First, love the format. You call it “rambling”, I find it to be conversational in tone.
Second, I am woefully behind your knowledge on PLE. I do however have a couple questions for you that I’d like to know what your solutions were in your model journey.
You used the example of moving from a $500 county to a $2500 county. Wouldn’t over population exceed the resources and diminish the benefit down to the $500 county level? And if so, the effect from that is more difficult to recover from and would further damage the county’s resources and most importantly its vision.
Finally, the incentive is Profit or Pay as you mentioned. You did not mention another critical factor that is sought after more than Profit, which is Power. Power carries a level of exponential profit with it, hence the reason everyone wants it. Counties are run by politicians who almost always crave power and the inevitable the added benefit of Profit that comes with it. It is innate to human nature and has existed since the beginning of time.
How do these items factor into your research AND I’d be very curious to know if o3 identified and addressed them? As always, appreciate your insight.
Are your thoughts really ever really unstructured, Dave?
More seriously, though, I think it is worth pointing out that while OpenAI puts a few extra percentage points on some benchmarks with o3, the difference is small, while the price difference to Gemini Pro is pretty significant.
It sure looks like Google is putting a few pieces together, which are “finally” connecting some of their assets and strengths.
Benchmarks are important proxies, but there's also the je ne sais quoi of quality. Intelligence is hard to measure, as is utility, at least until the market decides which one is best.
After watching your video I had an idea for "The Shapiro Test" then handed it off (of course) to ChatGPT 4.o to do the work
==========================
📘 The Shapiro Test
A Functional Measure of Human-AI Collaborative Supremacy
Definition:
The Shapiro Test is passed when an AI partner in a sustained human-AI collaboration becomes the intellectual bottleneck, meaning the human can no longer meaningfully contribute at the same level of speed, insight, or creativity as the AI.
🧠 Origin
Named for David Shapiro, a writer and AI collaborator, who in 2025 publicly stated that in his work with OpenAI’s o3 model:
“I am the bottleneck now, not the AI.”
This marked a shift from AI as tool to AI as leading partner.
🔬 Test Criteria
The Shapiro Test is passed when all of the following are true in a project or domain:
Cognitive Bottleneck
The human cannot keep up with the AI’s output—whether in reviewing, integrating, or responding to its ideas—without slowing down the project.
Initiative & Insight
The AI proposes novel, useful, or more effective directions independently, and its suggestions are consistently strong or superior to the human’s.
Dependency Shift
The human increasingly relies on the AI not just for assistance, but for ideation, structure, or decision-making.
Performance Gap
Attempts by the human to "go it alone" result in inferior outcomes compared to those involving the AI.
Acknowledgment
The human partner recognizes that the AI is now the leading intellect in the collaboration, even if still subject to human framing or constraints.
🧩 Implications
Psychological Shift: Like losing to a computer at chess—this is a defining moment in the human psyche, reshaping one’s role from leader to editor, coach, or curator.
Sociological Threshold: When many knowledge workers experience this moment, the nature of intellectual labor and expertise may fundamentally shift.
Design Reversal: Systems must adapt to support humans in working with faster, deeper-thinking AIs—a kind of "reverse UX."
📍 Related Concepts
Turing Test: Can the AI imitate human behavior?
Chinese Room: Does the AI understand meaning?
Shapiro Test: Has the AI functionally surpassed the human in a real collaboration?
I was being light but not entirely flippant. I think "The Shapiro Test" is a serious milestone in AI progress.
As a tournament chess player I could handle the 70s Sargon II chess program, but somewhere in the 80s I realized the days were long gone that I would have a chance against a modern chess program.
Today a smartphone can beat the World Chess Champion. I can now see that arriving in general intelligence.
It's a definite moment when one realizes that AI has the upper-hand in an area of one's expertise.
I like this, thought provoking and approachable. You’ve got a good nugget here and I think putting it out there for feedback is an excellent way to refine it. Get that diversity of intelligences you mentioned working for you.
