Reconstruction

After this lesson, the reader should understand:

• 01State the extractability thesis, and the claim it deliberately does not make.
• 02Explain why several rule sets can be equally consistent with the same behaviour.
• 03Express a reconstructed boundary with its uncertainty attached.

Two rules, one behaviour

A reconstruction yields "escalate above $48,000". An equally consistent reading is "escalate when the approver is not a director". How is the ambiguity resolved?

Not by the data alone. If directors happen to hold the higher approval limits, the two rules produce identical behaviour across every record in the log^[1], and no further volume of the same evidence will separate them. This is the non-identifiability problem familiar from inverse reinforcement learning^[2]: behaviour under-determines the rule that generated it. The discipline's answer is not to silently pick the more plausible reading. It is to surface both, name the confound explicitly, and route the question back to the institution — which knows which of the two it meant, and is the only party entitled to say. The strongest objection follows directly: if behaviour cannot pin down the rule, why reconstruct at all? Because under-determination is rarely total, and even when it is, the reconstruction has done real work. Real logs usually contain a few cases that separate the readings — a director approving an order above the line, an associate escalating one well below it — and a single such case can retire a hypothesis that thousands of ordinary cases could not. Where no such case exists, the reconstruction has still narrowed an open question ('how does procurement actually decide?') to a closed one with two named answers. That is a far better question to put to an institution than the one it started with. Philosophers of science have long argued that evidence never tests a hypothesis in isolation^[3]; the practical response has never been to stop testing, but to make the competing hypotheses explicit. The harder limit is in the last step. Routing the question back assumes the institution knows which rule it meant. Often it does not. Practice can emerge without anyone ever intending either reading — the approval pattern formed because directors happened to hold higher limits, and nobody chose whether seniority or amount was the real test. In that case asking 'which did you mean?' has no true answer, and the honest question is a different one: 'which do you now choose to adopt?' That is no longer discovery. It is a policy decision, and it needs an author with the authority to make it.

• What should a reconstruction do when two readings fit the evidence equally well?
• Is a reconstruction an accusation, a hypothesis, or a finding?
• Who owns a rule that was inferred rather than authored?
• How much confidence should a reconstruction carry before anyone acts on it?
• When an institution cannot say which of two readings it meant, is choosing one an act of interpretation or of legislation?

No responses yet. Be the first to answer one of the questions above.

Reconstruct a boundary, then attack your own result.

1. 01Infer a boundary from a real decision log.
2. 02Write it out as an explicit rule.
3. 03Propose a second rule that fits the same data equally well.
4. 04Identify the evidence that would separate the two.
5. 05State what you would require before presenting the first as a finding.
6. 06Search the log for any case that already separates the two rules, and say what it shows.