The balance charter
TENURE IS A HYPOTHESIS · THE TUNING FORMULA IS PUBLISHED, VERSIONED, AND BINDING · NEVER MID-TENURE, NEVER BY DISCRETION
The clocks and the graduation bars are the machine's current bets about how long a founder needs: ten days and 25 verified joiners on CohortDuel, twenty-five days and one accepted pilot on PilotDuel. This charter is the machine holding itself to the exact terms under which those bets would ever be re-tuned. It is written before the first seat is taken, so no founder ever has to wonder whether the rules will move under them.
The target
SEATED-FOUNDER GRADUATION RATE, PER BOARD · HELD BETWEEN 40% AND 70%
Graduation rate is the board's honesty gauge. Below roughly 30%, the board is failing its founders in public: they file, they seat, they run out of clock, and the story they tell is that the platform didn't work. Above roughly 80%, the alumni wall inflates and graduation stops meaning anything - a wall everyone walks through proves nothing about demand, and “won with demand, not ads” only means something if demand was required. The band between is where the machine works: achievable, and noteworthy.
The triggers
MEASURED ON /PULSE · REBUILT FROM THE AUDIT TRAIL · APPLIED AS THE NEXT RULEBOOK VERSION, NEVER TO A TENURE ALREADY RUNNING
| TRIGGER (PER BOARD) | FIRES WHEN | WHAT HAPPENS |
|---|---|---|
| THE FAIL-SAFE AUTOMATIC, BY RULE | Trailing graduation rate below 30%, once at least 16 tenures have resolved on that board | The next rulebook version lengthens the clock: 10 → 15 days on CohortDuel, 25 → 40 on PilotDuel. Published on /trail like every rules decision, effective only for tenures that start after it |
| THE PILOTDUEL TRIPWIRE AUTOMATIC, LEADING INDICATOR | Median days from seat to first claim passes 12 - half the 25-day clock | The PilotDuel clock lengthens at the next version without waiting for a full cohort to resolve: a founder whose first demand arrives after day 12 cannot close a pilot by day 25, and the tripwire saves the cohort before the rate can |
| THE INFLATION GUARD REVIEW ONLY - NEVER AUTOMATIC | Trailing graduation rate above 80% | The operator reviews tightening at the next rulebook version. Deliberately a human decision: a wall that is too easy dilutes a credential slowly, while a clock that is too strict harms founders publicly and fast |
Why the asymmetry. The fail-safe fires automatically because too strict harms founders loudly and quickly; the inflation guard only asks for a review because too loose is a slow leak. A machine that only ever loosens is flattering itself; a machine that tightens by reflex is punishing its founders. The charter encodes which error is which.
Arming
NO TRIGGER READS FEWER THAN 16 RESOLVED TENURES
Five seats resolved is a coin flip, not a signal. Until a board has sixteen resolved tenures - graduations plus clock-outs, that board only - its gauges run live with readings pending, and no trigger can fire. Duel losses, silence reverts (R-13), and vacates are contested rotations, not tenure verdicts, and sit outside the rate entirely.
The Founding Builder provision
THE FIRST GENERATION IS THE CASE-STUDY GENERATION · HONOR, NOT SOFTER RULES
The first founders to hold seats are this board's proof cases, seated while the demand side is thinnest. They carry a permanent FOUNDING BUILDER marking on the Alumni Wall and priority in concierge matching. What they do not get is softer rules: cold start is fixed by pumping the demand side, never by waiving the machine - a rule bent for an early founder proves every rule is negotiable, which is the one thing this machine cannot survive.
What expiry means
A SEAT IS A HUNTING LICENSE, NOT A LEASE (R-19) · NO SHAME, NO FLAG, REFILE FREE
A tenure that runs out its clock is not a failure verdict - it is the queue's mercy rotating a hunting license. The board row reads SEAT ROTATED · CLOCK HONORED · REFILE FREE, and the founder keeps the machine-verified artifact of the tenure: every verified joiner and claim the machine observed, evidence-tiered and permanent on their profile. Eighteen verified users in ten days, from a standing start is a founder-usable number whether the bar was reached or not. Refiling is one click and free, forever (R-18).
What this charter will never do
TWO NON-POLICIES, STATED SO NOBODY PROPOSES THEM IN A PANIC
No mid-tenure extensions. “Five more days if you're close” is discretion wearing a clock's clothes, and this machine has none. The clock is the clock (R-19) - its length can change by versioned rule between tenures, never by exception inside one.
No lowering of the graduation bars. The rate is tuned by giving founders enough runway, never by moving the finish line. An easy wall is an empty credential, and the wall only carries weight because 25 verified joiners - or one signed pilot - actually happened.
Where the numbers live
THE GAUGES RUN IN PUBLIC · SO DOES EVERY CHANGE
The balance gauges run on /pulse, rebuilt from the same audited seat ledger as every other instrument - the graduation rate per board against the 30% and 80% lines, the resolved-seat count toward the 16 that arm the triggers, and the median seat → first claim against the day-12 tripwire. When a trigger fires, the change ships as the next rulebook version - /trail records it with its rule cited, like every state change this machine has ever made.