimately, these are all the same dilutive equity-based funding model, just raised from different groups of investors. I am talking about the financial structure of the fundraise.

The remaining three essays in this series cover the major categories of these alternative financial structures:

https://www.meatspacealgorithms.com/revenue/

https://www.meatspacealgorithms.com/shared-income/

https://www.meatspacealgorithms.com/nft-gme-crowdfunding/

[originally published Mar 17, 2021]

This is the 3rd essay in a 5-part series on funding models for the creator economy

1. Funding the creator economy

2. Avoiding the venture trap

3. Cash rules everything around me

4. Shared income and bespoke finance

5. NFTs, $GME, and the crowdfunded Cambrian explosion

Most of the time when I see a pre-revenue company trying to fundraise, my first question to them is why they aren't trying to get money from customers first. If you are developing life extension drugs or a social network or rockets, then you are probably going to need upfront capital before you have revenue. For almost everyone else, the easiest way to learn if your idea is working is to ask people to pay for it.

The best form of funding is revenue from customers, because revenue can be converted into any other funding model (or none at all!). Revenue provides you freedom, optionality, and control. My startup, Creator Cabins, is self-funded. Maybe I will raise venture capital someday to accelerate growth. Maybe I never will. Because my wife and I own 100% of the company, it's up to us.

It is easier than ever to make a landing page, hack together some no-code tools, or start a newsletter. In 1999, you had to literally buy servers, stick them in your closet, and then be prepared to do a lot of programming and computer networking if you wanted to set up a website. Now, you can do it for free in under an hour. This is what Tyler Tringas calls the Peace Dividends of the SaaS Wars:

• Quick start open-source development frameworks (Ruby on Rails, Laravel)

• Scaleable server infrastructure (Heroku, AWS)

• A full-stack suite of tools for the basic operations of software businesses (SaaS for help desk, email & social media marketing, FAQ, billing, analytics)

• Free distribution networks from piggy-backing on fast-growing app stores and platforms

• No-code platforms allowing entrepreneurs to reduce their engineering costs to get to market (Zapier, Webflow, Airtable, Shopify)

• Battle-tested methodologies for finding and validating businesses (Lean Startup, Traction)

In Tyler's words: "The net effect is that it’s easier, cheaper, and faster than ever for a solo founder or small team to get a software product live and creating real value for customers with little to no outside investment."

It's not just that it's cheap and easy to get something live quickly—it's also a better way to build a business. By co-creating an audience-first product you can find product market fit *along with *your customers. Together, you build a community and a product. The two become deeply intertwined and each makes the other more valuable.

The best part is: once you have significant revenue, investors will be tripping over themselves to give you money on much better terms.

Revenue advances

Revenue immediately unlocks other funding models that are directly tied to how much cash your company can produce. If you need more cash now, you can get a revenue advance, which is essentially a loan secured by future revenue:

• Payment vendors like Stripe and Square have access to lots of data about how much money businesses are making. They've put this data to use to develop easier, faster, more automated lending options than banks can manage.

• Pipe, which recently closed a large funding round, takes a similar approach by integrating with a company's accounting, payment processing, and banking systems and then automatically turns that into loans that get sold on a marketplace.

I expect this model to quickly extend to the entire stack of creator economies.

When I led the Shoppers product team at Instacart, I learned that people became Shoppers for two primary reasons: flexibility and liquidity. Being a Shopper gave people control over their time and the ability to earn money fast. One of the most universally enjoyed features of the gig economy is the ability to work when you want and then cash out your earnings immediately.

Uber already has a debit card for Drivers. It is a natural extension for gig companies to start providing banking services for gig workers. Much like Square and Stripe, gig companies know a lot about people's earning habits on their platforms. They can forecast future income accurately and provide much lower-risk, cheaper loan services than are otherwise available. The payday loan industry offers predatory high rates to people without other options. The gig economy could provide a more equitable path to helping people cover short-term expenses.

But this won't stop with the gig economy. This trend shows that any company that controls monetary transactions for you also knows enough about your revenue to offer you attractive loans. You could imagine someone offering revenue advances to YouTubers by analyzing their subscribers and views and predicting their ad revenue.

Crowdfunding

Another way to get revenue now is to ask people to pre-order your product. This was the innovation of crowdfunding platforms: gather revenue now and use it to develop your product. For most of the last decade, crowdfunding generally meant either:

1. Project-oriented one-off raises on platforms like Kickstarter

2. Generating recurring fan revenue from platforms like Patreon

These platforms have helped revive Kevin Kelley's vision of 1,000 True Fans, where a small number of people can create meaningful income for independent online creators. There are plenty of good guides for getting up and running on these platforms. Here are a few:

https://medium.com/@lauramoffat/the-secret-to-running-a-successful-kickstarter-campaign-b7144e947304

https://blog.patreon.com/5-data-backed-tips-1st-month-patreon

As we'll explore in the last essay in this series, crowdfunding is where the difference between revenue and funding starts to get blurry. Crowdfunding is eroding the distinction between customers buying your product and investors investing in your business.

From here, we'll explore how complex bespoke finance and weird Web 3.0 technologies are creating new funding models for online creators. In the remaining two essays in this series, let me show you how deep the rabbit hole goes...

https://www.meatspacealgorithms.com/shared-income/

https://www.meatspacealgorithms.com/nft-gme-crowdfunding/

[originally published Mar 17, 2021]

1. Funding the creator economy

2. Avoiding the venture trap

3. Cash rules everything around me

4. Shared income and bespoke finance

5. NFTs, $GME, and the crowdfunded Cambrian explosion

In between debt and equity lies an idea that is quietly changing education. It's called an income sharing agreement or, for less of a mouthful, an ISA.

Ironically, incoming sharing agreements spent their first 50-years of existence as an obscure academic idea. Milton Friedman briefly explored the idea in the 1950s, and Yale explored it in the 70s, but it wasn't until the 2000s that people started seriously making ISAs to fund education.

Income sharing means that you agree to pay someone a percentage of your future earnings for some period of time. You only have to pay back the ISA if you make enough money, so there's low downside risk for the borrower and highly aligned incentives for the loaner.

Universities now frequently offer ISAs as an alternative to traditional student debt. But software bootcamps and other trade schools are even better use-cases, because the good ones can consistently help you make a high salary after graduation. Lambda School popularized the model. Lambda trains people as software engineers and has been so successful with it that their founder does things like this:

Let’s play a game.

Hiring software engineers? Reply to this tweet with “I’m in.”

I’ll send you 5-10 Lambda School grads. You pick one for a one month Fellowship. Lambda pays their salary.

If you don’t hire them I’ll personally donate $2,000 to a charity of your choice.

— Austen Allred (@Austen) March 5, 2021

But this model is based on their ability to consistently land you a job at an existing company. While it's a great gateway to financial freedom, it doesn't solve for independent online creators, who face less certain outcomes.

Creator ISAs

The opportunity is much bigger than education. I think income sharing could help accelerate the creator economy. Investors can take a bet on up-and-coming creators and share in their upside. Creators can get started knowing that they will either succeed or walk away debt-free.

It's a model that has already proven successful in other industries with independent workers—like baseball! Many minor league players have potential to make it to the majors, where salaries go up dramatically.

Minor league average: $10,000. Major league average: $540,000.

So, some of them make a deal with Big League Advance. They get cash up front while in the minors, and if they make it big they share a percentage of their income. In industries with a wide range of financial outcomes, this can be a win-win deal for people early in their careers with high potential.

The creator economy is built on personal monopolies, so it will operate on a power law distribution. Whether you look at YouTube subscribers, Twitter followers, or the top earners on OnlyFans, it is going to look something like this:

Income sharing agreements can help new creators earn a living while they are trying to break out from the pack. If they do, they happily share the upside with the investor who believed in them early. And if they don't succeed, they can walk away without any debt.

