Carlos Perez on How AI Solves the Cost of Coordination Problem
Discussion
From an X-thread by Carlos E. Perez:
"1/
Everyone's debating whether AI will take our jobs.
But we're missing the bigger story: AI is about to solve the problem that made us choose between markets and planning in the first place.
Thread on the coordination revolution no one's talking about
2/
Here's the paradox that should keep you up at night:
We can build enough housing for everyone. We have the construction capacity.
Yet homelessness persists.
Why? Not scarcity of resources. Scarcity of coordination.
3/
In 1945, economist Friedrich Hayek wrote "The Use of Knowledge in Society."
His core insight: No central planner can coordinate millions of people. Too much information. Too complex. Too fast-changing.
Markets won because prices solve the calculation problem automatically.
4/
The Soviet Union proved Hayek right.
They tried central planning. Result?
Empty shelves despite full warehouses
- Construction taking decades
- Systematic lying about production
- Eventual collapse
The calculation problem was computationally intractable for humans.
5/
So we accepted the trade-off:
✅ Markets coordinate efficiently
❌ But only for people with money ❌ And require artificial scarcity for profit ❌ And externalize environmental costs ❌ And optimize for quarterly returns
"There is no alternative," Thatcher said.
6/
But here's what changed:
Intelligence cost is approaching zero.
That constraint Hayek identified? The one that made central planning impossible?
It just became... solvable.
7/
Think about what AI + IoT enables:
• Real-time data from millions of sensors • Processing 10^15+ variable optimizations instantly • Preference inference without price signals • Continuous adaptation (not 5-year plans) • No human intermediaries to game the system
The calculation problem is solved.
8/
This unlocks three architectural possibilities that were science fiction 10 years ago:
- Algorithmic Abundance Management
- Hybrid Planning-Market Systems
- Evolutionary Stable Abundance
Let me break down each one.
9/
Solution #1: Algorithmic Abundance Management
Replace markets for commodities with AI coordination.
Example: Housing
- AI senses: "500K people need housing in Region X"
- Not: "How many can afford housing?"
- But: "What housing serves their actual needs?"
10/
The AI then:
- Identifies available land/materials
- Optimizes for cost, sustainability, preference
- Coordinates construction schedules
Matches people to units based on need + choice
Updates continuously based on feedback
Result: Everyone housed. No artificial scarcity needed.
11/
"But Soviet planning failed!"
Yes. Because:
❌ Human planners couldn't process the information ❌ Bureaucrats gamed the metrics ❌ Updates took years ❌ No way to know preferences without prices
AI eliminates ALL these constraints.
12/
Hayek: "Central planning can't aggregate local knowledge"
AI: "I can process local signals from 8 billion sources simultaneously"
Hayek: "Prices are the information transmission mechanism"
AI: "I can infer preferences directly from behavior + stated needs"
The debate just shifted.
13/
Solution #2: Hybrid Planning-Market System
"But who controls the AI? Sounds dystopian."
Fair. So add democracy.
Strategic Layer (40 years): Citizens deliberate on goals Tactical Layer (10 years): Experts model pathways Operational Layer (Daily): AI coordinates, humans choose
14/
Example: Healthcare
Citizens decide: "Universal healthcare, emphasizing prevention"
AI models:
- Option A: $800B/year, 2-day wait times, 8 years
- Option B: $600B/year, 5-day waits, 5 years
Democracy chooses values. AI optimizes execution.
15/
This is what China does (strategic coherence) + What the West needs (democratic legitimacy).
The best of both:
- Long-term planning
- Democratic goal-setting
- Transparent algorithms
- Individual choice within abundance
16/ Solution #3: Evolutionary Stable Abundance
"Can't we just fix capitalism instead of replacing it?"
Yes! Use AI to change the game theory.
Make abundance MORE profitable than scarcity.
17/
How? Three mechanisms:
Perfect transparency → Scarcity engineering gets detected instantly
Instant alternative coordination → Supply restrictions become unprofitable
Reputation systems at scale → Cooperation rewards exceed extraction rewards
18/
Example: Company tries restricting supply to inflate prices.
Pre-AI: Takes months for competitors to respond.
AI-enabled:
- Restriction detected in hours
- Alternative suppliers matched to consumers automatically
- New production spins up in days
Scarcity engineering becomes non-viable.
19/
This already happened with software!
Open source (Linux, Python, Apache) became MORE profitable than proprietary for many domains.
Why? Network effects + reputation > artificial scarcity.
AI generalizes this pattern to physical goods.
20/
Now here's where it gets wild:
These three solutions REINFORCE each other.
Solution #3 changes incentives → builds political support for #2 → enables deployment of #1 → makes #3 more stable
Virtuous cycle toward abundance.
21/
Compare this to alternatives:
❌ Pure markets: Efficient but exclude the poor, engineer scarcity ❌ Soviet planning: Failed calculation problem ❌ Social democracy: Better but still accepts artificial scarcity ❌ UBI alone: Provides income but landlords capture it
22/
AI solutions are different because they address the ROOT problem:
Not "how do we redistribute better?"
But "how do we coordinate for actual abundance instead of artificial scarcity?"
23/
The moral dimension is stark:
Once coordination becomes computationally trivial, maintaining artificial scarcity becomes a choice, not an inevitability.
We can no longer hide behind "market efficiency" when algorithmic coordination is provably more efficient.
24/
"This sounds utopian."
No. It's engineering.
The productive capacity exists. The sensor networks exist. The AI capabilities exist.
What's missing is the institutional architecture to combine them.
25/
Testable predictions:
If this is real, pilot projects should show:
- Lower cost per person served
- Higher satisfaction
- Less waste
- Faster adaptation to disruptions
These aren't philosophical debates. They're hypotheses to test.
26/
The implementation path isn't "wake up tomorrow in AI communism."
It's:
- 2025-2030: Pilot projects, proof of concept
- 2030-2040: Scale successful models
- 2040-2060: Abundance becomes default for basics 2060+: Post-scarcity civilization
Gradual. Empirical. Adaptive.
27/
Here's the thing that haunts me:
We're about to have the conversation about "AI replacing jobs."
But we're missing the bigger conversation about "AI replacing the coordination mechanisms that created artificial scarcity in the first place."
28/
The question isn't "will AI take my job?"
The question is "will we use AI to coordinate for human flourishing, or will we use AI to make scarcity engineering more sophisticated?"
That's the choice.
29/
And the clock is ticking.
Because every day we maintain artificial scarcity despite abundance capacity is a day we're choosing inequality.
Not because it's necessary. But because we haven't built the institutional architecture to do better.
30/
The Hayekian critique held weight in 1945.
It justified markets because central planning was computationally impossible.
But in 2025? That justification evaporates.
The calculation problem is solved.
What remains is political will and institutional innovation.
31/
I don't know if these solutions will work.
But I know they're testable in ways that ideological debates aren't.
Run the pilots. Gather the data. Let reality arbitrate.
If algorithmic coordination outperforms markets for commodities, we should use it.
32/
This is the most important economic transformation in human history:
Intelligence abundance → Coordination abundance → Material abundance
The question is whether we have the imagination to build it.
33/
[FINAL]
The future isn't "AI apocalypse" or "AI utopia."
It's "AI enables coordination architectures that were impossible before."
What we build with that capability is up to us.
Let's build something better than artificial scarcity."
(https://x.com/IntuitMachine/status/1981832241866244306)
More information
See also: