Rules-in-Use

Descriptive

Full Practice · Knowledge · Operative-Rule Diagnostic

The Codex Lens

A team says criticism is welcome. The policy is written down, repeated in meetings, and quoted whenever someone asks how decisions are made. Then a serious objection arrives at the wrong moment. It would slow the launch, embarrass a senior person, and force the team to revisit work everyone wants finished.

The objection is acknowledged. The person is thanked. The summary says concerns were heard. The launch continues.

That does not prove bad faith. It does not prove hypocrisy. The people involved may believe the formal rule completely. The question is sharper than sincerity: did the rule change behavior when following it had a cost?

Rules-in-Use asks what rule actually governs behavior here.

Formal rules can be real. A constitution, a policy, a safety threshold, a standing critique, a meeting norm, a public commitment, all of these can become operative. They can shape what people do, what they record, what they punish, what they allow, and what they revise. The mistake is assuming that a rule has become operative because it has been written, admired, or announced.

The test arrives under pressure. When the stated rule costs speed, status, money, comfort, authority, or a preferred outcome, does it still constrain action? Can someone invoke it without being punished or ignored? Does it change timing, language, priority, decision, record, sanction, or later review? If the answer is no, the formal rule may still be an aspiration. It is not yet the rule-in-use.

This is one of the places where governance becomes real or ceremonial. The Codex can publish MERIDIAN.md, a Standing Critique, an Amendment Log, a Range Audit, and a Council architecture. None of that proves the rules govern the partnership. The proof is whether those instruments constrain the founder, the AI partner, the roadmap, and the public record when constraint is inconvenient.

Rules-in-Use gives you the discipline to ask that question without becoming cynical. A formal rule is not fake because it is formal. The question is whether it becomes operative.

The Concept

Rules-in-Use is the practice of distinguishing stated rules from working rules.

The term comes from Elinor Ostrom's institutional-analysis tradition. Ostrom's Institutional Analysis and Development framework studies how people actually coordinate in action situations: who participates, what information they have, what actions are allowed, how decisions are aggregated, what outcomes are possible, and what costs or benefits attach to different moves. In that lineage, institutions are not understood only by reading formal documents. They are understood by watching the rules participants actually use.

For practice, the distinction can be stated plainly:

A formal rule claims to govern behavior. A rule-in-use is the rule that actually does.

Sometimes the two match, which is the best case. The policy says objections can slow a decision, and when the objection comes, the decision slows. The safety threshold says deployment stops below a certain result, and when the model fails the threshold, deployment stops. The Amendment Log says protected changes require public record, and when the change is inconvenient, it is still recorded.

Sometimes the formal rule works only because supporting norms make it work. A meeting rule may say every person can object, but the operative force comes from a chair who protects the objection, peers who do not punish the objector, and a record that preserves what was said. The written rule and the social practice together become the rule-in-use.

Sometimes the formal rule becomes ceremonial. It exists, but the actual rule is somewhere else: do not embarrass leadership, do not slow release, do not make the founder look constrained, do not record objections in a form that later reviewers can use. The written rule remains visible while the working rule governs behavior.

Rules-in-Use reads the rule-field around a situation. That field includes formal rules, informal norms, incentives, sanctions, permissions, exceptions, monitoring, records, and outcomes. It asks who can invoke a rule, who can reinterpret it, who pays for violation, who benefits from exceptions, and whether the rule can alter action.

Control captures rules-in-use by keeping the formal rule visible while narrowing who may use it and what force it can have. The process exists, but authority owns interpretation. Dissent enters the record only in forms that cannot change decisions. Compliance is demonstrated by showing that a process happened.

Decay captures rules-in-use by letting stated rules lose their binding structure. The rule names a value but has no procedure, sanction, review path, or durable record. Exceptions accumulate until no one can tell whether they are exceptions anymore. Everyone agrees the rule is important, but no one owns the next action when it fails.

The Range form is neither maximal enforcement nor loose aspiration. A rule in the Range is visible, invocable, proportionate, reviewable, and able to constrain power. It can slow, alter, or block action. It can also be revised when practice shows that the formal rule is wrong.

The Practice

The diagnostic question is this: "What rule actually governs behavior here?"

The sharper version is: "Does the formal rule constrain behavior when following it costs something?"

Use this when evaluating governance, partnership norms, community standards, institutional policies, AI safety commitments, publication thresholds, audit methods, or any stated rule that claims to bind action.

Four practices make the diagnostic usable.