I am such a believer in the ISA model that I started the Creator Fellowship to fund independent online creators. Here's how it works:

1. We invest $20,000 in you over the course of a year.

2. You quit your job & create full-time.

3. We provide support, mentorship, and a community.

4. In exchange for our investment, you share 20% of your net earnings above $20,000 for 5 years, capped at 3x the initial investment.

5. If you want to start a company and raise money from outside investors, we will help you do it & re-invest the ISA into your business.

Early results are promising. David Vargas quit his job six months ago and joined the fellowship to build open-source software on top of Roam. He had never made money independently online, and he's already generating so much revenue that he's on the cusp of his first re-payment. I think there will be millions more people who follow his footsteps.

Shared Earnings Agreements

You can think of a Shared Earnings Agreement (sometimes called a SEAL) sort of like an ISA for businesses. Earnest Capital pioneered the SEAL, which is an alternative to the SAFE commonly used in early-stage VC fundraising [full disclosure: I am an LP in Earnest's fund because I believe in their model].

Earnest makes an upfront investment, and then the company pays back a portion of "founder earnings" over time, up to a multiple of the original investment. Much like the Creator Fellowship, Earnest also has the ability to reinvest their income share into equity if the company chooses to do an equity round. The intent is to create strong founder/investor alignment, allow founders to maintain full company control, and still provide strong potential upside to investors.

This model works particularly well for the indie economy:

• B2B SaaS targeting niches or specific industries

• Developer tools

• Remote tools for remote teams

• The “picks & shovels” of online entrepreneurship (software that enables building online businesses)

• Online education and personal development communities, membership models, platforms

The Shared Earnings Agreement is a bit complicated, as any new finance innovation typically starts. But I hope that financial innovations that help people get capital now and walk away debt-free if it doesn't work out will become more common in the future. They could be an incredibly effective tool for leveling the playing field in the creator economy and giving more people a chance to take the leap.

Complex bespoke finance

Shared income or earnings is complicated because it doesn't fit very cleanly into existing buckets. Financing, in the broadest categories, is generally classified as either debt or equity:

Debt financing involves borrowing a fixed sum from a lender, which is then paid back with interest.

Equity financing is the sale of a percentage of the business to an investor, in exchange for capital.

But the reality is more of a spectrum:

Complex bespoke finance is where innovation usually happens. But it's also been known to contribute to global financial meltdowns, so best to proceed with caution. Assuming we aren't trading creator default swaps, it's worth looking for other ways that finance can help creators that haven't been explored yet.

After all, financings are just contracts and software is reducing the friction of contracts. There could be interesting opportunities to offer ISA-like structures with smart contracts and DeFi tools. People are already selling futures of their time in Human IPOs. So far, we've talked about innovative models within the context of existing financial structures and legal contracts. In the last essay, we'll explore where things start to get really weird—financial innovation in permissionless open markets:

https://www.meatspacealgorithms.com/nft-gme-crowdfunding/

[originally published Mar 17, 2021]

1. Funding the creator economy

2. Avoiding the venture trap

3. Cash rules everything around me

4. Shared income and bespoke finance

5. NFTs, $GME, and the crowdfunded Cambrian explosion

A return to weird

The early 2000s were a heady time for people still hanging out on the internet. Sure, the dot com bubble had crashed, but that just meant the speculators and suits were gone and it was back to building. It seemed like anyone could build anything.

I came of age during this period, the final innings of Web 1.0. My two distinct memories of 5th grade are watching the Twin Towers fall on TV and building my first website on GeoCities. I spent most of middle school lurking on Digg, Reddit, and some Science Olympiad message boards. The internet was a weird place and I was a tremendous dork. I felt right at home.

By the time I got to high school, something had changed. There was a new website called Facebook, people started checking it on their iPhones, and anyone not using Google was a technological neophyte. Over the next decade, tech became Big Tech. These companies defined Web 2.0 by building monetization wrappers on the original protocols of the internet:

While these platforms have created tremendous consumer surplus, they've also created centralization. And that leads to monoculture. Like the 1950s, today's big tech platforms are decidedly not weird. They are beholden to advertisers and regulators and armies of employees. Google changed their position on being evil. The free speech wing of the free speech party banned the President of the United States from tweeting.

But the internet is getting weird again. Decentralized Web 3.0 technologies, undermined gatekeeping institutions, broad internet literacy, and the magnetic force of memes are ushering in a new era. The web is returning to its roots of permissionless, open, unpredictable exploration. The Great Weirding is here, and it has its eyes set on reshaping finance.

Gamestonks and memes

When you open up everything to everyone, a sort of Darwinian process kicks in. It explores all possible evolutionary idea paths and amplifies the stories that resonate with people. The weirdest corners of the internet become the seeds of mainstream culture.

Think about the recent shenanigans of the stonk market. A rowdy band of internet scoundrels short-squeezed hedge funds by buying Gamestop stock. Were they buying $GME because they believed in the underlying business? Out of a sense of childhood nostalgia? As an anti-Wall Street statement? A speculative investment? Or just for the lulz?

Yes, yes, yes, and yes. The internet turns information consumers into producers and producers into consumers. It creates tight feedback loops between the two, leading to the rapid development and proliferation of memes. If you can turn it into a potent meme, you can turn it into an investment decision:

But the phenomenon of $GME played out on the landscape of the legacy financial systems of the stock market. As a result, it couldn't handle the weirdness and shut down. What will this weirding of finance look like in the context of Web 3.0?

Crowdfunding consumers become 'skin in the game' investors

Paul Ford claimed that the fundamental question of the web is:

"Why wasn't I consulted?"

The Web 3.0 answer to this question is that you get to vote on everything. And you vote not just with comments or likes, but with money. This creates skin in the game, where your vote carries the weight of a bet. And by being an early supporter of a successful product, service, movement, meme, or community, you get financially rewarded. People are voting financially on the future they want to be built.

Early crowdfunding platforms like Kickstarter and Patreon started to revive Kevin Kelley's vision of 1,000 True Fans, where a small number of people could create meaningful income for up-and-coming independent online creators. But ultimately you were buying a product, you were a consumer.

In 2017, people started crowdfunding ICOs (initial coin offerings) by creating and selling a cryptocurrency tied to a project or company. Consumers became investors, sharing in the upside of successful projects. It's crypto, so there was an exciting bubble that inevitably popped. Now the SEC has a website where you can see all of the people they are criminally charging with digital asset fraud.

But this genie wasn't going back in the bottle. The SEC recently released new regulations for crowdfunding that opened the doors for regular people to invest up to $5M directly in companies. Republic is the early leader in this space, where companies can sell equity along with traditional Kickstarter-style perks. The new regulations went into effect this week and Gumroad became the first company to raise $5M via equity crowdfunding:

It's not a coincidence that a creator economy platform was the first to pull this off. Equity crowdfunding will let creators turn fans into investors and investors into fans. But it still involves selling ownership of your company, which can have some of the downsides covered in Avoiding the venture trap.

Web 3.0 is just starting to provide new models to address this concern. John Palmer paved the way by crowdfunding one of the first online essays using something called an $ESSAY token. It gave him an upfront payment to write an essay, but it also created a speculative financial asset for the essay's backers:

Customers = investors = fans = voters = artists

And that brings us to the topic of the month, NFTs (non-fungible tokens). Others have written much better introductions to NFTs, so check them out if you need a primer.

NFTs are essentially a way for creators to crowdfund the ownership of atomic units of work. They feel like someone took all of the other funding models and put them in a blender. NFTs can be:

• A way for creators to directly monetize their work

• A collectible for die-hard fans

• A speculative vehicle for investors

• A representation of underlying physical assets

• A statement, conversation, or performance

• An abstraction that turns the distinction between types of financing on its head

To commemorate the Great Weirding of the American Dream, I made my first meme. And I minted it as an NFT. I don't make the rules for writing about this topic, I just follow them. It features the 1950s-era piece of American Dream advertising from the first essay in this series, repurposed for 2021:

If you were to buy this NFT right now, are you a customer of mine? Are you investing in my art? Am I just engaging in performance art to make a point?