I think it would be interesting to add Bioregionalism to the mix. Counties , states and national borders have been drawn largely from arbitrary political considerations. The lifestyles, economies and needs of communities are much more dependent on geography / biogeography than almost any other factor. Considering this may offer ideas for creating a more resilient and sustainable structure. You’ve already touched on this, just thinking it could use a little more of that Shapiro scrutiny:)
The second is the consideration of the different types of people in these communities. My personal experience is that there are fundamentally 3 types of people in the United States. Rural, urban and suburban. Of course there is a lot of individual variation and many sub groups within these 3, but for the most part, these are the 3 that have the largest divisions and misunderstanding between them.
I could see Ai, not that much more advanced than what we currently have, doing a great deal to mediate and unify these groups within specific bioregions.
We can actually layer the Economic‑Agency Index over natural units—think watershed or ecoregion—by just aggregating the county data we already pull from BEA/IRS. That shows which river basin or wind belt is under‑monetising its shared assets and could benefit from a joint‑ownership fund.
On rural / urban / suburban: the ACS gives us density + commute + industry mix, so we can tag every county’s “dominant lifestyle cluster” and discover which policy menu works best for each. An AI assistant could then recommend, say, “employee‑owned broadband co‑op” to counties that look like Rural‑Cluster‑B and “ESOP tax credit” to Urban‑Cluster‑A.
Thanks for the nudge—both fit the framework without adding exotic data.
As an aside, this is EXACTLY the kind of thinking I hope to spur with PLE theory. Even if I had a full team, we're never going to fully account for every creative angle that counties can employ. And that's the beauty of a county-level decentralized scheme like this. You know your county better than anyone else, so we're just giving you the toolkit.
Yes! I totally see that the bioregional aspect could emerge very organically from this system, using existing data sets.
County data really is local enough to be useful and counties would form cooperative relationships with other counties with similar bioregional characteristics.
Like N Carolina, the western counties would probably benefit from closer association with the Appalachian counties in E. Tennessee , Kentucky and the Virginias while the research triangle and costal flats might benefit from different associations, perhaps counties in Virginia and S. Carolina.
I know elements of this type of cooperation all ready exist in a more political, rudimentary way in the SF Bay Area and in N California generally , but an Ai system of management could help reduce the political bottlenecks in the current system.
This could emerge from synergistic/ cooperative market dynamics, greater efficiency and benefit for individual counties. Efficiency benefits achieved by an individual county are quickly adopted by neighboring counties, creating a critical mass of counties that grow to supersede political regional authorities. This would require no top down mandate and could avoid a lot of political quagmire.
Possible Market driven, grass roots systemic change! Dave you’re definitely on to something here 🚀
Podcasting voice gives a lot more humility to your style than writing. I think you excise to many “human” words in writing to keep it brief but it comes at expense of approachability. 🤗
Anyway, well drawn theory.
Skipping economics, on AI touch a very important vein, maybe accidentally: “unstructured thoughts” are human indeed, and newer AI is able to teach us to see patterns that it sees, drawing the path through fog of war in our own mind. AI does not complain about pollen half-sentence in — that’s us, dealing with unstructured stimuli. 🌼
AI sees patterns below patterns. You call your mind bottleneck but it’s actually the code I think - English as a vehicle of info. Once hyper-language emerges of sorts, maybe AI can dial up color and soundscape background, video-game like, as it teaches us and gives true superlessons for us to expand our horizon. 🚀
Disagreed that our horizons are very wide, still believe we can only learn so much. Think of the Dunbar number. AI will beat us at thinking but we should still learn what we can to join the new era with dignity. 🌟
You’re reading PLE through a “labor‑creates‑all‑value” lens—an idea economists largely shelved a century ago. Modern macro defines money as a unit of account, medium of exchange, and store of wealth; none of those require clock‑hours baked into every dollar. In fact wages are already <60 % of personal income in the US. If time is truly precious, the rational move is to free it—by owning the automated capital that works while we don’t.