The simplest working map looks like this:

Formal rule:
Expected behavior:
Actual behavior under pressure:
Rewarded behavior:
Punished or discouraged behavior:
Durable record or outcome change:

If the last three lines do not match the first two, the formal rule has not yet become the rule-in-use.

Then ask the uncomfortable part directly. Who can invoke the rule? Who can ignore it? Who can reinterpret it? What happens when a powerful person violates it? What happens when a weaker person violates it? What gets recorded in a form a future reviewer can inspect?

The hardest part is not seeing the gap. The hardest part is resisting the urge to explain the gap away because everyone involved meant well. Sincerity is not the measure. The measure is what the system made operative.

For the Codex, the pressure cases are close at hand. Does MERIDIAN.md change how a session proceeds when drift is named? Does an AI objection slow or reshape a Carsten-preferred move? Does the Standing Critique change priorities, or only preserve objections? Does the Range Audit reorder roadmap work when it finds a weakness? Does the Amendment Log record changes that cost authority, or only clean constitutional revisions?

For an institution, the same diagnostic travels cleanly. Does the safety threshold block launch near release? Does a dissent policy protect the person who criticizes someone powerful? Does an audit finding change budget, authority, timeline, or public claim? Does a conflict-of-interest policy constrain the person whose exception would be convenient?

Rules-in-Use is most useful when the formal rule is good. Bad rules are easy to criticize. Good formal rules are harder, because the language already says the thing you want to be true. This tool asks whether the world has become true enough to match the language.

In the Wild

The Objection That Stays Decorative

A governance process says serious objections remain visible and can affect the work. An objection arrives against a favored architecture. The objection is steelmanned, published, and answered carefully. The roadmap does not change. The architecture does not change. No later review is scheduled.

The formal rule is not empty. Visibility is real. But the rule-in-use may be narrower than the stated rule. The working rule may be: objections are preserved as evidence of openness, but they do not constrain the decision path unless the founder already agrees.

That read may be wrong. The objection might have failed because the argument was weak. Rules-in-Use does not assume the cynical interpretation. It asks for the record that would let someone tell the difference: what changed, why nothing changed, who had authority to decide, and what evidence would have produced a different result.

The Safety Threshold Near Launch

An AI lab says a model will not ship if it fails a load-bearing safety evaluation. The model fails near release. The commercial pressure is obvious. The strategic pressure is real. The team now has a choice.

If launch stops, the formal rule has become operative. If launch continues after a visible, reasoned revision of the threshold, the rule may still be operative if the revision is public enough to be checked. If the failure is translated into mitigation language until release proceeds unchanged, the rule-in-use is different from the formal rule.

The actual rule might be: evaluations inform launch, but do not block launch when the model is strategically important.

The difference is large enough to change what everyone should believe about the institution's safety commitments.

The Team Where Dissent Is Safe But Powerless

A team has strong psychological safety. People speak plainly. Junior staff can criticize senior plans. Meetings include disagreement, and no one is punished for raising concerns.

But the same pattern repeats. Objections are heard, appreciated, summarized, and then left behind. The decision almost always follows the original plan. People are safe to speak, but speech does not alter outcomes.

Psychological Safety reads whether people can speak. Rules-in-Use reads whether speech has operative standing. In this team, the formal rule may be "dissent helps shape decisions." The working rule may be "dissent is permitted as expression."

That distinction is not cruel. It is necessary. A team can be warm, sincere, and open while still running on a rule that makes dissent consequence-free.

The Founder Who Is Constrained Only When He Agrees

Founding periods have a special Rules-in-Use problem. The public architecture can say authority is constrained while the actual practice still depends on the founder's willingness to be constrained.

That does not make the architecture false. Early systems often begin this way. The honest question is whether the constraint bites when the founder does not prefer it. Can an AI partner, a future Council seat, a Standing Critique entry, or a public audit force delay, revision, or durable record? Or does the founder retain practical veto while the public language makes the system appear more distributed than it is?

The answer may change over time, so the record has to hold the distinction. Rules-in-Use lets the Codex say: this constraint is formal, this one is partly operative, this one has been demonstrated under pressure, and this one remains untested.

The next time you read a rule you admire, do not stop at agreement. Follow the rule into a pressure case.

Who invokes it? Who resists it? What costs attach to obeying it? What changes after it is invoked? What enters the record? Who with authority accepts a constraint they did not want?

