The Tinkering SocietySolving Fractured Liquidity with Size Lending
Most successful first generation DeFi products were pool-based. Aave struggled to get traction with a P2P model (as ‘EthLend’) until they conceptualized their pool-based model with variable rates. The aggregated liquidity and simplicity (along with a confluence of DeFi Summer hype and low interest rates) built Aave into a product currently valued at more than $1B. Recently, we’ve seen order book DEXs like GMX and dYdX thrive where we once only had pool-based DEXs like Uniswap. And yet lending...
The Tinkering SocietyMilestones Towards a Long Position
As mentioned previously, I believe crypto-specific blow-ups and forced liquidation is behind us. Yet the gm Portfolio is in cash with the thesis that crypto will go down with the rest of the public markets should the economy enter a recession and public equities suffer a bear market. Here are some things we are monitoring as milestones along a way towards taking long positions in crypto: 1. Yield curves steepen (un-invert). Currently, the US Treasury 2-year/10-year and the 3-month/10-year spr...
The Tinkering SocietyFair Launches and Neutrality
Are fair launches really that equal? And, if not, how can we improve them?(Fair) Launch CodesTo add context, here’s a quick primer on the concept. These are token distribution models which are designed to favor no individual or group. There are no founders’ allocations, seed round, or ICO that provides preferential coin or token access. Yearn.Finance is a prime example. Zero $YFI was allocated to presales or ICOs, or even the founder. Early distribution was primarily shared out between the fi...
We’re builders and thinkers on a mission to further develop the crypto ecosystem through protocol research and incubation.
The Tinkering SocietySolving Fractured Liquidity with Size Lending
Most successful first generation DeFi products were pool-based. Aave struggled to get traction with a P2P model (as ‘EthLend’) until they conceptualized their pool-based model with variable rates. The aggregated liquidity and simplicity (along with a confluence of DeFi Summer hype and low interest rates) built Aave into a product currently valued at more than $1B. Recently, we’ve seen order book DEXs like GMX and dYdX thrive where we once only had pool-based DEXs like Uniswap. And yet lending...
The Tinkering SocietyMilestones Towards a Long Position
As mentioned previously, I believe crypto-specific blow-ups and forced liquidation is behind us. Yet the gm Portfolio is in cash with the thesis that crypto will go down with the rest of the public markets should the economy enter a recession and public equities suffer a bear market. Here are some things we are monitoring as milestones along a way towards taking long positions in crypto: 1. Yield curves steepen (un-invert). Currently, the US Treasury 2-year/10-year and the 3-month/10-year spr...
The Tinkering SocietyFair Launches and Neutrality
Are fair launches really that equal? And, if not, how can we improve them?(Fair) Launch CodesTo add context, here’s a quick primer on the concept. These are token distribution models which are designed to favor no individual or group. There are no founders’ allocations, seed round, or ICO that provides preferential coin or token access. Yearn.Finance is a prime example. Zero $YFI was allocated to presales or ICOs, or even the founder. Early distribution was primarily shared out between the fi...
We’re builders and thinkers on a mission to further develop the crypto ecosystem through protocol research and incubation.
Subscribe to The Tinkering Society
Subscribe to The Tinkering Society
Share Dialog
Share Dialog
<100 subscribers
<100 subscribers
As a simple mind experiment, let’s plug in our brains and visualize an extreme case of AI disruption.
In this not-so-impossible future, everything we know today is getting commoditized, reduced to fixed computation costs.
This is already happening with:
Repetitive tasks (data entry, administrative)
Well-documented tasks (monthly reports, taxes)
Routine software development (websites, single-page applications, data analysis)
Early design work (product, visual, conceptual)
Light research and analysis
But we can imagine this trend continuing, with many advanced tasks semi- or fully automated. In a few short years, it’s likely that 90% of the computer work we do today on a daily basis has been taken on by the machine.
If one thing is certain, it is that humans are notoriously good at leveraging excess creativity (surplus of time → increased capacity for creative thinking), and this won’t change with the introduction of mainstream AI. So what’s next?
The “prompt engineers” of today are the “context engineers” of tomorrow. Humans will take a more product-management type role controlling and adjusting the flow of tasks.