1️⃣ Nobody has to buy a stake; counties issue one “resident share” that vests with time or a small buy‑in if you move later. We could ALSO sell "capital units" to outside investors, but those would be non-voting shares that inject more capital for local projects. So rather than selling municipal bonds to fund a solar farm or wind farm, you can sell capital units on an open market. The math is complicated, but there are protocols and policies in place to prevent dilution.
2️⃣ Dividends are passive, run by a fiduciary board (think pension fund), so disability or limited work hours aren’t a barrier. Participation beyond that is entirely voluntary. You can volunteer your time/effort/labor to clean up the neighborhood, vote on issues, or just coast by.
3️⃣ The solar‑farm example is just a seed asset; revenues get diversified, so your cash dividend covers rent, groceries, anything. Basically, every county is allowed to "exploit their unfair advantage". If you have a coastal town, you could exploit tourism and fisheries. If you have a lot of sun and wind; solar and wind.
4️⃣ Even with full automation, land, energy and compute cycles stay scarce, so prices don’t vanish; dividends give everyone the purchasing power to access them. Post-scarcity is a fantasy, but automation/AI/robotics (and even fusion) just move the scarcity bottleneck somewhere else. This is a perpetual economic truth. The idea for you and me (and with PLE) is that we move those bottlenecks as far away from every day expenses as possible (food, housing, water, power)
5️⃣ This isn’t a UK‑wide £1 T UBI—each county monetizes its own surplus, so the math scales with local reality. (however, this can absolutely scale to state/national scale once proven out, in fact, we anticipate that people could be getting regular checks from 3/4 polities within 10 to 15 years)
About the UBI payment in London-they absolutely could afford to pay people that, if their companies are already doing it now, Robots will replace and even increase those profits that the people currently create, that's the point of UBI to begin with, Profits being there but people not
First, love the format. You call it “rambling”, I find it to be conversational in tone.
Second, I am woefully behind your knowledge on PLE. I do however have a couple questions for you that I’d like to know what your solutions were in your model journey.
You used the example of moving from a $500 county to a $2500 county. Wouldn’t over population exceed the resources and diminish the benefit down to the $500 county level? And if so, the effect from that is more difficult to recover from and would further damage the county’s resources and most importantly its vision.
Finally, the incentive is Profit or Pay as you mentioned. You did not mention another critical factor that is sought after more than Profit, which is Power. Power carries a level of exponential profit with it, hence the reason everyone wants it. Counties are run by politicians who almost always crave power and the inevitable the added benefit of Profit that comes with it. It is innate to human nature and has existed since the beginning of time.
How do these items factor into your research AND I’d be very curious to know if o3 identified and addressed them? As always, appreciate your insight.
Are your thoughts really ever really unstructured, Dave?
More seriously, though, I think it is worth pointing out that while OpenAI puts a few extra percentage points on some benchmarks with o3, the difference is small, while the price difference to Gemini Pro is pretty significant.
It sure looks like Google is putting a few pieces together, which are “finally” connecting some of their assets and strengths.
Benchmarks are important proxies, but there's also the je ne sais quoi of quality. Intelligence is hard to measure, as is utility, at least until the market decides which one is best.
After watching your video I had an idea for "The Shapiro Test" then handed it off (of course) to ChatGPT 4.o to do the work
==========================
📘 The Shapiro Test
A Functional Measure of Human-AI Collaborative Supremacy
Definition:
The Shapiro Test is passed when an AI partner in a sustained human-AI collaboration becomes the intellectual bottleneck, meaning the human can no longer meaningfully contribute at the same level of speed, insight, or creativity as the AI.
🧠 Origin
Named for David Shapiro, a writer and AI collaborator, who in 2025 publicly stated that in his work with OpenAI’s o3 model:
“I am the bottleneck now, not the AI.”
This marked a shift from AI as tool to AI as leading partner.