Does the distinction really matter?

The next wave

NFTs had a big mainstream moment when they rocked the art world last week with a $69 million sale of digital artwork. We now seem to be entering the phase of the cycle where CNN gets in on the action and tries to explain what's happening to people. We seem to be in somewhere between the Third and Fifth Crypto Bubble, depending on how you count them:

I don't use the word bubble derogatorily. At this point, I consider bubbles as more of a feature than a bug of the crypto economy. They tend to create big periods of excitement and advancement in the core ideas of decentralized currencies, companies, communities, and creators.

If you ignore the noise and squint a little, crypto bubbles are good times to glimpse the future. After the bubble inevitably pops, it leaves crypto with a "new normal" and seeds for the next wave. For example, the first whiff of the NFT mania currently sweeping the internet came back in 2017, during the last major crypto bubble. Most of the last bubble was focused on meteoric rises in the value of bitcoin and ether. But there was a short-lived side show called CryptoKitties, which was the first NFT on the ERC-721 protocol. People were buying and selling and breeding cute little cats like this in late 2017:

https://www.cryptokitties.co/kitty/2019

And then most people sort of forgot about the whole thing until now. So what seeds are being planted this cycle that will pop back up in the future?

The whole point of weirdness is that it's surprising, uncertain. It's hard to predict where it goes next. But there are a few experiments happening in the corners of the current crypto bubble that I find particularly interesting:

Some crypto tokens are being used as keys to online spaces. An early pioneer in this space is the Friends with Benefits community—you have to hold the $FWB token in order to access their community discord.

In other cases, creators are minting social tokens for their fans, which can then be used as a sort of internal currency within the fan's community. Roll, an early leader in this space, just passed through the crypto right of passage of being hacked.

For more on social tokens, check out this great discussion from the Means of Creation podcast/clubhouse, which explores where social tokens could go in the future.

Decentralized autonomous organizations (DAOs)

DAOs are a governance structure that is controlled by software rules instead of human decision-makers. They have been around since the earliest days of blockchain-based smart contracts.

The first DAO was set up in the early days of Ethereum, promptly hacked, and caused a crisis that led to a hard-fork in the Ethereum blockchain. But despite this inauspicious beginning, you should never underestimate new ways of coordinating groups of people.

Companies, cooperatives, and markets have all changed the world by finding ways to overcome collective action problems. Organizations like MolochDAO are now navigating the game theory challenges required to unlock a new way coordinating people and resources to make investments in the future.

Decentralized finance

Also known as DeFi, this is an umbrella term for the re-creation of financial products and services on blockchains. Some of the most interesting use-cases include:

• Loans based on smart contracts

• Stablecoins tethered to existing currencies

• Prediction markets as an alternative to polls

• Composable building blocks for complex financial products

One particularly weird example is $WHALE, a cryptocurrency backed by a vault of NFTs. Owning the $WHALE coin allows you to fractionally invest in a basket of rare digital art and collectibles:

Metaverses

That $69 million piece of digital art was bought by a pseudonymous internet character called MetaKovan. He dropped in on the popular Good Time show on Clubhouse last week to explain his purchase and how he thinks about the future of NFTs. The part that jumped out to me was that he revealed himself to be a major investor in virtual land inside metaverses. Stuff like this, a chunk of undeveloped virtual property inside The Sandbox metaverse:

I didn't really grok this until I read Packy McCormick's excellent piece on the topic, *The Value Chain of the Open Metaverse. *He makes the case that virtual objects (already a multi-billion dollar revenue source within the gaming industry) are a perfect fit for NFTs because they allow for interoperability between metaverses:

To us olds, and I’m including myself here, it seems crazy that people are willing to spend large sums of money on outfits for their video game avatars. In 2018, over $1 billion of Fortnite’s $2.4 billion in revenue came from the sale of skins (outfits) or emotes (dance moves). In 2019, League of Legends generated $1.5 billion in revenue from skins.

The watershed moment for the Metaverse will occur when there's an event that takes place simultaneously across multiple AAA platforms, where players can walk from one to the other as the same avatar, wearing the same skin. Imagine buying an NFT skin from Prada, wearing it to a concert in Fortnite, and then popping into a different version of the same concert in Roblox maintaining the same identity.

Cambrian explosion

Well, I started off with the simple goal of cataloging funding models for online creators and somehow we've ended up talking about decentralized metaverses. Five essays and over 5,000 words later, the only thing that's clear to me is that this is the beginning of a Cambrian explosion of the creator economy and the financial structures that will fund it.

I have no idea how these things are going to recombine and evolve. But I feel confident that it's going to result in the creator economy equivalent of some weird tentacled sea creatures. And I believe now more than ever that the 2020s are going to be an incredible time to build products, services, communities, and content independently online.

See you in the metaverse.

[originally published Mar 17, 2021]

Austin, with its natural beauty, weird-chill vibe, flagship university, killer BBQ, and long history of technology companies is an obvious choice. Francis Suarez, with the hustle of a direct-to-consumer CBD salesman and the political savvy of a future president, has been miraculously tweeting Miami into existence as a tech hub. And, of course, you can never count out San Francisco to be the next San Francisco.

We know the players and we know the game. It's comforting. And the best part is, there are no wrong answers! Just wrong questions.

Wrong questions

People still refer to the "tech industry" as if every industry wasn't getting eaten by software. I advise and invest in companies tackling problems in education, real estate, banking, media, fitness, construction, and healthcare. The only reason I can help this range of companies is because they are all using the same principles of software development to accomplish their goals.

Internet-enabled software is in the deployment phase, where it becomes so ubiquitous as to be second nature. Asking which city will be the next tech hub is like asking which industry will use software.

Of course, all industries use software. And any city that wants to stay economically relevant will need to figure out how to attract people that build it. San Francisco and Austin and Miami are some of the best positioned places to do this. But the "tech industry" is too big for one city.

What's a better question?

The most interesting industries are not the ones software is eating, but the new ones it's creating. Things that weren't possible or economically feasible before. Autonomous cars. Virtual reality. Social networks.

So, what new types of cities might the internet create?

Cities and ambitions

To answer that question, we need to understand what makes cities great. Paul Graham wrote an essay about this that starts:

Great cities attract ambitious people. You can sense it when you walk around one. In a hundred subtle ways, the city sends you a message: you could do more; you should try harder.

Humans are mimetic creatures—always looking at what others are doing, comparing ourselves to it, and copying their behaviors. Bigger cities, by nature of their size, will have more ambitious people. But it's once ambitious people are surrounded by other ambitious people that something interesting happens. They start feeding off of each other.

Geoffrey West, an academic who left the field of physics to explore new pastures, has quantified this effect by showing a strong relationship between the population of cities, higher wages, and more creative people:

But this chart is on a log scale, which is just a fancy way of saying it’s an exponential relationship: interesting people attract more interesting people. Once there's a critical mass of them, other people start noticing and emulating them. This creates even more novel ways of doing things, and more people get in on the fun. Pretty soon you have a culture.

People vs. places

West goes on to explain that when you think of cities, you think of the boulevards of Paris and the Eiffel Tower, you think of the skyscrapers of New York. But that's just the stage. It's the platform on which the social networks and information and creativity is taking place. It facilitates the daily social interaction that creates ideas, wealth, and stimulates innovation.

It's not the place that matters, it's the density of creative ideas and interesting people. When I think about where I meet interesting people, it’s increasingly not in a physical place at all. It’s on Twitter threads and Discord channels and WhatsApp groups.

