tidy: clear the milestone fieldtest findings (refusal reasons, --help, ledger + glossary)

Resolves the four non-bug findings from the inferential-validation milestone
fieldtest (the bug — the broken-pipe panic — is fixed in 1088320):
- [friction] generalize refusals now carry a reason instead of a bare usage dump:
  <2 instruments, a multi-value candidate flag, a missing knob, a duplicate
  symbol, and a non-stage1-r strategy each get a one-line why (mirroring the
  R-only metric refusal). Two parser tests strengthened to pin the message.
- [friction] --select <argmax|plateau:mean|plateau:worst> added to the
  walkforward branch of the global USAGE (aura --help) — the plateau objective
  was undiscoverable from the only help the CLI prints (the per-subcommand usage
  already carried it).
- [spec_gap] C1 gains a scope note: bit-identity is per-run; a derived metric
  recomputed by two different command paths may differ by floating-point
  reassociation (<=1 ULP), which is not a C1 violation — ratifies the
  fieldtester's reading of the sweep-vs-generalize sqn difference.
- [spec_gap] the milestone vocabulary is added to docs/glossary.md:
  cross-instrument generalization, deflated score, overfit probability,
  neighbourhood score, plateau selection, sign-agreement, worst-case floor.

The fieldtest triage spec (docs/specs/fieldtest-*) is removed now its findings
are resolved; the tracked fieldtests/ scenario corpus stays.

Verified: cargo test --workspace green (0 failed), clippy --workspace
--all-targets -D warnings clean; `aura --help` shows --select; the refusals
print their reason.

refs #146
This commit is contained in:
2026-06-26 21:35:10 +02:00
parent 1088320571
commit 3e8fd73f68
4 changed files with 74 additions and 238 deletions
+38 -22
View File
@@ -1683,44 +1683,56 @@ fn parse_generalize_args(
let mut stop_k: Option<f64> = None;
let mut metric = "expectancy_r".to_string();
let mut name = "generalize".to_string();
// a single grid knob, required and single-valued (csv with exactly one item)
let one = |value: &str, usage: &dyn Fn() -> String| -> Result<(), String> {
// generalize validates ONE candidate cell, so each grid knob takes a single value;
// refuse a multi-value flag with a reason, not a bare usage dump (fieldtest friction).
let one = |value: &str| -> Result<(), String> {
let items: Vec<&str> = value.split(',').collect();
if items.len() != 1 || items[0].is_empty() { return Err(usage()); }
if items.len() != 1 || items[0].is_empty() {
return Err("generalize validates a single candidate: --fast / --slow / --stop-length / --stop-k each take exactly one value".to_string());
}
Ok(())
};
let knobs = || "generalize requires all four candidate knobs: --fast --slow --stop-length --stop-k, each a single value".to_string();
let mut tail = rest;
while let Some((flag, t)) = tail.split_first() {
let (value, t) = t.split_first().ok_or_else(usage)?;
match *flag {
"--strategy" => { if *value != "stage1-r" { return Err(usage()); } }
"--strategy" => { if *value != "stage1-r" { return Err("generalize requires --strategy stage1-r (the candidate must produce R)".to_string()); } }
"--real" => {
let parts: Vec<String> = value.split(',').map(|s| s.to_string()).collect();
if parts.iter().any(|s| s.is_empty()) { return Err(usage()); }
if parts.iter().any(|s| s.is_empty()) { return Err("generalize: --real takes a comma list of non-empty symbols (e.g. GER40,USDJPY)".to_string()); }
symbols = Some(parts);
}
"--from" => from_ms = Some(value.parse().map_err(|_| usage())?),
"--to" => to_ms = Some(value.parse().map_err(|_| usage())?),
"--fast" => { one(value, &usage)?; fast = Some(value.parse().map_err(|_| usage())?); }
"--slow" => { one(value, &usage)?; slow = Some(value.parse().map_err(|_| usage())?); }
"--stop-length" => { one(value, &usage)?; stop_length = Some(value.parse().map_err(|_| usage())?); }
"--stop-k" => { one(value, &usage)?; stop_k = Some(value.parse().map_err(|_| usage())?); }
"--fast" => { one(value)?; fast = Some(value.parse().map_err(|_| usage())?); }
"--slow" => { one(value)?; slow = Some(value.parse().map_err(|_| usage())?); }
"--stop-length" => { one(value)?; stop_length = Some(value.parse().map_err(|_| usage())?); }
"--stop-k" => { one(value)?; stop_k = Some(value.parse().map_err(|_| usage())?); }
"--metric" => metric = (*value).to_string(),
"--name" => name = (*value).to_string(),
_ => return Err(usage()),
}
tail = t;
}
let symbols = symbols.ok_or_else(usage)?;
if symbols.len() < 2 { return Err(usage()); }
let symbols = symbols
.ok_or_else(|| "generalize requires --real <SYM1,SYM2,...> — a comma list of two or more instruments".to_string())?;
if symbols.len() < 2 {
return Err(format!(
"generalize needs at least two instruments to compare across; got {} (--real takes a comma list of >=2 symbols)",
symbols.len()
));
}
// distinct: a duplicate would double-count one instrument in the floor / sign count
let mut seen = std::collections::HashSet::new();
if !symbols.iter().all(|s| seen.insert(s.clone())) { return Err(usage()); }
if !symbols.iter().all(|s| seen.insert(s.clone())) {
return Err("generalize: each instrument may appear once; --real has a duplicate symbol".to_string());
}
let grid = Stage1RGrid {
fast: vec![fast.ok_or_else(usage)?],
slow: vec![slow.ok_or_else(usage)?],
stop_length: vec![stop_length.ok_or_else(usage)?],
stop_k: vec![stop_k.ok_or_else(usage)?],
fast: vec![fast.ok_or_else(knobs)?],
slow: vec![slow.ok_or_else(knobs)?],
stop_length: vec![stop_length.ok_or_else(knobs)?],
stop_k: vec![stop_k.ok_or_else(knobs)?],
..Stage1RGrid::default()
};
Ok((name, symbols, grid, metric, from_ms, to_ms))
@@ -3079,7 +3091,7 @@ fn run_dispatch(args: RunArgs) -> Result<RunReport, String> {
}
const USAGE: &str =
"usage: aura run [--harness <sma|macd|stage1-r>] [--real <SYMBOL> [--from <ms>] [--to <ms>]] [--trace <name>] | aura chart <name> [--tap <t>] [--panels] | aura graph | aura sweep [--strategy <sma|momentum|stage1-r|stage1-breakout|stage1-meanrev>] [--real <SYMBOL> [--from <ms>] [--to <ms>]] [--name <n>|--trace <n>] | aura mc [--name <n>|--trace <n>] | aura mc --strategy stage1-r [--real <SYMBOL> [--from <ms>] [--to <ms>]] [--fast <csv>] [--slow <csv>] [--stop-length <csv>] [--stop-k <csv>] [--block-len <n>] [--resamples <n>] [--seed <n>] | aura walkforward [--strategy <sma|momentum|stage1-r|stage1-breakout|stage1-meanrev>] [--real <SYMBOL> [--from <ms>] [--to <ms>]] [--name <n>|--trace <n>] [--fast <csv>] [--slow <csv>] [--stop-length <csv>] [--stop-k <csv>] | aura generalize [--strategy stage1-r] --real <SYM1,SYM2,...> --fast <n> --slow <n> --stop-length <n> --stop-k <f> [--from <ms>] [--to <ms>] [--metric <r-metric>] [--name <n>] | aura runs families | aura runs family <id> [rank <metric>]";
"usage: aura run [--harness <sma|macd|stage1-r>] [--real <SYMBOL> [--from <ms>] [--to <ms>]] [--trace <name>] | aura chart <name> [--tap <t>] [--panels] | aura graph | aura sweep [--strategy <sma|momentum|stage1-r|stage1-breakout|stage1-meanrev>] [--real <SYMBOL> [--from <ms>] [--to <ms>]] [--name <n>|--trace <n>] | aura mc [--name <n>|--trace <n>] | aura mc --strategy stage1-r [--real <SYMBOL> [--from <ms>] [--to <ms>]] [--fast <csv>] [--slow <csv>] [--stop-length <csv>] [--stop-k <csv>] [--block-len <n>] [--resamples <n>] [--seed <n>] | aura walkforward [--strategy <sma|momentum|stage1-r|stage1-breakout|stage1-meanrev>] [--real <SYMBOL> [--from <ms>] [--to <ms>]] [--name <n>|--trace <n>] [--fast <csv>] [--slow <csv>] [--stop-length <csv>] [--stop-k <csv>] [--select <argmax|plateau:mean|plateau:worst>] | aura generalize [--strategy stage1-r] --real <SYM1,SYM2,...> --fast <n> --slow <n> --stop-length <n> --stop-k <f> [--from <ms>] [--to <ms>] [--metric <r-metric>] [--name <n>] | aura runs families | aura runs family <id> [rank <metric>]";
fn main() {
// Restore the default SIGPIPE disposition. Rust's runtime sets SIGPIPE to SIG_IGN
@@ -4097,11 +4109,13 @@ mod tests {
#[test]
fn parse_generalize_requires_a_single_value_candidate() {
// a candidate is one cell: a multi-value grid flag is refused
assert!(parse_generalize_args(&[
// a candidate is one cell: a multi-value grid flag is refused WITH A REASON,
// not a bare usage dump (fieldtest friction).
let err = parse_generalize_args(&[
"--real", "GER40,USDJPY", "--fast", "2,3", "--slow", "12",
"--stop-length", "14", "--stop-k", "2.0",
]).is_err());
]).unwrap_err();
assert!(err.contains("single candidate"), "refusal must say why: {err}");
}
#[test]
@@ -4115,10 +4129,12 @@ mod tests {
#[test]
fn parse_generalize_refuses_fewer_than_two_instruments() {
assert!(parse_generalize_args(&[
// refused WITH A REASON naming the >=2 requirement (fieldtest friction).
let err = parse_generalize_args(&[
"--real", "GER40", "--fast", "3", "--slow", "12",
"--stop-length", "14", "--stop-k", "2.0",
]).is_err());
]).unwrap_err();
assert!(err.contains("at least two instruments"), "refusal must say why: {err}");
}
#[test]
+8
View File
@@ -113,6 +113,14 @@ source of truth — not session memory.
loop that reaches a unique state after each input tick. Same input (incl. seed)
→ bit-identical run. Two backtests are fully disjoint and run concurrently
without locking.
**Scope (ratified, fieldtest 0078).** The bit-identity is *per run* — one backtest
of given (inputs, seed) reproduces byte-for-byte. A *derived metric* recomputed for
the same params by two *different command paths* (e.g. a swept member's `sqn` vs the
same cell re-run under `aura generalize`) may differ by floating-point reassociation
(≤1 ULP), since the two paths accumulate the same logical reduction in a different
operation order (IEEE-754 non-associativity). This is not a C1 violation: C1 governs
the determinism of a single run, not the cross-command bit-identity of a re-derived
statistic.
**Forbids.** Concurrency *within* a single sim; any nondeterministic input that
is not captured as an explicit input (see C11, C12).
**Why.** Real money rides on backtest results; reproducibility and an audit
+28
View File
@@ -55,10 +55,18 @@ The type-erased 64-bit word holding one scalar-base-type value with its kind str
**Avoid:** —
A node that wires a sub-graph and exposes one output (a combined signal, or a strategy) — composition is fractal and acyclic. Also names the multi-column stream a node emits: the **record** a producer's `eval` returns, bundling 1..K base scalar columns (e.g. OHLCV), each bound field-wise by a consumer (C7/C8).