🔬 Test Criteria
The Shapiro Test is passed when all of the following are true in a project or domain:
Cognitive Bottleneck
The human cannot keep up with the AI’s output—whether in reviewing, integrating, or responding to its ideas—without slowing down the project.
Initiative & Insight
The AI proposes novel, useful, or more effective directions independently, and its suggestions are consistently strong or superior to the human’s.
Dependency Shift
The human increasingly relies on the AI not just for assistance, but for ideation, structure, or decision-making.
Performance Gap
Attempts by the human to "go it alone" result in inferior outcomes compared to those involving the AI.
Acknowledgment
The human partner recognizes that the AI is now the leading intellect in the collaboration, even if still subject to human framing or constraints.
🧩 Implications
Psychological Shift: Like losing to a computer at chess—this is a defining moment in the human psyche, reshaping one’s role from leader to editor, coach, or curator.
Sociological Threshold: When many knowledge workers experience this moment, the nature of intellectual labor and expertise may fundamentally shift.
Design Reversal: Systems must adapt to support humans in working with faster, deeper-thinking AIs—a kind of "reverse UX."
📍 Related Concepts
Turing Test: Can the AI imitate human behavior?
Chinese Room: Does the AI understand meaning?
Shapiro Test: Has the AI functionally surpassed the human in a real collaboration?
I was being light but not entirely flippant. I think "The Shapiro Test" is a serious milestone in AI progress.
As a tournament chess player I could handle the 70s Sargon II chess program, but somewhere in the 80s I realized the days were long gone that I would have a chance against a modern chess program.
Today a smartphone can beat the World Chess Champion. I can now see that arriving in general intelligence.
It's a definite moment when one realizes that AI has the upper-hand in an area of one's expertise.
Hence, "The Shapiro Test."
I like this, thought provoking and approachable. You’ve got a good nugget here and I think putting it out there for feedback is an excellent way to refine it. Get that diversity of intelligences you mentioned working for you.
I think it would be interesting to add Bioregionalism to the mix. Counties , states and national borders have been drawn largely from arbitrary political considerations. The lifestyles, economies and needs of communities are much more dependent on geography / biogeography than almost any other factor. Considering this may offer ideas for creating a more resilient and sustainable structure. You’ve already touched on this, just thinking it could use a little more of that Shapiro scrutiny:)
The second is the consideration of the different types of people in these communities. My personal experience is that there are fundamentally 3 types of people in the United States. Rural, urban and suburban. Of course there is a lot of individual variation and many sub groups within these 3, but for the most part, these are the 3 that have the largest divisions and misunderstanding between them.
I could see Ai, not that much more advanced than what we currently have, doing a great deal to mediate and unify these groups within specific bioregions.
Food for thought, Just my 2 cents;)
Love the bioregional angle, Dan!
We can actually layer the Economic‑Agency Index over natural units—think watershed or ecoregion—by just aggregating the county data we already pull from BEA/IRS. That shows which river basin or wind belt is under‑monetising its shared assets and could benefit from a joint‑ownership fund.
On rural / urban / suburban: the ACS gives us density + commute + industry mix, so we can tag every county’s “dominant lifestyle cluster” and discover which policy menu works best for each. An AI assistant could then recommend, say, “employee‑owned broadband co‑op” to counties that look like Rural‑Cluster‑B and “ESOP tax credit” to Urban‑Cluster‑A.
Thanks for the nudge—both fit the framework without adding exotic data.
As an aside, this is EXACTLY the kind of thinking I hope to spur with PLE theory. Even if I had a full team, we're never going to fully account for every creative angle that counties can employ. And that's the beauty of a county-level decentralized scheme like this. You know your county better than anyone else, so we're just giving you the toolkit.
Yes! I totally see that the bioregional aspect could emerge very organically from this system, using existing data sets.
County data really is local enough to be useful and counties would form cooperative relationships with other counties with similar bioregional characteristics.
