docs: the objection register

[EvanWinget]

What changed

One new document, docs/objection-register.md, plus its entry in docs/README.md. Nothing normative changes: no spec, no implementation, no vectors.

The register lists every objection on the public record to a Lisp VM for Bitcoin, or to the conditions architecture in particular, with BitLisp's answer, a status, and where the answer lives (spec section, divergence row, decision number, vector file). Where no answer exists it says so and names the phase that owes one.

• Section 1, O1 to O10: objections to any expressive VM (reviewability pricing, audit burden, witness-supplied code, static typing, ZK, termination, macros, names, debuggability, activation).
• Section 2, O11 to O22: objections to the CLVM-derived VM (serialization, peak memory, evaluation order, surplus arguments, softfork semantics, the CLVM defect history, tag domain separation, numbers, BLS, costing, budget commitment, bllsh divergences).
• Section 3, O23 to O36: objections to the conditions architecture (novel validator, the bundle model, positional identity, evidence contamination, the designers' lean, narrow margins, coin set vs UTXO, direct assertions, ephemeral coins, tokens, cultural headwind, self identity, thin aggregation record, what has been built).
• Section 4: the OPEN and PARTIAL rows grouped by owing phase. This must agree with the execution plan and is the objection-side view of it.
• Section 5: four objections no thread has raised, pre-registered so the essay states them first.

Sources and policy

Sourced from the nine public threads read in full on 2026-08-16 (evaluation doc sections 8.4 and 8.5) and the two exchanges the evaluation doc already records in section 6, cited only as that document records them. Attributions are to public posts. No private correspondence is quoted or newly characterized.

Why now

The execution plan's Phase 5 item 4 (track objections and revise the evaluation doc against them) has no artifact until hostile review starts. This register is that artifact written ahead of time: the essay's skeleton, and the checklist for the direct exchange with the introspection architecture's author. A conceded objection stated first by the project is worth more than one stated first by a reviewer.

Review guide

One commit. Read docs/objection-register.md top to bottom. The things worth checking:

1. Every "where" cell points at something that exists on main (or on PR 60 for C24 and decision 28, see merge order below). Spot-check a few against docs/vm-record.md, docs/condition-record.md, spec/VALIDATION.md, and docs/bitcoin-script-successor-evaluation.md.
2. Every ANSWERED status has a vector or invariant named. If a row claims ANSWERED with only prose behind it, it should be RECORDED.
3. Section 4 against docs/execution-plan.md Phase 4 and Phase 5 bullets: nothing owed here that the plan does not carry, and nothing the plan carries that the register omits. If the plan needs a skeleton edit to match, that goes direct to main separately per the plan-update rule.
4. The CONCEDED rows (O4, O10, O25, O26, O31, O33) are the ones the essay will publish as misses. Confirm each is conceded in the evaluation doc already and not newly here.

Verify: ci/lint/lint.sh (prose, whitespace, codespell). No tests are affected.

Merge order

References to C24 and decision 28 (rows O34 and the section 3 preamble) resolve once PR 60 (identity-assert) merges. Merge this after PR 60, or read those two references against that branch.