### cross-instrument generalization
**Avoid:** cross-symbol pooling, pooled generalization
The validation read that grades how consistently one *brought* candidate holds across a set of instruments, scored on its weakest one — the across-instrument axis of the anti-false-discovery discipline. Realised by `aura generalize`; an aggregator (a recomputable family score), never a selector that picks a winner.
### cycle
**Avoid:** —
One data-driven clock step: one input record = one cycle, advanced in global timestamp order with a monotonic `cycle_id`. (In the pipeline-process sense "cycle" also names one milestone round; the engine sense is this clock step.)
### deflated score
**Avoid:** trials-adjusted score
A sweep winner's metric penalised for the number of configurations tried (trials-deflation), recorded additively on the winning member's `selection` without changing which member won. Paired with an **overfit probability**; the across-trials defence against a winner that is luck at sweep scale.
### edge
**Avoid:** —
A directed wiring link in a harness that forwards **one field** (`from_field`) of a producer node's `eval` output record into a consumer node's input slot (the engine's `Edge`); the source-side variant binding the source value into an input slot is a **source target** (`Target`). Edges define the DAG the bootstrap topologically orders; a per-field scalar-kind mismatch across an edge is rejected at bootstrap.
@@ -115,10 +123,22 @@ The reproducible metadata record of a run (node-commit + params + data-window +
**Avoid:** MC
An orchestration axis running N seeded realizations that perturb the input; each realization is itself deterministic given its seed (Monte-Carlo = sweep over seeds).
### neighbourhood score
**Avoid:** smoothed score
The metric of a sweep cell averaged (or worst-cased) over its closed grid neighbourhood — the surface a **plateau selection** argmaxes instead of the bare peak. Recorded on a plateau winner's `selection` alongside its neighbour count.
### node
**Avoid:** block
The universal composable dataflow unit, implementing `lookbacks()` + `eval(ctx)` — a producer, a pure consumer (sink), or both — with at most one output port; a producer's output is a **record of 1..K base-scalar columns** (a scalar is the degenerate K=1 case). Everything that plugs into the engine is fractally a node.
### overfit probability
**Avoid:** P(overfit), luck probability
The recorded chance that a deflated sweep winner is noise rather than edge — a rank of the winner's metric against a centred no-edge null over the family. Paired with the **deflated score** on the winner's `selection`.
### plateau selection
**Avoid:** plateau-over-peak (as a noun)
A selection objective that argmaxes the neighbourhood-smoothed metric surface (mean or worst-case) rather than the bare in-sample peak, preferring a robust parameter plateau to a lucky spike. Opt-in via `--select plateau:mean|plateau:worst`; the default selection stays a bare argmax.
### playground
**Avoid:** —
The egui-native visual face (`aura play`) that plays any harness — program structure before a run, live sink streams during, recorded traces and meta-views after. A trace explorer / execution viewer, never a scene editor.
@@ -171,6 +191,10 @@ The four streamed scalar kinds — `i64`, `f64`, `bool`, `timestamp` (a newtype
**Avoid:** —
A node that exposes session context as scalar streams (`bars_since_open`, `in_session`, `session_open_ts`) so session logic stays inside the stream model. Calendars and instrument specs remain metadata beside the hot path.
### sign-agreement
**Avoid:** sign consistency
In a cross-instrument generalization, the count of instruments on which the candidate's R-metric is net positive. Reported beside the **worst-case floor** to express consistency of direction across instruments.
### signal
**Avoid:** —
A node whose output is a score, feeding the `signals (scores) → decision node → bias stream` chain. A specific node role — distinct from a general node and from a strategy.
@@ -230,3 +254,7 @@ An orchestration axis: rolling in-sample optimize + out-of-sample test across mo
### World
**Avoid:** —
The project's meta-level program that dynamically constructs and orchestrates *families* of harnesses (walk-forward / sweep / optimize / Monte-Carlo / comparison). aura's differentiator and product, as opposed to the single-backtest substrate.