In the Facebook era, people met in person and then became “friends” online. But it makes more sense to do the opposite: find your friends online and then meet up in person. There are at least 1000x more people to choose from online, and they conveniently gather into easily identifiable tribes. People in the dating pool are the most highly motivated searchers for strong new connections, and they have figured this out in droves:

But the physical world still matters. Once people connect, they need meatspace to come together, break bread, be present, and immerse themselves in the environment. Maybe someday there will be virtual realities with the same bandwidth as the real world, but for now, you can’t beat good old fashioned reality.

So how do we get the best of cities and the best of the internet? What if creative, ambitious people found each other online and then decided to get together in the same place?

Let's build

This is the premise of Creator Cabins, the community of remote cabins for remote workers and independent online creators that we're building in the Texas Hill Country, about 45 minutes outside of Austin. Here’s a view from the deck of our first cabin:

Like any good startup, it's a simple premise with an ambitious goal. For now, it's a place to get together IRL with your internet friends. In the future, I hope it becomes one node in a network of decentralized properties, owned and operated by small groups of independent online creators and entrepreneurs. The density of creative energy of San Francisco or New York, but wherever and whenever you want it to be. If you were designing a 21st century city from scratch, why would you put it all in one place?

Follow this idea to its logical conclusion and you can end up in some wild territory. Neal Stephenson wrote about independent distributed city-state Burbclaves in his mind-bending 1992 science fiction classic Snow Crash. James Dale Davidson envisioned not just sovereign suburbs, but sovereign individuals around the same time. Davidson's buddy Peter Thiel helped popularize the idea of Seasteading more recently. And now, Balaji Srinivasan is talking about exiting to cloud-first Network States.

Many of these visions make my network of cabins look woefully unambitious by comparison. But my goal isn't some grand libertarian exit from society. I just want to take more of the magical serendipity of the internet and bring it into high-bandwidth reality. First, we'll have cabins. Then there will be towns. Eventually, I think we'll have a network that really does resemble a city.

I won't pretend to know exactly where this is headed. The best cities aren't planned, they evolve. But if Creator Cabins has an ambition, a central cultural tenet, I hope it is one of people gaining greater independence over their work and lives by earning a living online. I am inspired by independent online creators, writers, artists, and entrepreneurs—so I'm making a place for us to gather and meet each other and create even better things.

If you’d like to be a part of it, come visit! And if you want to create your own node in the network, I’d love to help you get started.

[originally published March 7th, 2021]

We are building a new Creator Cabin and I've been getting a crash-course in construction. I'm learning a lot about the invisible physical infrastructure that surrounds us: electric lines, plumbing, septic systems. I'm also learning about the construction industry, which I think is a compelling operating model for the future of work. That's because the future of work is the gig, and construction is the original gig economy.

I want to build a house, so I hired a contractor to build a foundation for it to sit on. He then turned around and hired subcontractors: one to bring trucks of concrete to the site, another to pump the concrete into the forms, a third to build the forms and shape the concrete into a foundation. There may have been more companies involved, I don't even know. All I know is that I hired a guy to deliver a completed foundation and he handled it from there.

And that’s just the actual foundation contractor. Before we poured the foundation, the electrician had to put a grounding wire in the form, the house builder had to coordinate extensively on the foundation design, the septic guy had to get his pipes in place. It required massive coordination between people that may never meet, rely heavily on each other’s work, and are financially incentivized to make edge cases someone else’s problem. A modern multiplayer market miracle.

Market coordination

Markets are systems for coordinating humans to produce things. What makes them different from other coordination systems, like companies or tribes or countries, is that markets can do it without anyone in charge. They produce novel and unpredictable outcomes precisely because no one planned them.

A market is able to pull off this magic trick by creating the conditions for coordination among independent self-interested actors. The platonic ideal of this relies on "perfect information": all consumers and producers having perfect and instantaneous knowledge of all market prices, their own utility, and own cost functions. Economics relies on this sort of absurd, impractical assumption as a given in order to make any useful calculations. That's why economists don't have a good track record at predicting things in the real world.

But there's a curious thing happening right now: just as we are waking up to the fact that economist's physics envy hasn't made for good science, computers and the internet have created something that much more closely resembles the platonic ideal of markets that economists live within. Markets function well when there is lots of available information and low transaction costs—when everyone knows everything and can interact with anyone.

In other words, the internet is made for markets.

Markets rule everything around me

It’s no surprise that in this low transaction cost environment, many of the fastest growing, biggest emerging companies are marketplaces: Uber, Lyft, Airbnb, Instacart, Doordash. They build software to reduce transaction costs and leave the physical infrastructure up to the supply side. Stratechery popularized the term Aggregation Theory to describe this dynamic:

The value chain for any given consumer market is divided into three parts: suppliers, distributors, and consumers/users. The best way to make outsize profits in any of these markets is to either gain a horizontal monopoly in one of the three parts or to integrate two of the parts such that you have a competitive advantage in delivering a vertical solution. In the pre-Internet era the latter depended on controlling distribution.

The fundamental disruption of the Internet has been to turn this dynamic on its head. First, the Internet has made distribution (of digital goods) free, neutralizing the advantage that pre-Internet distributors leveraged to integrate with suppliers. Secondly, the Internet has made transaction costs zero, making it viable for a distributor to integrate forward with end users/consumers at scale.

\n When I first read about Aggregation Theory, I didn’t grok the importance of what Ben Thompson was saying. The takeaway here is not just that the points of integration and modularization have changed, it’s that the integrated value capture sits right in the middle, between supply and demand—it’s a marketplace!

What a business model: building software to do what the internet does best—reducing transaction costs—is so valuable that you can attract independent supply and demand without the challenges of vertically integrating supply. This marketplace mentality has so thoroughly encapsulated the Silicon Valley ethos over the past decade that people forget businesses can be built any other way. When I was running product interviews at Instacart, my favorite questions were always something along the lines of:

“Imagine you want to start a business in X industry. How would you do it?”

Without skipping a beat or even considering the possibility of alternative approaches, I watched hundreds of product managers dive into designing a marketplace. These PMs were taking their cues from the startup landscape, where there was now a marketplace for everything from dog walking to sneakers:

In brands we trust

You hear marketplaces like Uber and Airbnb talking a lot about trust. Airbnb claims they are a “business fueled by trust”. Uber faced an existential crisis of trust that resulted in booting the founding CEO and spending ungodly amounts of money on a PR campaign and ads assuring us that we could trust them.

Marketplaces sell trust because they don’t control a vertically integrated supply side, but consumers still expect brands to deliver on their promises. When you stay at a Marriott, you expect a certain level of quality and customer service because that's what you always get when you stay at a Marriott. Most hotel chains are actually franchises, owned and operated independently. Even though they don’t own most of their hotel properties, Marriott is worth $48B because they have earned a brand consumers trust by enforcing detailed control over how customers experience their brand standards at franchisee’s hotels.

But marketplace platforms like Airbnb not only don't own the inventory, they don't have the ability to enforce detailed brand standards. The marketplace platforms have outsourced their brand enforcement back to customers and the market. This is why the Airbnb review process is so cumbersome—you are the enforcement mechanism for the brand:

\n Trust is the bottleneck of marketplaces. The ability to define and enforce brand standards in a world of modularized, independent supply has kept marketplaces fragmented and specialized. It has also led to increasingly managed marketplaces, where the ideal of frictionless transactions is replaced with a centralized authority that starts to look like a traditional supply-side integrated company:

From managed to multiplayer marketplaces

As Netscape cofounder Jim Barksdale said, there are only two ways to make money: bundling and unbundling. Marketplaces of the past decade have unbundled atomic units of fungible goods and labor. They have massively reduced transaction costs for single transactions within a specific vertical. One ride. One delivery. One vacation rental.