I think “context engineering” will become the new “prompt engineering”. Context management is an ongoing evaluation of progress towards both high and low level goals. While next-gen AI will surely be able to handle simple context, human input and alignment towards proper contextual nuance will be essential. The question of “where should we spend our time/resources” often relies on intuition, on-going reflection, a mission or raison d'être, and an understanding of time management. These human strengths will endure for quite some time.
And then there’s the physical world.
A foreign concept to many of us that live in the reflections of our laptops, the real world remains very real and difficult to navigate. Interfacing between the digital and physical worlds is complex and requires advanced edge case management (like that time it took 10 trips to home depot to fix the garage door despite having tried to think ahead at every step).
Almost everything involving manipulating or interfacing with the real world will need humans, at least until the humanoids extend a helping hand.
And speaking of humanoids, we can extend the above thought experiment to robotics as well. At some point in the future we will have a similar wave in robotics to that of LLMs today, where we find that robots can suddenly do almost every simple task that humans can.
What then becomes valuable is guiding the robots and managing their tasks and context.
This is a big one in my opinion and the most under-explored.
Value increases with scarcity, and amid a new world of machine-generated content and guidance, there is less of “that which is human”.
Let’s take a simple example: you’re hiring for a new role at your company. All of the job candidates are machine-polished on fine digital paper, and are even prompted on the video call interview with on-screen reminders and suggested responses.
What do we look for to choose a candidate? We might look for personality, signs of abstract thinking, spontaneity, and passion. Or in other words, humanness. It may even be the case some day (soon?) that video interviews are no longer enough and we must get together in person to measure these characteristics.
So here we may have personality valued much higher than it is today (witty thinking, passion, humor).
And where do we develop personality? In a large part from other humans. Through networking, authenticity, shared creative experience, and teamwork.
As a simple mind experiment, let’s plug in our brains and visualize an extreme case of AI disruption.
In this not-so-impossible future, everything we know today is getting commoditized, reduced to fixed computation costs.
This is already happening with:
Repetitive tasks (data entry, administrative)
Well-documented tasks (monthly reports, taxes)
Routine software development (websites, single-page applications, data analysis)
Early design work (product, visual, conceptual)
Light research and analysis
But we can imagine this trend continuing, with many advanced tasks semi- or fully automated. In a few short years, it’s likely that 90% of the computer work we do today on a daily basis has been taken on by the machine.
If one thing is certain, it is that humans are notoriously good at leveraging excess creativity (surplus of time → increased capacity for creative thinking), and this won’t change with the introduction of mainstream AI. So what’s next?
The “prompt engineers” of today are the “context engineers” of tomorrow. Humans will take a more product-management type role controlling and adjusting the flow of tasks.
I think “context engineering” will become the new “prompt engineering”. Context management is an ongoing evaluation of progress towards both high and low level goals. While next-gen AI will surely be able to handle simple context, human input and alignment towards proper contextual nuance will be essential. The question of “where should we spend our time/resources” often relies on intuition, on-going reflection, a mission or raison d'être, and an understanding of time management. These human strengths will endure for quite some time.
And then there’s the physical world.
A foreign concept to many of us that live in the reflections of our laptops, the real world remains very real and difficult to navigate. Interfacing between the digital and physical worlds is complex and requires advanced edge case management (like that time it took 10 trips to home depot to fix the garage door despite having tried to think ahead at every step).
Almost everything involving manipulating or interfacing with the real world will need humans, at least until the humanoids extend a helping hand.
And speaking of humanoids, we can extend the above thought experiment to robotics as well. At some point in the future we will have a similar wave in robotics to that of LLMs today, where we find that robots can suddenly do almost every simple task that humans can.
What then becomes valuable is guiding the robots and managing their tasks and context.
This is a big one in my opinion and the most under-explored.
Value increases with scarcity, and amid a new world of machine-generated content and guidance, there is less of “that which is human”.
Let’s take a simple example: you’re hiring for a new role at your company. All of the job candidates are machine-polished on fine digital paper, and are even prompted on the video call interview with on-screen reminders and suggested responses.
What do we look for to choose a candidate? We might look for personality, signs of abstract thinking, spontaneity, and passion. Or in other words, humanness. It may even be the case some day (soon?) that video interviews are no longer enough and we must get together in person to measure these characteristics.
So here we may have personality valued much higher than it is today (witty thinking, passion, humor).
And where do we develop personality? In a large part from other humans. Through networking, authenticity, shared creative experience, and teamwork.
No activity yet