# Memoization — Shared Evidence, Computed Once

> **Law:** asking never writes, so within one response every structural read is
> pure. A memo may skip a probe. It may never change what inference computes.

**Why.** One question is read by up to eight mechanisms, and the same analysis
(the consensus climb, the weave, a resonance query) must not be paid eight
times. Nor may whoever asked first be billed for everyone.

## `Precomputed` — one response, one container (`mind/pipeline-mechanism.ts`)

`think` creates it before any mechanism runs. It is the only place a response's
shared evidence lives.

**Eager**, populated before any `floor` or `run`:

- `rec`, the recognition;
- `computed`, every mechanism's `parse` spans;
- `guide`, the query gist;
- `k = 2·recallQueryK`.

**Lazy**, cached by promise, so the first caller starts the work and every later
caller awaits it:

- `attention()`, the consensus climb;
- `weave()`;
- `resonance()`, the single top-`k` ANN query;
- `frames()`;
- `spanShapedOf(anchor)` / `spanShapedAll()`;
- the window identities `queryWindows`, `queryResolved` and `windowsOf`;
- `reachMemo`.

A mechanism that never asks pays nothing, and two that ask the same question pay
once. A `floor` checks `worthRunning` before it first touches an expensive
analysis (`mechanism-market.md`). Each shared analysis is charged to its own
meter phase through `Precomputed.shared`, never to whichever mechanism touched
it first (`meter.md`).

## Mind memos — `beginResponse` → `endResponse` (`mind/mind.ts`)

`respond` takes fresh maps. `respondTurn` reuses the conversation's maps, which
are content-keyed across turns.

| Memo                                                   | Key                   | Lifetime                                       |
| ------------------------------------------------------ | --------------------- | ---------------------------------------------- |
| `perceiveMemo`                                         | bytes + boundary set  | response or conversation                       |
| `recogniseMemo`, `climbMemo`                           | bytes                 | response or conversation                       |
| `canonMemo`                                            | bytes                 | response, when a `canon` is set                |
| `_resolvedSubtrees`                                    | tree node (`WeakMap`) | response or conversation                       |
| `_edgeChoice` (the pick memo), `_edgeAsked`            | node / question       | response; cleared at the end                   |
| `sharedReachMemo`, probes, derivations (`traverse.ts`) | node / form           | on write, or whole at its budget (`caches.md`) |

`foldTree` takes the `_resolvedSubtrees` fast path only when no visitor is
passed. A walk that emits sites always descends in full, and the cache only
elides store probes.

## The trace boundary

A traced response must emit every step and still give the same answer.

- **Bypassed under trace:** the pick memo (`guidedNext`), `sharedReachMemo` and
  the derivation memos. Each is fresh or absent, and `chooseNext` recomputes the
  same pick from the store and the question.
- **Always consulted:** `perceiveMemo`, `recogniseMemo` and `climbMemo`.
  Bypassing `recogniseMemo` re-runs recognition over a warm subtree cache and
  emits 5 sites instead of 31: a change in the answer, not just a slowdown.
- **The pick memo is cleared when the climb publishes its points.** A pick made
  earlier read less evidence, and keeping it made traced and untraced responses
  disagree.
- **The exact tier's pick made before the climb is provisional.** It serves
  until the points arrive and is then made once more, because a co-instance read
  from them may name a step or outrank what witnessing named. Keeping it made
  traced and untraced responses disagree (`test/157`).

## Adding a shared analysis

Add one lazy method to `Precomputed`, never a memo map elsewhere, and guard it
behind `worthRunning` in `floor`.

## Pins

- `test/42` — recognition is idempotent under trace: same site count, same
  cached object.
- `test/155.4` — traced and untraced responses agree once the climb publishes
  its points.