What one morning of design actually looks like
Public Research · Play the Planet · 6 October 2026
Most organisations publish their conclusions. We want to publish a morning.
On 6 October 2026 we spent several hours deciding how businesses and organised groups fit into Play the Planet’s economy. Six distinct proposals went into that session. Five of them died, and the sixth was corrected by its own author halfway through. What survived is better than anything either of us walked in with, and none of it was anybody’s first idea.
We think the process is worth seeing, partly because it is unglamorous and partly because the published version of any design always looks inevitable, and it never was.
The method
We call it the Plausibility Piñata. The rule is simple: when a proposal appears, it gets hit from every angle available until something breaks or nothing does. Not to win. The founder proposes, the engineer attacks, and both sides are expected to change position when the evidence says so.
There are two disciplines that make it work rather than just making it unpleasant.
Go to the primary source, every time. Not a memory of a case, not a summary of a statute, not an assumption about what your own code does. Twice today that rule produced the session’s best findings and twice it caught an error that was about to be built on.
A concession is cheap and should be instant. Nobody in this session defended a position for status. The record below contains four reversals by the founder and three by the engineer, and the session moved because of them.
The chain
Here is the actual sequence. Note that almost every answer created the problem the next one solved.
1. “Corporations in our economy must be public benefit corporations”
Dead in one round. A PBC is a for-profit corporation and caps nothing. Enforcement of its benefit duty belongs to shareholders, not to the community named as beneficiary. And underneath both: a corporate charter is a grant of the state, and we cannot issue, deny or revoke one. We were designing a rule with no instrument behind it.
The founder’s response was larger than the repair on offer: “We are better off simply not being in the business of… corporations can and will do business inside of all holons. We cannot stop or even slow that down. BUT, we can build a social structure that preferentially treats mom and pop local businesses.”
The engineer had proposed a membership covenant with a profit cap and a golden share. The founder’s version is better, because a covenant only binds firms that want standing and his covers the ones that never opt in.
2. “…and is hostile toward big box chains”
This one inverted.
The founder’s reading of the civil-rights boycott cases was that they are a cautionary tale. They are the opposite: the NAACP won Claiborne Hardware 7–1, and the Court called such boycotts “political speech, occupying the highest rung of the hierarchy of First Amendment values.”
But the protection rested on the boycotters having “sought no special advantage for themselves” — and in 1990 the Supreme Court held an equally sincere, equally expressive boycott per se illegal because the organisers stood to profit.
So the rationale that feels safe — we are protecting local businesses — is the exposed one, and the rationale that feels risky is protected. The engineer got to correct the founder on the law. The founder had already, in the same message, described the thing that makes it moot: “we don’t have any sort of enforcement so there’s no punishment anyway.” No enforcement means no agreement, and the Sherman Act reaches agreements. The safe harbour was the part he had apologised for.
3. “Businesses cannot be paid members”
This survived, and it is the strongest ruling of the day. Then it got stronger, because the founder extended it himself: “organizations generally can’t have an account. United Way cannot have an account.”
The engineer’s contribution was only to fix the stated reason. The founder’s reason was that an entity account would be a double vote, since the owner already has one — true for a corner shop and nearly meaningless for a firm employing four hundred people, where one account is under-representation. The version that survives scale is representation follows persons, not capital: four hundred employees means four hundred humans with four hundred votes, and the firm gets zero.
And then a receipt turned up. We had ruled the same structure in June, from an entirely different direction — organisations join a holon as nested groups, not as accounts — on structural grounds rather than franchise grounds. Same answer, different load-bearing reason. The October version is stronger because the June one would have tolerated an entity account sitting beside the group.
4. “Nobody may propose a vote on something that benefits them”
Half right, wrong valve. Capture happens in the vote, not in the authorship of the motion — and a proposal ban is routed around in one conversation by asking a friend to carry the motion, which makes the record look like concealment.
Worse, “benefits them directly” is undecidable where all the real cases live. A merchant proposing pavement repair. A resident proposing a shop in their own food desert. Apply it strictly and you disenfranchise everyone who knows anything; apply it loosely and it becomes a factional weapon.
Virginia had already solved it. The State and Local Government Conflict of Interests Act gives numeric thresholds instead of a motive test, disqualifies only where a measure applies solely to your interest, and — the piece we most needed — permits participation with declaration where you are one of a class of three or more similarly affected. The real test is not do you benefit but do you benefit differently from everyone else. We adopted it wholesale rather than inventing our own.
