This essay intends to answer three key questions about blockchains:
Where will productive assets come to back blockchain finance?
Are blockchains good for anything meaningful, or are they just fintech with extra steps?
Why haven’t blockchains produced significant productive assets up to now?
The dollar system ("TradDollar") is both over and under estimated. To some, it represents an irresistible force, a requirement for doing any business at all. To crypto purists, it is often seen as a fragile relic, one that will be swept away by "trustless" cryptocurrency systems.
In reality, the TradDollar is simply a tool. It enables efficient trade in high marginal cost ("heavy") goods by leveraging the strength of institutions (American ones in particular) in lending, contracts, and capital formation. And, up to now, it has also been used for commerce in low marginal-cost "light" goods and services, including investment in software development.
However, the improvement of blockchain scaleability primitives, and the rise of operating systems that make blockchains more programmable, will open the door for a competing set of rails: the BlockDollar. This infrastructure will outcompete in “light” businesses that are more naturally funded via equity than debt and require frictionless iteration & feedback loops to reach maximum potential.
This concept needs very little explaining: some goods or services cost something to deliver, and some cost almost nothing. A DVD delivered to your house costs something in physical transport; a streamed video costs only electricity. This is often described as "atoms vs. bits", or "physical vs. virtual", but even purely virtual things can have high marginal costs (a given financial arbitrage requires increasing capital to increase profit; online performance advertising is similar).
From here on, I will call things "heavy" (high marginal cost) or "light" (low marginal cost). Importantly, some things that are light in theory can become heavy in practice: Twitter sells advertising that costs it ~nothing to deliver, but has to use a sales-intensive process to sell it, making the true advertising+sales product heavy.
Not everything virtual/information-based is light, but to be light it is almost always necessary to be virtual (there are probably exceptions; I can't think of them right now).
Heavy things have a predictable relationship to money: they often want to borrow money to deliver their heavy payload, but have a high probability that that money can be paid back in a clear timeframe. This makes them natural candidates for debt:
My company is worth $10M
It delivers heavy things that cost $1M to make, and earn $2M upon sale
I borrow $1M from you to deliver my heavy things
I make $2M, and pay you back $1.1M I make $900K here upon completion instead of $1M, but I also was able to keep 100% of my company's equity, as opposed to giving 10% of it to investors. You were extremely happy to make 10% on a fairly quick turnaround with a high chance of payback.
The above is a simplification, but it's easy to see why debt is appealing when the payback is very likely and transactions are repeated over and over.
However, if the payback probability is unknown, lending that money to me is a terrible deal for you. You make $0.1M profit when paid back, but lose $1M whenever I fail to pay. You need the "true" probability of my delivering to be over 90%; if it turns out to be 50%, you will get rekt.
In that example, you got rekt because the downside was up to $1M, but the upside was capped at $0.1M. If the upside were uncapped, you could make completely different calculations. Enter equity.
Equity lets me provide uncapped upside. I can sell a part of my company to you, and the bigger the business grows, the more that's worth. You can then calculate the risk-reward: how big I need to get for you to make your money back, to make a 10x profit, etc.
Just as heavy things naturally "want" debt, light things are perfect for equity. Light things are information, and so equity financing is usually the "0 to 1" process of finding a new information configuration. Examples include:
inventing a physical machine
developing a new piece of software
marketing to a set of customers to bring them permanently into a network
finding a business process that people are happy to pay for
Often, a business uses equity to discover something light, and then moves to debt to fund the heavy things that light information enables. The Dutch company ASML makes the incredibly complex lithography machines used to create high-end chips. Those are very heavy (and physical!) things. However, they all come from a light process:
In 2012 Intel, TSMC, and Samsung all invested in ASML to help the company finish the EUV project that had started 11 years earlier: there were very real questions about whether or not ASML would ever ship, or die trying, while it was clear that immersion lithography was reaching the limits of what was possible.
Equity funding created the light information used to make further heavy things. Now that most of what ASML does is heavy, it carries significant debt to enable that ongoing production.
Things can also move in reverse: light, virtual things can become heavy over time. This article argues that SaaS, as it becomes more predictable and less novel, will move to debt financing as opposed to equity. This makes perfect sense: debt works well when the risk profile is known, and the additional spend is for heavy things like sales and marketing--not "0 to 1" things like creating new business processes and software.
With the heavy/light and debt/equity concepts in place, we can now move to the current dollar system (the "TradDollar"), and better understand the unique value it provides.
Debt and equity are abstract concepts. In order for you to actually be able to lend me money, or invest in my company, we need to come to agreement on a number of things:
what exactly are you buying when you invest in my company's equity?
what currency am I paying you back in, if I'm borrowing?
who judges whether I've paid you back when I borrow?
what happens if you want new investors?
how do I know I'll get rewards in proportion to my equity investment?
how do you get the information to know whether to lend to me in the first place?
when I make money in this system, in what do I store it?
In order to transact heavy things at scale, you must have answers to these questions. By providing strong answers to them, the TradDollar has built significant network effects: it is more appealing to use because more people use it. But what is "it" here?
The TradDollar provides:
the unit of account for lending—the vast majority of loans globally are denominated in dollars.
settlement infrastructure for those dollars, primarily via US banks
evaluation infrastructure for creditworthiness
predictable legal infrastructure, especially in the US and US-aligned countries
assets: US treasuries used to play the largest role; now, US-based assets (stocks, real estate) are used to store large amounts of values
There’s a popular story that the evil US government forces everyone to use the dollar at gunpoint, and while that story has aspects of truth, it misses the substantial benefits that financial network effects provide for a lending-based system. And because any system that has to move heavy things will be lending-based, TradDollar rails provide real value to users.
Are TradDollar rails ideal for all use cases? Do they have complete lock-in? In order to explore that question, we need a new concept: Hardness.
In this highly recommended essay, Josh Stark defines "hardness" as the ability to "make the future certain in specific ways." He identifies three base forms in which those commitments can be fixed: atoms, institutions, and blockchains.
The best known example of atom hardness is the use of gold as money: there is a limited amount of gold that has been dug up, and it's hard to rapidly increase that. The atom properties of gold let it secure a commitment to not debase a money supply too fast.
Institutional hardness is found in the current legal, financial, and corporate systems. Groups of humans are arranged in ways that allow commitments to carry forward into the future with some degree of reliability. This type of hardness is great for transacting with heavy things. It allows commitments to be made, while also allowing human judgement to make evaluations and intervene if the system is running poorly.
TradDollar rails are the instantiation of institutional hardness at global scale, with institutions that are reliable enough to let heavy things move around in great amounts.
Finally, we have blockchain hardness. This hardness uses blockchains to store and enforce commitments of asset ownership or code execution, taking out (most of) the human judgement present in institutional hardness. Human social consensus is the fallback layer for exceptional events like attacks on the chain or infrequent upgrades, but everything else runs programmatically.
Because it eliminates the friction present in human institutions, blockchain hardness can be extremely efficient and programmable. While that efficiency has obvious advantages, it generally needs to be self-contained: blockchain hardness can’t evaluate conditions in the physical world, and so any references to the physical world must filter through human institutions. This is poorly suited for heavy things and lending, which require frequent evaluation of states outside a self-contained virtual world.
So institutional hardness is necessary for dealing with heavy things, and the TradDollar is a powerful implementation of institutional hardness with unassailable (or at least, hard to beat) network effects.
This raises the question: what is blockchain hardness better at than institutional, if anything? If the answer is "nothing", crypto is a waste of time.
Fortunately (for my bags), blockchain hardness excels in one huge, critical area: creating and improving light things.
The TradDollar has to make a lot of tradeoffs in order to be good lending/settlement rails. Two of the biggest are interventionability and regulatory burden.
Interventionability (a very real word) is the ability to pause the system at any point, to make it less programmable & automatic, i.e. higher friction.
The most obvious example of interventionability is funds settlement: at any point in the process of sending money between banks, any bank in the chain can halt the process, and request additional documentation. Similarly, credit card processing has many chokepoints where a halt can happen.
For a system that needs to extend credit to heavy things, this makes sense. There are a lot of judgement calls as to whether what a company is doing is legitimate or fraudulent, and letting humans into the decision loop both handles edge cases and lowers the incentive to commit fraud in the first place.
However, there is a real cost to interventionability: efficiency is capped at the maximum throughput of the human institutions involved. Even light processes that don't need intervention have to limit themselves to that efficiency level. When everything has to run on the same rails, everything has to be limited to those rails’ maximum throughput.
Regulations let business processes and facts be exposed as an API to institutional hardness. There are many ways of organizing human groups to pursue shared goals (friend groups, mafia, cults, DAOs, internet forum memberships), but incorporating your group in Delaware or BVI allows it to interact with financial and legal systems in ways those systems are set up to handle.
Beyond defining corporate entities, regulatory institutions also govern how a business can operate, and what it is allowed to do in the course of operations. For example, in the US, businesses must keep detailed accounting records ("how"), and cannot sell equity to the public without an onerous registration process (“what”). Regulations impose substantial cost on businesses, but they also make its activity legible to the institutions that businesses depend on.
Any activity running on TradDollar rails must bear its regulation costs (and receive the benefits). If too many exceptions were allowed, firms of all types would take advantage of those exceptions ("regulatory arbitrage”), rendering the activity illegible and incompatible to the rails.