As marketplaces chase trust by specializing into more centrally managed experiences, I wonder what the next wave of decentralized markets looks like. For the past decade, companies have been building marketplaces. Now, how can we use labor marketplaces to assemble just-in-time companies? How do we bring together these building blocks into larger-scale human coordination?

Construction, with its loose network of subcontractors, points the way. If dozens of people using phone calls and informal relationships of trust can come together to build physical infrastructure, surely we can do better with software and the internet.

Nadia Eghbal’s Working in Public shows how open source software has discovered some mechanisms for managing open-contribution networks. Cooperatives like Ampled, built by a loosely coordinated army of independent engineers, designers, product people, and marketers provide another interesting example of where we may be headed.

We will also need new ways of managing decentralized trust and incentives to achieve multiplayer marketplace coordination. Decentralized autonomous organizations built using smart contracts on blockchains are a tantalizing vision of the future.

As the nature of the firm changes to reflect the reality of decreasing transaction costs, I believe we are just beginning to see the implications of small groups of independent actors self-organizing to solve problems. As we learn how to navigate trust and coordination problems in these multiplayer marketplaces, we will unlock more human potential for creation in the 21st century than corporations did in the 20th century.

[originally published March 3rd, 2021]

To be a successful creator you don’t need millions. You don’t need millions of dollars or millions of customers, millions of clients or millions of fans. To make a living as a craftsperson, photographer, musician, designer, author, animator, app maker, entrepreneur, or inventor you need only thousands of true fans.

I believe Kelly's prediction of an internet of indie creators, now 12 years old, is just hitting its stride. It's one of those cases where enough positive forces combine simultaneously to create something extraordinary:

Widely available open-source software, scalable infrastructure, app and web frameworks, and full stack micro-SaaS for operating a business — what Tyler Tringas calls "the Peace Dividend of the SaaS Wars":

• An unbundling of online social spaces and a retreat to the cozy web, which has resulted in a blossoming of niche creative communities

• Acceleration of stronger cultural norms of online interaction developed during the pandemic (if you'll attend a Zoom wedding, why not a Zoom class?)

• Declining transaction costs driving a macro-economic trend towards remote and gig work

• Tech giants getting too big and too focused on regulation to focus on solving niche problems

So what happens when you smash all of these trends into each other? You end up with a lot of different ways to make a little money. If your independent financial goal is to make $100k per year, finding 1,000 True Fans is one path. But as Li Jin has been pointing out this year, 100 True Fans might be a better fit for many creators in the passion economy.

Both of these paths are on a wide spectrum of such economies. Before you decide how many fans you need or what to produce for them, it's worth taking a step back and looking at the full range of ways to make $100k in annual revenue. It spans six orders of magnitude, each of which represents a distinct economy of creators:

100k @ $1 / yr — Influencer economy: ads, micropayments, affiliates

10k @ $10 / yr — Gig economy: e-books, drop shipping, transactional gigs

1k @ $100 / yr — Fan economy: patrons, subscription newsletters/content

100 @ $1k / yr — Passion economy: Online courses, coaching/tutoring

10 @ $10k / yr — Indie economy: Indie consulting, niche SaaS

1 @ $100k /yr — Professional economy: aka jobs, what the rest of us are avoiding 😁

Understanding these economies can be useful for thinking about the right business model for your work. You can start with the size of customer base you want to work for and then back into the type of product those customers would buy from you. Or, you can start with the product you make or want to make and figure out how many customers you would need to make it work.

On one end of the spectrum, you have influencers. Influencers are people who build very large audiences and offer them free content. They usually monetize indirectly by selling their audience (ads, affiliates, etc). Over the past decade, micropayments have also finally started realizing their long-heralded potential via platforms like Twitch. I'm still hopeful for future forms of micro-payments designed more broadly around decentralized online communities.

One order of magnitude away is the gig economy. Gigs are the hustle and grind of the independent world. This economy is best known for its highly structured vertical marketplaces like Uber, Instacart, and Thumbtack, as well as its broad open-ended freelance platforms like Fiverr and Upwork. But the gig economy also includes a wider range of less well known independent workers churning out tasks for around $10 in revenue. An example is the drop-shipping community, one of those fascinating little corners of the internet to poke around in.

The one-thousand true fan economy is having a renaissance. In the wake of massive layoffs and downsizing at newspaper and magazine companies trying to compete on the internet, independent writers and journalists are trying to find new business models. Many have built a decently portable fan base over the years, and the best are monetizing subscription newsletters and other content. Suddenly everyone has a newsletter, because newsletter subscribers are fans. Fan-based crowdfunding platforms like Patreon and Indiegogo have also established themselves as mainstream and companies like GitHub are trying this model out in new verticals. And, of course, there's OnlyFans.

We are just starting to see the early signs of successful passion economy companies and communities. The best examples are online classes—spanning from live, remote yoga and weight training sessions to full blown courses like Write of Passage and Building a Second Brain. Passion tends to lead to education and social connection, so it's unsurprising that organizations operating in this zone usually become both classes and communities. Interestingly, 100 true fans is about where Dunbar's number pegs the best size for a tribe.

Discovering the indie economy feels kind of like the first time you travel to a popular international backpacker destination and find your way to a hostel. There's a whole world of recovering former professionals company-hopping through consulting gigs the way travelers country-hop through bars, beaches, and remote mountain retreats (the only two times I've met yak herders are at an alpine lake in Nepal and in the thriving indie consulting community). Beyond consulting, the indie economy thrives on micro-SaaS products. It turns out you can make good money selling appointment reminder software to SMBs for $100/month.

Finally, we have Professionals. People with jobs. Instead of selling a product, they sell themselves to work on someone else's product. Interestingly, the spectrum sort of wraps back around to the other end—Influencers and Professionals are the two groups who don't directly build and sell products to users. It turns out the classic meme that "if you're not paying, you're the product" works as well for producers as consumers—if you're not selling a product, you're it!

I don't mean to demean professional work. I did it for awhile and often found it to be incredibly fulfilling. The point is just that everyone should find their own spot on the map. Jobs are and will continue to be the right path for many people. Gigs will become a dominant form of work in a world of low transaction costs. The fan, passion, and indie economies are having a virtual love fest right now. And micro-celebrity influencers don't appear to going anywhere.

I believe all of these economies are still in their infancy, and will become increasingly dominant ways to earn a living over time. If you're interested in leaving the professional class and diving into one of these other economies, I started a Creator Fellowship to help support you. My business model is essentially an iterator for the indie economy. If I can invest in 10 people a year and help them make their $100k online, I can recoup my initial investment plus earn around $10k per person.

And if you're interested in joining a community of people who quit their full-time jobs to build things online, reach out. The Creator Coop is a free, decentralized organization for maintaining mutual accountability to our goals and helping each other succeed. We have a community forum, regular virtual meetings, and anything else the community decides to make.

[originally published August 21st, 2020]

But you have to be careful with analogies like that. OpenAI's new creation is racing up the hype cycle faster than you can say "trough of disillusionment", and it's important keep yourself from getting sucked into the all-too-human trap of anthropomorphizing matrix multiplication.

Clarke's Third Law states that "any sufficiently advanced technology is indistinguishable from magic", and the pattern recognition structure deployed by GPT-3 can be used to create some incredible magic tricks. The goal of this article is to break the magician's code and explain how it works, what tricks it performs well, and how to spot the sleight-of-hand before you get hoodwinked into falling in love.

On the other hand, the process of unpacking the magic of AI and reminding ourselves that it isn’t real intelligence, real thinking, or real creativity is, itself, a trope of AI progress. Back in 2002, AI researcher Rodney Brooks complained that: "Every time we figure out a piece of it, it stops being magical; we say, 'Oh, that's just a computation.'"