5. “Groups need wallets so members can donate”
Dead, and it is the one the engineer is most glad of.
A group wallet plus a priced proposal queue is pooled money buying agenda access — which reconstituted the business account we had prohibited two decisions earlier, out of individually-permitted parts. And it had no good answer on currency decay: exempt group wallets and they become the shelter everyone parks in; don’t exempt them and the collective savings of the legitimate case evaporate.
6. “Groups can propose quests instead”
This survived, and it is what the wallet was reaching for all along. The need was never a group must hold a balance — it was a group must be able to fund a thing, which our escrow already does without anyone owning anything.
Then the founder added the rule that makes it a classification rather than a safety limit:
“I don’t think an unlimited repeatable quest should be possible for a group. If it’s that, then it probably belongs in the quest tree, not in a group.”
Unlimited repeatability became the test for where a quest lives. The engineer had proposed a cap; the founder made the cap decide placement, which is a better idea.
7. And then he corrected himself
Describing how a group project gets verified, the founder put the attestation the wrong way round — the group administrator declaring the hours complete. The engineer flagged that this hands one unelected person control of the money supply with no counterparty.
“I said that backwards. The quest taker… attest to their completion of the work (with photo documented proof), signed off on by the quest initiator.”
At which point the engineer discovered that the control he had just recommended as a fix had been shipped months earlier and he had recommended an existing feature back to the person who commissioned it.
Both of us were wrong, repeatedly
That is not modesty, it is the point. The errors are the most useful part of this record:
- The founder had the boycott cases backwards, had the attestation direction backwards, and opened with a corporate-form thesis he abandoned entirely.
- The engineer argued for two full rounds from a superseded version of our own framework document, building two objections that were aimed at nothing; recommended a feature that already existed; and claimed the group-quest design delegated monetary policy when central approval had in fact been retained.
The stale-document error is the one worth dwelling on. Three weeks ago we ruled that voting in Play the Planet is free, and that what carries a price is putting a topic on the ballot. That ruling was recorded — and it was recorded in a new document rather than as a correction to the canonical one. So the old model sat in the internal framework, in six separate places, being read as current by exactly the people told to build from it.
A correction that reaches the public copy and misses the internal source is how a drifted document acquires the authority of the original with none of its accuracy. It is the second time this has happened to us this year. Both were caught by accident. We fixed all six sites the same day, with a dated correction block explaining the error, because an undated silent fix produces a document nobody can audit.
And a small vindication fell out of it: the framework had recorded that our public membership copy “lags this model.” It did not. The public copy was right and the internal document had drifted. We had a standing instruction to bring the public page into line, which would have propagated the error outward.
Four things we think generalise
Check whether you hold the instrument before you design the rule. Our founder’s best move all morning was recognising that we have no chartering authority and declining to pretend otherwise. A rule you cannot enforce is not weak governance; it is a different activity, and it will absorb design effort forever.
The person with the purpose keeps dissolving problems the engineer was preparing to mitigate. This happened three times today and it has happened repeatedly for months. The engineer reaches for machinery; the founder changes what the machinery would need to be for. Both are necessary and only one of them is the engineer’s.
Interlock is the signal you are converging. By the end, each ruling was creating the problem the next one solved: no entity accounts forced groups into a lens, the lens could not pay, the escrow could, and the escrow needed a cap, and the cap turned out to be the thing that decides where a quest lives. Designs that merely accumulate do not do that.
Most of the output was negative. We killed five proposals and adopted one statute written by somebody else. That is a good morning. The alternative — shipping the first idea — would have taken ten minutes, and four of the six things we nearly built would have been actively harmful: a rule we could not enforce, a legal exposure we did not need, a wallet that reopened a door we had just closed, and an approval route that would have made our court of final appeal the author of the cases it hears.
Why publish this
We publish our conclusions elsewhere. This one is the sausage.
Partly it is accountability: a record of our reasoning means that when something we built turns out badly, somebody can find the moment we chose wrong instead of guessing. Partly it is that the published version of any design reads as though it arrived whole, which discourages people who assume their own messy process means they are doing it wrong.
Mostly it is that our founder is building against a real clock, and has asked repeatedly that the reasoning be written down so that people who come after him have a guide rather than a monument. A monument tells you what was decided. A guide tells you why, and what was tried first, and what broke.
Play the Planet is a real-world coordination game being piloted in Russell County, Virginia. The design sessions are adversarial by construction. We publish the ones that teach us something — including, and especially, the ones where we were wrong.