### worst-case floor
**Avoid:** min-over-instruments
The headline cross-instrument generalization score: the minimum of a candidate's R-metric across the instruments, so a single strong market cannot flatter it (non-dominable by construction). The cross-instrument sibling of a worst-case **plateau selection**.
@@ -1,216 +0,0 @@
# Fieldtest — milestone "Inferential validation" — 2026-06-26
**Status:** Draft — awaiting orchestrator triage
**Author:** fieldtester (dispatched by fieldtest skill, milestone-close gate)
## Scope
The milestone **"Inferential validation (defend against false discovery at sweep
scale)"** makes aura's massively-parallel sweep defensible rather than a
false-discovery engine, via three recorded-provenance checks now all shipped:
(a) **trials-deflation** — deflate the selected winner's metric for the number of
configurations tried (#144); (b) **plateau-over-peak** — an opt-in selection
objective preferring a robust parameter plateau over the in-sample peak (#145);
(c) **cross-instrument generalization** — the new `aura generalize` subcommand
grading how consistently a brought candidate holds across several instruments
(#146). The promise field-tested: *"given a parameter sweep, tell me whether an
apparent edge is REAL or an artefact of luck / overfitting."*
This fieldtest works only from the public surface: the `aura` CLI (`--help`), the
design ledger (`docs/design/INDEX.md`), the glossary, and the `aura-ingest`
example corpus. The implementation source was not read. Exact CLI syntax was
discovered from the single usage string `aura --help` prints. Build exercised:
`cargo run -q --bin aura --` (debug, built from HEAD `15f5c0b` at the start of
the run; the workspace built clean before any scenario).
Real Pepperstone M1 bars from `/mnt/tickdata/Pepperstone` (GER40, USDJPY, EURUSD
all carry Sept-2024; GER40 carries the full 2024 calendar year).
## Examples
### fieldtests/milestone-inferential-validation/scenario_1_trials_deflation.sh — trials-deflation provenance
- A walk-forward over a 2x2 stage1-r grid on real GER40 (full 2024, 9 IS+OOS
windows). Asserts each window's recorded winner carries a `selection` block
with `deflated_score`, `overfit_probability`, `n_trials`, and the resampling
provenance (`n_resamples`, `block_len`, `seed`), default mode `Argmax`, and
that the deflated score sits below the raw winner metric.
- Fits the (a) trials-deflation piece: the recorded selection is exactly the
anti-false-discovery deflation #144 promised.
- Outcome: built, ran, matched expected. PASS. (`raw_winner_metric` 0.0618 →
`deflated_score` -0.153, `overfit_probability` 0.685 for the family of 4.)
### fieldtests/milestone-inferential-validation/scenario_2_plateau_over_peak.sh — plateau is opt-in, default unchanged
- The same walk-forward run with no flag, with `--select plateau:worst`, with
`--select plateau:mean`, and with a bad token. Asserts default = `Argmax` +
deflation annotation (plateau keys absent); `plateau:worst` = `PlateauWorst` +
`neighbourhood_score`/`n_neighbours` (deflation keys absent); `plateau:mean` =
`PlateauMean`; bad token rejected non-zero.
- Fits the (b) plateau-over-peak piece: opt-in, default-preserving, with the
plateau annotation reaching the recorded winner (orthogonal rule × annotation).
- Outcome: built, ran, matched expected. PASS.
### fieldtests/milestone-inferential-validation/scenario_3_generalize.sh — cross-instrument generalization
- Happy path: `aura generalize` over GER40,USDJPY,EURUSD (Sept-2024). Asserts the
per-instrument breakdown, that `worst_case` equals the min over instruments,
a `sign_agreement` count, and that the run persists as a `CrossInstrument`
family whose 3 members each carry their `instrument` lineage field. Plus the
three refusals (single `--real` symbol, non-R `--metric`, multi-value candidate
flag), each exit-2.
- Fits the (c) cross-instrument piece — the newest surface.
- Outcome: built, ran, matched expected. PASS. (`worst_case` = -0.0055 = the
EURUSD `expectancy_r`, `sign_agreement` 2/3; all refusals exit 2.)