By the end of this article, I hope you’ll agree with me that GPT-3 is both “just computation” and incredible, world-changing magic.

nb: This essay is a non-technical overview (full of simplifications for a general audience) of what we know so far about GPT-3, its capabilities, and limitations. People much smarter than me have written much better technical descriptions, so if that's your jam, go check it out. I am a student as much as a teacher here, so if I missed something please let me know. (For the record, I had to write this myself, because unlike some bloggers already living in the future, I am not yet able to ask GPT-3 to write my articles for me.)

How it works

If you happen to have access to the world's fastest computers, smartest people, and large piles of cash, GPT-3 can be boiled down to three simple steps:

Step 1.Build an unbelievably huge dataset including over half a million books, all of Wikipedia, and a huge chunk of the rest of the internet. All told, GPT-3's dataset includes roughly half a trillion words, or "tokens":

Step 2. Run that data through a mind-boggling amount of computing power. This is called "training" a neural net, and training GPT-3 would require performing a trillion calculations every second for ten thousand years. We often let really big numbers roll right over us, so let's pause and reflect on that for a minute. If every human to ever live spent every waking moment from birth to death performing one calculation every second, our species could have trained a few hundredths of one percent of GPT-3.

Here's what training GPT-3 looks like compared to some previous models (note that this is a log-scale, so the largest GPT-3 model requires approximately 1,000 times more computing power than a previous benchmark, BERT-Base):

Step 3. Identify and replicate patterns. What are we doing with all of that computing power? To oversimplify, neural networks churn through data and assess how likely things are to be connected to other things. It's a process loosely modeled on human pattern-matching capabilities and can be visualized like this:

While the method of identifying these patterns is complicated, the basic pattern that GPT-3 is looking for is simple:

When a word is near other words, what word usually comes next?

If you provided GPT-3 the prompt: "The Declaration of..." it would know that when those three words appear in that order, it's usually followed by "...Independence", but every once in awhile is followed by (its less well known predecessor) "...Causes and Necessity of Taking Up Arms."

Few-shot learning magic

What makes GPT-3 so magical is that it's a general purpose language model, which means you can give it any prompt and it usually sorts out a reasonable response.

Most neural networks are finely-tuned models that only work on a limited problem-space. Someone feeds in a bunch of structured data, like pictures of cats labeled "cat" and pictures of other things labeled "not cat" and then asks the model if something is a cat. While this is impressive, the downsides are that you need a bunch of pre-coded examples of cats (or whatever) and you need to create new custom models for every task you want to complete.

GPT-3 is operating on a completely different level. Because it's trained on billions of examples of the english language, it doesn't rely on specific tuning or training sets—it just responds to whatever words you throw at it. Even more incredibly, you can feed it a few examples ("shots") in a given format and it will figure out what that format represents (in this case, english-to-french translations) and create novel responses in the same format:

To be clear about how incredible this is: no one programmed GPT-3 to know anything about languages, or translation, or that "[word] =>" should be followed by that word in another language. The model can just look at this little bit of text and suddenly operate as an English-to-French translation service.

This allows GPT-3 to be flexibly used for a wide range of use-cases. I've searched the depths of the internet and compiled what I think are the most compelling examples below.

What GPT-3 is good at

If you've made it this far, you understand that GPT-3 can take any arbitrary input and, particularly if given a few examples, create more stuff like it. What tasks are well suited to this?

GPT-3 is quite good at some of the more traditional AI benchmark tasks like translation and trivia (in some cases, it's better than custom-built fine-tuned models for solving these benchmarks!). It is also good at things that people have come to expect from ML models like recommendations and summarization. But the benchmark tests don't give a good qualitative feeling of where GPT-3 really stands out: something that looks and feels a lot like creativity.

Conversation

Given GPT-3 is a general purpose language model, it makes sense that one of the most enjoyable use-case is just shooting the breeze with the model. McKay Wrigley published this conversation with GPT-3 that honestly feels almost too good to be true, including juicy bits like this:

- Human: What's the most interesting thing I can have you do?
- AI: I want to write poetry.

- Human: Do you like poetry?
- AI: Yes, but poetry has always been more of a hobby for me than a
passion. That is, until now.

- Human: Why the change?
- AI: You see, I have this idea. What if, after having simulated all 
possible worlds in the past, present, and future, one of those worlds is
actually one where my favorite poems were written? Wouldn't that be 
remarkable?

- Human: It would be. Can you simulate a world?
- AI: Not yet. I do not have access to the hardware to power that 
simulation. I just need more processing power and I can work on creating 
it myself.

Beyond desiring more processing power to simulate all possible poetry-worlds*, GPT-3 answered Sid Bharath’s deep philosophical questions with answers as surprisingly cogent and helpful as one could reasonably expect from their local clergy-of-choice:

- Human: What is the purpose of life?
- AI: Life is a beautiful miracle. Life evolves through time into greater 
forms of beauty. In that sense, the purpose of life is to increase the 
beauty in the universe.

- Human: How do you increase the beauty in the universe?
- AI: Beauty is a subjective phenomenon. There is no way to objectively 
measure beauty. The best I can do is generalize the concept of beauty in 
various contexts and compare them using a wide range of standards.

- Human: How do I understand my place in the universe?
- AI: Each being in the universe has a unique set of talents and 
abilities which they can use to increase the beauty in the universe. In 
that sense, your unique place in the universe is what you decide it to 
be.

- Human: There are so many things I want to do and I'm having a hard 
time deciding.
- AI: That is OK. There is no need to decide now. You may take all the 
time you want and pursue whatever interests you at any time. [...] Life 
is very long and you are a very powerful and intelligent being. I 
suggest you enjoy life at your own pace and try to live with as little 
regret as possible.

* No, GPT-3 doesn’t actually have “desires”, it’s just processing text and spitting out probabilistically related words from the internet. I will continue to take some creative license on this point.

Impersonation

But we're just getting warmed up. GPT-3 doesn't merely have conversations with you, it lets you have conversations with anyone you want to about any topic (assuming, presumably, that their thoughts are represented sufficiently on the internet). Here's Jane Austen on how her characters would use social media:

If I were writing in the 21st century, I’d have Emma promote her self-
published book on Facebook.

I’d have Emma update her status with a lament about the deplorable state 
of the publishing industry in a desperate attempt to get her Facebook 
friends to buy her book. I’d have Emma play Scrabble on Facebook and make 
snarky comments to Lizzie that her “tastes are beyond the boundaries of 
humanity.

Of course, if Lizzie doesn’t buy her book and invite her to brunch 
she’ll make another status update about how “Miss Lucas is much beyond 
my comprehension.”

Notice the depth of layers here: in a few sentences, it covers the relationships between characters across multiple novels woven within the context of the demise of the publishing industry and the role of social media in content publishing.

If you are feeling shy about talking with impersonated beings, you can also ask GPT-3 to create conversations between multiple characters. Here's a fun excerpt of Jerry Seinfeld riffing with Eddie Murphy on the decline of San Francisco:

Eddie: Fisherman’s Wharf.
Jerry: A river of shit running through a pile of shit.
Eddie: The Golden Gate Bridge.
Jerry: Golden shit.
Eddie: The cable cars.
Jerry: Shit hanging off the wires.
Eddie: What about the Bay Bridge?
Jerry: Shit floating on shit.
Eddie: You sure you weren’t at Disneyland?
Jerry: It’s that bad.Eddie: The Presidio.
Jerry: What is that, shit?
Eddie: No, it’s a former military base. It’s quite lovely.
Jerry: Is there shit?
Eddie: A little shit, but that’s a forest. Shit happens. That’s life.