The TradDollar's primary task is to link a virtual world of finance and business to a world of heavy, often physical things as they move around. It uses human institutions to make this link, and most of its costs come from the challenges of reliably maintaining it.
But what about economic activity that doesn’t need to maintain information about external facts? Activity that simply exists or doesn't?
Two types of light things exist in this way: code and content. They can be delivered for almost no marginal cost and once people have them, the transaction is completed.
"Code & content are light" is hardly a new insight, so why take so much time building up to it? I did so because their lightness has two very important properties:
better suited to equity financing
self-justifying, no “real world link" needed: once the customer has the code/content, it can just be used
In turn, this means that many forms of code and content don't need the benefits of TradDollar rails, but must pay all their costs. They don't need to make their activity legible, because they're not borrowing against it, and they don't need institutions to verify that the product was delivered, because that can be regulated programmatically.
If TradDollar rails add high friction to the creation and delivery of code/content, it's likely some forms of economic activity are being stifled and not happening. What might those be? Unfortunately, any answer here will be mostly speculative, because we're talking about a counterfactual: what are things that could be happening, but aren't?
However, we can at least look at existing code/content economic patterns, and get some good ideas of what those missing things might be.
Looking at the internet today through the lens of light things, a few points jump out immediately:
People really like code, and people really like content.
They really want to fund new code and content.
The whole experience is silo'd.
People use Substack, Twitter, Instagram and many more to generate a vast stream of content to loyal audiences. A large portion of the productive population writes software (or manages its writing) for fun and money, and organizes companies to do the same. Crowdfunding and patronizing content creators is a thing, as is investing in companies for profit, fun, and status.
But the whole feels less than the sum of its parts. Successful content creators cannot crowdsource improvements to the tools they create on, and raising money happens in very limited and proscribed ways. Profits from any of these endeavors don't have ways to flow back into the ecosystems that created them.
Another way of saying this: people love light things, produce tons of them, and want to pay for more. However, because light things aren't connected to blockchain hardness, and instead run on trad rails, a lot of potential activity is lost.
While we can’t know exactly what we're missing out on due to light things not being connected to blockchain hardness, we can analyze the costs of this situation:
interest in and use of a platform do not lead to people building improvements and integrations for the platform, because that's not technically possible.
there's an artificial separation between developers (employed by the platform's company, and working on its backend) and its content creators/users, who cannot make improvements to the platform itself.
it's harder to attract high performers to these platforms, because their paths to monetizing high performance are limited. When monetization is possible (as in Youtube), that money does attract high numbers of high performers, although they are limited to pure content creation and can't improve the software itself.
no possibility for capital formation in communities. Even if improving/composing the underlying software were doable, it's hard to organize it and give upside, which in turn feeds back to the problem of attracting high performers.
high performing people and groups can't compound on their success easily, because of the lack of capital formation and lack of upside.
If this friction were removed, we'd likely see what I call Engines: communities that can invest in themselves, produce real, light goods, accumulate resources if successful, and then re-start the process, further growing the community and drawing resources to it.
Engines are a compelling pattern that arises naturally whenever light things are connected to money rails. Even in trad land, where there's more friction, powerful Engines arose in the Bay Area. Startup participants would raise (equity) funding, generate software, exit and be wealthy, and then fund the next generations of startups. Critically, it wasn't just about money: the people who had created startups had better expertise in both choosing and advising future ones.
In crypto, Ethereum arose as a community-funded Engine, and then immediately started self-funding the projects that would build on top of it, creating the ICO boom. This trend petered out due to lack of scaleability and programmability of the underlying Eth rails (which I address in the Operating System section later on), but the desire to iterate in this way was clear.
More speculatively, Justin Murphy's excellent “Social AI" essay, walks through what a self-improving Reddit-type-forum might feel like, as well as what Ethereum with better programmability and useability would mean.
Light things want to iterate and self-improve: to turn into Engines. On TradDollar infra, they can only do so at very large, coarse scale, due to the friction involved. Blockchain hardness promises to alleviate this friction, but so far has not been able to close the loop and become fully scaleable and programmable.
The stage is now set to discuss the BlockDollar: the infrastructure to which Engines will gravitate.
To recap, blockchain hardness provides the possibility for a form of hardness that is efficient and programmable. As financial infrastructure, it's not well-suited to "heavy" uses, but it does unlock promising ground for light, self-improving engines of code & content.
Now, let's move from "why would you want a BlockDollar?" to "what will it actually look like"?
I chose the word "dollar" deliberately for both Block and Trad Dollars, to emphasize that this is not a crypto-utopian vision of immoral fiat being smoten by divine wrath. Rather, it acknowledges the success and utility of the dollar system in its sphere, while looking for opportunities in its efficiency limitations.