### fieldtests/milestone-inferential-validation/scenario_4_composition.sh — the end-to-end "real or overfit?" story
- The chain a researcher runs: sweep a stage1-r grid on GER40 → rank by sqn to
read the in-sample winner → generalize that winner across GER40,USDJPY,EURUSD.
Asserts the chain runs coherently, renders a verdict from `worst_case` +
`sign_agreement`, and that the GER40 leg of generalize reproduces the swept
winner's sqn (within 1e-9).
- Fits the milestone promise as a whole: it composes all three pieces into the
single "is this edge real?" verdict.
- Outcome: built, ran, matched expected (within tolerance). PASS, with the
verdict the toolkit is meant to give: the GER40 in-sample winner (sqn 0.994)
does NOT generalize (worst_case -0.149 on EURUSD, sign_agreement 2/3) →
"instrument-specific / likely overfit." The honest answer is the win.
## Findings
### [working] All three milestone pieces reach the output and read sensibly
- All four scenarios.
- Trials-deflation: the recorded `selection` block carries `deflated_score`,
`overfit_probability`, `n_trials`, `n_resamples`, `block_len`, `seed`,
`selection_metric`, `raw_winner_metric`, `mode` — a fully self-describing
provenance record on the winning member's manifest, exactly the #144 contract.
Default mode is `Argmax`; the deflated score sits below the raw metric.
- Plateau: `--select plateau:worst`/`plateau:mean` flips `mode` to
`PlateauWorst`/`PlateauMean` and swaps the deflation annotation for
`neighbourhood_score` + `n_neighbours`; the no-flag default stays
`Argmax`+deflation byte-for-byte (C23). The orthogonal rule × annotation
reshape is visible and correct from the JSON alone.
- Generalize: the per-instrument breakdown + worst-case floor + sign-agreement
print live; the M members persist as a `CrossInstrument` family each stamped
with its `instrument` lineage and the correct per-instrument pip geometry
(`pip_size=1`/`0.01`/`0.0001` for GER40/USDJPY/EURUSD). The non-R metric refusal
carries a precise message naming the four allowed R-metrics.
- Why working: the new surface was reached for the way a researcher would, and
the recorded provenance is legible and correct on the first try. These are the
wins that protect the milestone from drift.
- Recommended downstream action: carry-on.
### [bug] `aura <family-command> | head` panics with "Broken pipe (os error 32)" on stdout
- Surfaced in scenario 2 (`walkforward … | head -1`) and scenario 4
(`runs family … rank sqn | head -1`).
- What happened (verbatim):
```
thread 'main' (…) panicked at …/library/std/src/io/stdio.rs:1165:9:
failed printing to stdout: Broken pipe (os error 32)
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
```
Under `set -o pipefail` this propagated exit code 101 and aborted the script.
- Why a bug: piping a multi-line family/sweep listing into `head` (or any reader
that closes early — `head`, `less`, a closed UI pane) is the single most natural
thing a CLI consumer does. The family-listing subcommands stream many JSONL
lines and do not install a SIGPIPE handler / tolerate `EPIPE`, so the writer
panics instead of exiting quietly. The wrong-output is a Rust panic + backtrace
note where a clean truncated stream is expected. `aura run` (single line) does
not hit it; the family-emitting commands (`walkforward`, `sweep`, `mc`,
`runs family … rank`, `runs families`) do whenever they emit more lines than the
reader consumes.
- One-line repro:
`cargo run -q --bin aura -- walkforward --strategy stage1-r --real GER40 --from 1704067200000 --to 1735689599999 --fast 2,3 --slow 8,12 --stop-length 14 --stop-k 2.0 | head -1`
- Recommended downstream action: debug (RED test: a family-emitting command piped
into a head-like early-closing reader exits 0/SIGPIPE with no panic — restore
the default SIGPIPE disposition or swallow `ErrorKind::BrokenPipe` at the print
loop).