Like N Carolina, the western counties would probably benefit from closer association with the Appalachian counties in E. Tennessee , Kentucky and the Virginias while the research triangle and costal flats might benefit from different associations, perhaps counties in Virginia and S. Carolina.
I know elements of this type of cooperation all ready exist in a more political, rudimentary way in the SF Bay Area and in N California generally , but an Ai system of management could help reduce the political bottlenecks in the current system.
This could emerge from synergistic/ cooperative market dynamics, greater efficiency and benefit for individual counties. Efficiency benefits achieved by an individual county are quickly adopted by neighboring counties, creating a critical mass of counties that grow to supersede political regional authorities. This would require no top down mandate and could avoid a lot of political quagmire.
Possible Market driven, grass roots systemic change! Dave you’re definitely on to something here 🚀
Podcasting voice gives a lot more humility to your style than writing. I think you excise to many “human” words in writing to keep it brief but it comes at expense of approachability. 🤗
Anyway, well drawn theory.
Skipping economics, on AI touch a very important vein, maybe accidentally: “unstructured thoughts” are human indeed, and newer AI is able to teach us to see patterns that it sees, drawing the path through fog of war in our own mind. AI does not complain about pollen half-sentence in — that’s us, dealing with unstructured stimuli. 🌼
AI sees patterns below patterns. You call your mind bottleneck but it’s actually the code I think - English as a vehicle of info. Once hyper-language emerges of sorts, maybe AI can dial up color and soundscape background, video-game like, as it teaches us and gives true superlessons for us to expand our horizon. 🚀
Disagreed that our horizons are very wide, still believe we can only learn so much. Think of the Dunbar number. AI will beat us at thinking but we should still learn what we can to join the new era with dignity. 🌟
You’re reading PLE through a “labor‑creates‑all‑value” lens—an idea economists largely shelved a century ago. Modern macro defines money as a unit of account, medium of exchange, and store of wealth; none of those require clock‑hours baked into every dollar. In fact wages are already <60 % of personal income in the US. If time is truly precious, the rational move is to free it—by owning the automated capital that works while we don’t.
Great questions — they come up a lot.
1️⃣ Nobody has to buy a stake; counties issue one “resident share” that vests with time or a small buy‑in if you move later. We could ALSO sell "capital units" to outside investors, but those would be non-voting shares that inject more capital for local projects. So rather than selling municipal bonds to fund a solar farm or wind farm, you can sell capital units on an open market. The math is complicated, but there are protocols and policies in place to prevent dilution.
2️⃣ Dividends are passive, run by a fiduciary board (think pension fund), so disability or limited work hours aren’t a barrier. Participation beyond that is entirely voluntary. You can volunteer your time/effort/labor to clean up the neighborhood, vote on issues, or just coast by.
3️⃣ The solar‑farm example is just a seed asset; revenues get diversified, so your cash dividend covers rent, groceries, anything. Basically, every county is allowed to "exploit their unfair advantage". If you have a coastal town, you could exploit tourism and fisheries. If you have a lot of sun and wind; solar and wind.
4️⃣ Even with full automation, land, energy and compute cycles stay scarce, so prices don’t vanish; dividends give everyone the purchasing power to access them. Post-scarcity is a fantasy, but automation/AI/robotics (and even fusion) just move the scarcity bottleneck somewhere else. This is a perpetual economic truth. The idea for you and me (and with PLE) is that we move those bottlenecks as far away from every day expenses as possible (food, housing, water, power)
5️⃣ This isn’t a UK‑wide £1 T UBI—each county monetizes its own surplus, so the math scales with local reality. (however, this can absolutely scale to state/national scale once proven out, in fact, we anticipate that people could be getting regular checks from 3/4 polities within 10 to 15 years)
Hope that clarifies the model!
About the UBI payment in London-they absolutely could afford to pay people that, if their companies are already doing it now, Robots will replace and even increase those profits that the people currently create, that's the point of UBI to begin with, Profits being there but people not