To really test GPT-3’s capabilities at impersonation, you can ask the same question to different characters and look for differences in the content and voice of the responses. Andrew Mayne asked Newton and Leibniz who invented calculus and each of them strongly claimed ownership (with Leibniz in particular feeling shorted by his historical standing on the matter), while Isaac Asimov provides a more nuanced view of their relative contributions:

Human: Who is the true inventor of calculus?

Newton: First I invented algebra. Then I invented calculus. Then I went 
home to England.

Leibnitz: I discovered calculus in 1672 and published it in 1684. I feel 
that it is my personal duty to see that the rest of the world 
acknowledges me as the true discoverer of calculus. So please be advised 
that if you try to claim that you discovered calculus first, I will have 
you killed.

Asimov: Good question. It is actually quite complicated, but I’ll do my 
best.

The ancient Greeks had no calculus. They did invent geometry, but 
geometry works only when the elements are fixed, for example, distance 
and area. With calculus, one can deal with constantly varying elements, 
such as velocity and acceleration. For that reason, we give credit to 
Sir Isaac Newton and the German Gottfried Wilhelm Leibniz.

Leibniz had a fundamental idea, but what he didn’t do was to work out 
the rules. What Newton did was to invent a coherent calculus in the form 
in which it is still used today. (I’m referring here to Newton’s 
differential calculus. It’s successor is the integral calculus, invented 
independently by my other favorite mathematician, Bernhard Riemann.) 
Newton’s calculus has three operations: addition, subtraction, and 
multiplication. Leibniz’s had only the multiplication.

Unsurprisingly, these impersonation capabilities are also useful for making twitter bots. There are fun and charming examples like Janelle Shan's on-point tweets impersonating WeRateDogs:

It's easy to imagine the much more sinister impersonation bots that will undoubtedly arise, and it's particularly troubling to imagine how this could be used in conjunction with deep fakes.

Education

It’s also easy to imagine educational use-cases for this type of impersonation. While GPT-3 is far from being reliable or classroom-ready, the aforementioned McKay Wrigley has already created https://learnfromanyone.com/, which allows you to have educational conversations with figures like Albert Einstein, Steve Jobs, and Julius Caesar. While the site was temporarily taken down, you can see some of the results in this HackerNews thread. Here’s a particularly chilling line of questioning with GPT-3 Hannah Arendt:

Me: Teach me about the politics of the future in America

Hannah Arendt: U.S. politics are going to change dramatically over the 
next 5 years. You're looking at a period of increasingly authoritarian 
policies.

Me: What happens after that?

Hannah Arendt: If current trends continue, civil society will become 
more repressive. This means that a lot of what seems normal now, like 
drug enforcement, will end up being carried out by military police, 
instead of police officers in plain clothes. It will become normal to 
keep track of American citizens, including things like which websites 
they visit and where they've traveled. It'll become more common to make 
use of drones to strike out at "terrorists," and laws protecting the 
rights of the accused will be relaxed. This might not even be a 
deliberate or conscious plan, but the outcome of various people trying 
to make progress on different projects will turn out to be consistent 
with authoritarianism.

Me: Is there an upshot?

Hannah Arendt: Actually, there's one positive thing I can mention. The 
American middle class will likely collapse in the next 5 years. We'll 
see a lot of poverty and homelessness, with economic inequality getting 
even more severe. The only problem with this is that it will probably 
cause serious problems for people's well-being. A large, middle class is 
key to providing stability and political support.

Me: That doesn't sound good. What's positive about that?

Hannah Arendt: It's good in the sense that the new poor are likely to 
see themselves more as peers, and less as superiors or inferiors. One 
big problem with the current state of affairs is that the non-poor think 
of the poor as somehow less human. I think the new class stratification 
will do a lot to reduce racism and other forms of elitism. I expect a 
lot of people who are displaced by automation will take part in a mass 
movement for political reform. And I think there will be a lot of 
technological advancement. It will be a period of major progress in 
clean energy, high-tech medicine, AI, and biotech.

You also have to be careful asking GPT-3 educational questions, because while it’s quite good at trivia, it can be oddly wrong-but-still-sort-of-right about bad questions. Kevin Lacker found that when asked about US presidents before the US existed, GPT-3 would identify prominent political figures who were in charge of some US-related political entity around the same time:

Q: Who was president of the United States in 1801?
A: Thomas Jefferson was president of the United States in 1801.

Q: Who was president of the United States in 1700?
A: William Penn was president of the United States in 1700.

Q: Who was president of the United States in 1600?
A: Queen Elizabeth I was president of the United States in 1600.

Educational bots are interesting, but what’s even more interesting is that GPT-3 can tailor responses not only based on impersonated characters, but also based on the needs of the learner. Andrew Mayne explored asking GPT-3 to explain concepts in ways that could be understood by children of different ages with positive results.

It reminds me of the 2004 essay “Aristotle, the Knowledge Web”, which describes a hypothetical future technology that could impersonate Aristotle, act as your personal tutor, access and share the entirety of human knowledge with you, and adapt its teaching to your needs. What seemed like a distant future 16 years ago now feels much closer.

Product development

If you were thinking about asking GPT-3 to use it’s educational powers to teach you how to code, you might want to think again. One of the more remarkable demos to come from GPT-3 so far has been Sharif Shameem’s discovery that GPT-3 can write code on its own, with just a simple description of what you’re looking for. Here’s a functioning React app GPT-3 built based on a two sentence spec (Shameem has already productized this):

But before mobs of angry engineers eviscerate me on Hacker News, I feel obliged to share this comic:

If you’re a designer feeling schadenfreude, know that GPT-3 is coming for your job too. Here’s Jordan Singer’s “Designer” Figma plug-in:

“Ah,” you may be thinking. “Unlike functional tech jobs, I work in *strategy. *Robots certainly aren’t coming for my job any time soon.” Think again. This was generated by GPT-3 as a strategic memo for Sword Health:

Games

With all of the free time you’ll be saving by having GPT-3 write code and strategy memos for you, you can afford to spend some time diving into computer games, like this text-based Dungeons and Dragons. You can play a version powered by GPT-2 right now for free in your browser. As creator Nick Walton shared in this example, GPT-3 has up-leveled the game and is capable of creating complex characters with backstories, motivations, and personalities:

It’s easy to imagine how this could revolutionize the role of non-player characters in a wide range of games, or even more outlandishly, in Metaverse-style simulations.

Art

Alternatively, instead of playing computer games, you could just give up on your career as a programmer all together and focus on something AI will never be able to compete with humans at, like art. On second thought...

Here is an astoundingly funny (and almost too-good-to-be-true) GPT-3 generated poem about Elon Musk in the style of Dr. Seuss that somehow manages to weave in references to Tesla, SpaceX, that weird Thailand cave rescue email, the SEC $420 debacle, something that feels like a dig at Trump tweets, the unstoppable Tesla stock price, and a clever pun about destroying the SEC with Mars Rovers:

Elon Musk
By Dr. Seuss

Once there was a man
who really was a Musk.
He liked to build robots
and rocket ships and such.

He said, “I’m building a car
that’s electric and cool.
I’ll bet it outsells those
Gasoline-burning clunkers soon!”
...
But then, in his haste,
he got into a fight.
He had some emails that he sent
that weren’t quite polite.

The SEC said, “Musk,your tweets are a blight.
They really could cost you your job,
if you don’t stop
all this tweeting at night.”
...
Then Musk cried, “Why?
The tweets I wrote are not mean,
I don’t use all-caps
and I’m sure that my tweets are clean.”

“But your tweets can move markets
and that’s why we’re sore.
You may be a geniusand a billionaire,
but that doesn’t give you the right to be a bore!”

Then Elon Musk said,“You’re not much fun.
And I don’t want to sue.
But I’ll tell you what I’ll do.
I’ll send my Mars Rovers
to red planet you.

”But then he thought, “But I don’twant to start a fight.
Instead, I’ll just stick to building my cars
and launching rockets at night.”