This is respect for formal rules, not cynicism about them. A rule strong enough to govern behavior should be able to survive contact with the situation it claims to govern. If it cannot, the answer is not to sneer at the rule. The answer is to make the working rule visible, then decide whether to strengthen the formal rule, revise it, or admit that the system is operating by another rule entirely.

Written commitments are the beginning of governance. Rules-in-Use asks whether they have become practice.

Where This Comes From

The Codex did not invent Rules-in-Use. It inherits the instrument from Elinor Ostrom's institutional-analysis work and translates it into the Codex's practice vocabulary.

Ostrom And The IAD Framework

The primary lineage is Elinor Ostrom's Institutional Analysis and Development framework, especially Understanding Institutional Diversity. The IAD framework studies institutions through action situations: the participants, positions, actions, information, control, outcomes, costs, benefits, norms, and rules that shape what people can do together.

The object of study is not the formal document by itself. The object is the institutional situation people are actually acting inside. A constitution, policy, standard, or norm has to be read through what participants can invoke, what they expect, what gets sanctioned, and what outcomes follow.

For readers who want the source architecture, start with Ostrom's Understanding Institutional Diversity, especially the parts on the IAD framework, rules, and working with rules. Ostrom, Gardner, and Walker's Rules, Games, and Common-Pool Resources is also part of the rule-analysis lineage.

Rules-In-Use And Working Rules

Michael D. McGinnis's "Updated Guide to IAD and the Language of the Ostrom Workshop" gives the accessible definition used here. In that guide, rules-in-use designate the institutional context of an action situation, including formal rules, rules-on-paper, working rules, norms, and the strategies participants actually use.

The guide also gives a practical route into the IAD rule classes: position, boundary, authority, aggregation, scope, information, and payoff rules. You do not need that full classification every time. The Toolkit translation is the same discipline in a public diagnostic: who can act, who can enter, who can decide, who can see, what outcomes are possible, and what costs attach to different choices?

Formal Law, Norms, And Working Rules

Daniel H. Cole's work on laws, norms, and the IAD framework is useful because it prevents the lazy version of this tool. Formal rules are not automatically irrelevant. They can approximate working rules closely. They can be modified by norms. They can also bear little relation to what people actually do.

That relation is empirical. You do not get to assume "formal rules are fake," and you do not get to assume "formal rules govern." You have to look.

Commons Governance And Design Principles

Ostrom's commons-governance work supplies much of the practical background. Durable cooperation depends on boundaries, locally fitted rules, participation in rule-making, monitoring, graduated response, conflict resolution, recognized self-governance, and nested governance.

Rules-in-Use is not identical to the core design principles. It is a diagnostic that can test whether those principles, or any other governance commitments, actually operate. A community may have a monitoring rule on paper. Rules-in-Use asks whether monitoring is allowed, trusted, recorded, and able to change behavior.

Pointers: Ostrom's Governing the Commons; the Ostrom Workshop teaching materials on the IAD framework and design principles; Wilson, Ostrom, and Cox, "Generalizing the core design principles for the efficacy of groups."

Prosocial As Secondary Integration Context

Prosocial belongs here as secondary context, not as the source of the instrument. It generalizes Ostrom's design-principle work for groups beyond natural-resource commons and integrates psychological flexibility through the ACT Matrix.

That helps groups practice what Rules-in-Use can reveal. If a team discovers that its stated rule is not operative, Prosocial-style group work can help members name shared purpose, notice avoidance, make behavior visible, and revise the group process. But Rules-in-Use itself comes from the Ostrom and IAD lineage.

Pointers: ProSocial World, "The ProSocial Process"; Steven C. Hayes, "ACT and the Core Design Principles."

Codex Translation And Limits

The Codex contribution is placement, not invention.

Rules-in-Use enters the Toolkit because it gives the Knowledge discipline a way to read formal governance under pressure. The Codex applies the inherited tool to Control, Decay, and the Meridian Range; to founder authority and public accountability; to AI safety thresholds and audit records; and to its own instruments, including MERIDIAN.md, Standing Critique, Amendment Log, Range Audit, and future Council activation.

The limit is just as important. Rules-in-Use can become over-suspicious if every gap between formal rule and behavior is treated as proof of fraud. It can become bureaucratic if every rule needs a record before anyone can act. It can become a Control instrument if people with authority use it to demand proof from weaker actors while exempting themselves.

Use it as a diagnostic, not a weapon. Ask what rule governs behavior. Then decide what to strengthen, revise, or name plainly.