In addition, it's possible (likely?) that this light rails will choose a unit of account that has high external adoption, and the US dollar is the best candidate for that. That doesn't preclude crypto L1 assets and other tokens as long-term stores of value (I think that's likely), but does mean that various forms of stablecoins may be with us for awhile in the light world of Engines. Analogously, the TradDollar system uses "unstaked" dollars (h/t @TheStalwart) for unit of account, while storing value in much more volatile stonks, bonds and real estate.
The BlockDollar will need an underlying settlement layer to provide its hardness. Any decentralized, programmable, and widely adopted Layer 1 will work for this; Ethereum is by far the most likely to win given its headstart and network effects.
Ethereum's role in this system will be providing hardness and a long-term monetary storage asset. Staking and burning allow it to distribute upside as the blockchain economy grows, and in turn make its hardness more attractive. By analogy with the traditional rails, it would be as if index funds, banks, and the Fed were fused into one thing that could be bought into by the general populace.
On top of Eth (and layer2s), you can then build all the pieces of global state that Engines require: equity, assets, permissions, agreements, etc.
If all this sounds to you suspiciously like 2017-era ICO promises…you're not wrong. Those projects were very much selling the promise of Engines, but were missing second critical BlockDollar component: an operating system.
Blockchain hardness, in isolation, is sterile. The only thing it can easily enable is self-referential financial activity: tokens that power systems that provide liquidity for trading tokens. This sterility arises because blockchains can arbitrate the agreements and incentives that drive Engines, but can't represent the very thing Engines are made of: code and content. (To be more precise, they can store small amounts of code and content, but not at nearly the scale required to be interesting or useful.)
Current computing primitives aren't good enough either: it's hard to compose software in the cloud (think of the difficulty of your SaaS apps talking), and it's hard to get those apps to talk to blockchains in interesting ways.
To make code+content more "liquid", and let it feed back into Engines, we need operating systems that seamlessly unite all the pieces of the application stack:
networked identity
global state read/write (blockchain access)
software hosting and distribution
software upgrades
inter-app communication and composability
You can think of this as taking everything that is light, and drawing it into one place to make it programmable.
People in crypto broadly agree that making blockchains programmable will be a huge unlock, and signficant work is going into this area. Below are some possible approaches, in what I see as increasing order of promise:
Glue things together more betterly. I am skeptical of this approach, because it's the somewhat-failed status quo, and critique it in theory here and with an illustration here.
Create a full-service web2 programming and hosting platform that integrates crypto. Replit's future vision is the most promising imo.
Use a P2P OS and integrate an Eth L2 with it natively. Uqbar Network on Urbit is the furthest along in this direction.
Regardless of which approach wins, significant dev talent is attacking the problem of connecting light things programmatically to blockchains, and a breakthrough is a matter of when, not if.
We started this essay by asking "what are blockchains good for?" and "what types of productive activity can happen only/best on blockchains?"
The BlockDollar answers that: low-friction blockchain hardness is the perfect infrastructure for self-improving Engines that operate on the big light things: code, content, and capital.
This answer departs heavily from the traditional crypto evangelist view that "decentralized financial systems" will sweep all before them on their own. Finance is rails, but something has to run on those rails for them to be valuable. In addition, it suggests that current DeFi was a false fork: it looked like a massive leap forward, but can't drive real adoption on its own, without anything to be financing.
Our final question we began with was "why hasn't this happened yet?" Historically, huge productivity unlocks come from not just one transformative technology, but from the convergence of multiple. As blockchain hardness becomes scaleable, new operating system primitives can complete that convergence, creating a powerful new whole, the BlockDollar.
I've talked about early forms of these ideas a couple times on the Network Age podcast:
[convergence of multiple unlocking technologies, with Josh Rosenthal] (https://rss.com/podcasts/thenetworkage/626289/)
Multiple elements of this topic deserve longform followups. Here is a short FAQ that I get. Reach out to me on Twitter @basilesportif or Urbit (~timluc-miptev) to add your own questions, and I'll address them in Tweet or longform.
Are you saying the world doesn't need credit?? We're never going back to a pre-credit world; lending activity is too critical.
Does it matter which unit transactions happen in? Does the dollar accrue value because purchases have to be in the dollar?
Doesn't the US just force everyone to use the dollar? How can you compete with that?
What about stablecoins? Aren't they putting TradDollar back into the mix, even down to having real-world links?
Isn't Stripe an example of TradDollar rails being successfully used to massively unlock light things?
What about social media? It seems to be wildly successful, is obviously light, and runs almost completely without crypto.
Where do games fit into this model?