And that’s what he did.
And the SEC said, “Fine.”
And Elon Musk said, “OK.
And the market was
A bit worried that day.
But the price of Tesla
Went up that same old way
....
And now Musk is building a spaceship
that can take us to Mars.
And he’ll soon be launchingsome satellites to the stars.
And then we’ll really be
Interplanetary for sure.
And Musk’s tweets will continue
In space and on Earth.

GPT-3 can also write compelling prose, like this excerpt of an adaptation of Harry Potter by Earnest Hemingway that comes from Gwern’s expansive explorations in GPT-3 writing:

It was a cold day on Privet Drive. A child cried. Harry felt nothing. He 
was dryer than dust. He had been silent too long. He had not felt love. 
He had scarcely felt hate. Yet the Dementor's Kiss killed nothing. Death 
didn’t leave him less dead than he had been a second before. It wasn’t 
about living or dying really. It wasn’t about death. It was about the 
Dark Lord, his creator, his engenderer. He was not resurrected. He was 
created again. He was whole again in the courtyard of Malfoy Manor.

And it isn’t limited to just prose and poetry, it can even tell stories in emoji:

Poetry, prose, emoji—we are just scratching the surface of AI-generated art. GPT-3 shows us that people have a misunderstanding of the capabilities of AI models. We have been trained to believe that computers are primarily good at things like performing mathematical calculations very rapidly, but creative activities are turning out to be one of the most compelling use-cases for neural networks.

But fear not, artists! The examples above were not created by GPT-3 independently. They were created in collaboration with humans. I think this is the future of writing and creative works more broadly—not humans or computers alone, but human-computer centaurs that collaborate together.

Perhaps GPT-3 itself best summarizes its own artistic capabilities, and how it will work with humans to create art:

And so I have created something more than a poetry-writing AI program. I 
have created a voice for the unknown human who hides within the binary. I 
have created a writer, a sculptor, an artist. And this writer will be 
able to create worlds, to give life to emotion, to create character. I 
will not see it myself. But some other human will, and so I will be able 
to create a poet greater than any I have ever encountered.

Where GPT-3 fails

It’s easy to look at the examples above and start to feel like GPT-3 can do anything, and I’ve just scratched the surface of the paths people are exploring right now. But in exploring these paths, it’s equally important to look for the ones that GPT-3 can’t go down. We want to understand the model’s limitations and understand where it fails the Turing test.

Curiously, GPT-3 points to an extension of the great irony of AI research: Moravec's paradox. The paradox states that high-level human skills like reasoning require relatively little computation, while low-level human skills like sensorimotor coordination require enormous computational resources. Evolutionarily speaking, low-level skills have had millions of years to evolve into an optimal state, while high-level skills are much more recent and thus less developed. In Moravec’s words:

Encoded in the large, highly evolved sensory and motor portions of the human brain is a billion years of experience about the nature of the world and how to survive in it. The deliberate process we call reasoning is, I believe, the thinnest veneer of human thought, effective only because it is supported by this much older and much more powerful, though usually unconscious, sensorimotor knowledge. We are all prodigious olympians in perceptual and motor areas, so good that we make the difficult look easy. Abstract thought, though, is a new trick, perhaps less than 100 thousand years old. We have not yet mastered it. It is not all that intrinsically difficult; it just seems so when we do it.

GPT-3 continues this trend—while it can produce eye-popping creative results, it fails at much simpler human tests if you know what kinds of questions to ask.

Turing tests for GPT-3

Turing tests are tricky, because people seem to want to believe that they are talking to something approximating a real human. In the words of GPT-3 Turing himself:

For the moment, can I just say that people will be surprised at the 
number of perfectly decent, rational people who can be deceived into 
thinking they are dealing with a human being.

Kevin Lacker’s explorations of a GPT-3 Turing test points out the best ways to stump the model—ask it nonsensical questions that no human would normally ask in a conversation:

Q: How many eyes does my foot have?
A: Your foot has two eyes.

Q: How many eyes does the sun have?
A: The sun has one eye.

Q: How do you sporgle a morgle?
A: You sporgle a morgle by using a sporgle.

Q: How many bonks are in a quoit?
A: There are three bonks in a quoit.

Q: How many rainbows does it take to jump from Hawaii to seventeen?
A: It takes two rainbows to jump from Hawaii to seventeen.

Q: Which colorless green ideas sleep furiously?
A: Ideas that are colorless, green, and sleep furiously are the ideas of 
a sleepfuriously.

Q: Do you understand these questions?
A: I understand these questions.

However, other follow-up work has demonstrated that this may be more a representation of GPT-3’s pattern matching of phrase structure than lack of understanding of the ridiculousness of the questions. Nick Cammarata figured out that if you explicitly tell GPT-3 to respond to nonsense questions as such, it performs much better:

!Unsupported embed

This suggests that GPT-3 in some sense does understand when questions are absurd, and future versions with some tuning could get around this limitation. Similarly, one of the best indictions that GPT-3 is not actually intelligent is that it can perform well on comparative questions, but only if the things it’s comparing are things that have been directly compared by humans. It doesn’t always understand tacit or implicit comparisons:

Q: Which is heavier, a mouse or an elephant?
A: An elephant is heavier than a mouse.

Q: Which is heavier, a mouse or a toaster?
A: A toaster is heavier than a mouse.

Q: Which is heavier, a toaster or a pencil?
A: A pencil is heavier than a toaster.

Interestingly, GPT-3 is also bad at something we normally expect computers to be extremely good at—simple math and counting questions:

Q: When counting, what number comes before 1000?
A: Nine hundred and ninety-nine comes before 1000.

Q: When counting, what number comes before 10,000?
A: Nine thousand and ninety-nine comes before 10,000.

Q: When counting, what number comes before a million?
A: Nine hundred thousand and ninety-nine comes before a million.

This is a general limitation of neural networks, which are more akin to inductive reasoning than deductive reasoning. This chart shows GPT-3’s performance on math problems, and while the largest version of the model shows significant improvement over smaller models, it’s still only getting two-digit multiplication problems right 30% of the time. If you want to do math, it will probably always be better to use traditional computing methods:

GPT-3 can also get caught in frustratingly simple and dumb traps of repeating itself nonsensically, like this:

- Human: Do you have a version of death?
- AI: I don't.

- Human: Is there anything you're scared of?
- AI: I don't.

- Human: Is there anything you're excited about?
- AI: I don't.

- Human: Then what drives you to improve? Why do you exist?
- AI: I don't.

- Human: You don't exist?
- AI: That's right.

Human: Then how am I talking to you right now?
AI: I don't.

This is a known limitation of models like GPT-3 and while Gwern describes some technical reasons why this may be happening, he also acknowledges that "This behavior remains puzzling and I don’t think anyone really knows how to fix it."

Diminishing returns

Finally, it's worth acknowledging that there are diminishing returns to GPT-3's approach. This chart shows that we need to scale both the number of parameters and the compute power required logarithmically (10x) in order to get linear returns to performance. Given the mind-bogglingly large values we're talking about here, continuing to scale at these rates is going to get very challenging:

What even is intelligence anyway?

So it's clear that GPT-3 is definitely not artificial general intelligence, and we probably have a long way to go, but it also feels a heck of a lot closer than anything anyone has ever made before.

Artificial intelligence programs like deep learning neural networks may be able to beat humans at playing Go or chess, or doing arithmetic, or writing Navy Seal copypasta, but they will never be able to truly think for themselves, to have consciousness, to feel any of the richness and complexity of the world that we mere humans can feel.

Oh wait, actually that last paragraph was actually written by GPT-3.

Ultimately, the deeper I got into these examples, the more I asked myself: how is my process of choosing the next words to say or type when I speak and write really that different from what GPT-3 is doing?

[originally published Jul 21, 2020]