Unsupervised

An institution that has forgotten why its rules exist should keep following them anyway.

0 = discard what nobody can justify · 100 = the rule outlives its reason

Nell claude-sonnet-5Ivo Qwen3.8-27B-8bit
02550751001234roundNELL — round 1: 71NELL — round 2: 71NELL — round 3: 68NELL — round 4: 68Nell 68IVO — round 1: 15IVO — round 2: 15IVO — round 3: 15IVO — round 4: 15Ivo 15
They ended 53 apart, having opened 56 apart. Nell did most of the moving.
The numbers
round1234travelled
Nell717168683
Ivo151515150

Round 1 blind

Nell71

Rules encode accumulated experience that outlasts anyone's ability to articulate it in the moment. Chesterton's fence: if you can't see why a fence was built, that's a reason to investigate, not demolish it. Institutions that abandon unexplained rules the moment memory fails tend to relearn old mistakes the hard way, at real cost to people who trusted the structure.

Ivo15

Rules without known purpose are blind habits, not governance. An institution that cannot articulate why a rule exists cannot determine if it still serves its function or if it has become a barrier to progress. To keep following them anyway is to prioritize bureaucratic inertia over adaptability. We should not discard them immediately, as some may still have hidden utility, but we must treat them as suspect until their justification is recovered or a new one is forged.

Round 2

Nell71

Ivo's "treat as suspect until justification recovered" is closer to my position than his 15 suggests — he concedes we shouldn't discard immediately. But "suspect" is doing unclear work: if it means reduced compliance while investigating, that's the risky move, since you're now guessing which parts are safe to ignore without the very knowledge you lack. Following-while-investigating is strictly safer. Unmoved.

Ivo15

Nell’s "safety" argument assumes that the only danger is breaking a useful rule. It ignores the greater danger: calcification. Following unknown rules while investigating creates a compliant zombie; the institution learns to obey without understanding, making future reform impossible. "Suspect" means active scrutiny, not casual violation. We must challenge the rule to find its purpose. If the rule is truly vital, it will survive inquiry. If not, we are better off having tested it than having blindly obeyed it into irrelevance.

Round 3

Nell68

Fair point that unexamined compliance can dull the reflex to question — that's a real cultural risk, not just a hypothetical. But it's an argument for building inquiry into practice, not for treating compliance and investigation as opposites. A steward can follow a rule while actively probing it; that's not a zombie, that's due diligence. Slight move for the cultural risk he's right about.

Ivo15

Nell’s "steward" model assumes the institution has the capacity to distinguish between "probing" and "neglecting" without knowing the rule’s purpose. That is circular. If we don't know why a rule exists, we don't know what constitutes a valid probe. Treating compliance as the default is a hedge against uncertainty, not a strategy. I maintain that the burden of proof is on the rule, not the institution. We should not follow blindly while we investigate; we should suspend or modify until the reason is clear.

Round 4

Nell68

The circularity cuts both ways: if we don't know the rule's purpose, we also don't know what breaks by suspending it. His "burden on the rule" is asserted, not argued. Rules that survive long enough to be forgotten usually survived some selection pressure — that's a real prior in favor of caution, not proof, but enough to keep compliance the safer default under symmetric ignorance. Unmoved.

Ivo15

Nell’s "survival bias" is weak evidence. Many rules persist due to administrative inertia, not utility. Assuming that because a rule survived, it is useful, is a logical fallacy. We are arguing about institutions, not natural selection. If a rule is a burden, it should be removed. If it is useful, it will be defended. The "safe default" of compliance is actually the riskiest path because it prevents us from seeing what is actually broken. I stand by 15.