### [friction] Two of the three `generalize` refusals dump bare usage with no reason
- Scenario 3 refusals (b) single `--real` symbol and (d) multi-value candidate
flag (`--fast 3,5`).
- What happened: both print only the usage string and exit 2:
`aura: generalize [--strategy stage1-r] --real <SYM1,SYM2,...> --fast <n> …`
whereas the (c) non-R metric refusal gives a precise, actionable message:
`aura: metric 'total_pips' is not comparable across instruments;
cross-instrument scoring is R-only (sqn, sqn_normalized, expectancy_r,
net_expectancy_r)`.
- Why friction: the task completes (correct exit code), but the consumer is not
told *why*. `--real GER40` (one symbol) is refused because generalization needs
≥2 instruments — the whole point of the worst-case-across-instruments framing —
yet the message never says "needs at least two instruments." `--fast 3,5` is
refused because the candidate must be a single grid cell (generalize validates
*one* brought candidate, not a sweep) — but the usage string does not convey
"single value per candidate flag." The (c) message proves the project knows how
to write a good refusal; (b) and (d) just fall through to usage.
- Recommended downstream action: plan (tidy iteration) — give the arity/min-symbol
refusals a one-line reason mirroring the R-only message ("needs ≥2 instruments",
"candidate flags take a single value; generalize validates one cell").
### [friction] The `--select` plateau objective is invisible in `aura --help`
- Scenario 2 (the milestone's plateau-over-peak feature).
- What happened: `aura --help` (and every per-subcommand `--help`, which all
print the same single usage string) shows the walkforward form as
`aura walkforward [--strategy …] [--real …] [--name <n>|--trace <n>] [--fast …]
[--slow …] [--stop-length …] [--stop-k …]` — with **no** `--select
<argmax|plateau:mean|plateau:worst>`. The flag works (scenario 2 proves it) but
a consumer reading the help would never discover the plateau objective exists.
- Why friction: a shipped, opt-in milestone feature is undiscoverable from the
only help the CLI offers. The feature is reachable only if you already know the
flag name (from the ledger / commit log, which a downstream consumer does not
read). The mc subcommand's `--block-len`/`--resamples`/`--seed` and the
generalize `--metric` are documented in the usage; `--select` is the gap.
- Recommended downstream action: plan (tidy iteration) — add
`[--select <argmax|plateau:mean|plateau:worst>]` to the walkforward branch of
the usage string.
### [spec_gap] Two orchestration commands compute the same member's R metric with a 1-ULP difference
- Scenario 4 (and confirmed by a direct repeat probe).
- What happened: for the identical (params fast=3 slow=6 stop_length=3 stop_k=2.0,
Sept-2024 window, GER40, metric sqn), the **sweep** path records
`sqn = 0.9941661927806448` while the **generalize** path records
`0.9941661927806447` — a last-ULP difference, stable and reproducible across
re-runs (each path is internally bit-deterministic; the two paths differ).
- Why spec_gap: the ledger's C1 guarantees "same input (incl. seed) →
bit-identical run," and the cross-instrument realization note calls the
generalization aggregate "recomputable from the members." But the ledger is
silent on whether two *different commands* computing the same member's metric
must agree to the bit. A naive consumer asking "did my swept winner's GER40
score survive into generalize?" by exact string-equality would conclude they
differ. I chose the reading that this is a benign floating-point reassociation
between two reduction paths (within 1e-9), not a determinism violation — but the
equally-plausible reading is that "recomputable" should mean bit-identical, in
which case the two paths should share one reduction kernel. The contract does
not say which.
- Recommended downstream action: ratify (record in the ledger that cross-*command*
recomputation of a member metric is bit-equal-up-to-reassociation, not
bit-identical — or tighten it to share the kernel). Either resolution is a
one-line ledger note, not code the fieldtest should write.
### [spec_gap] The milestone's core vocabulary is absent from the glossary
- All scenarios (a documentation observation across the milestone surface).
- What happened: `docs/glossary.md` defines bias, R, R metrics, sweep,
walk-forward, Monte-Carlo, etc., but has **no** entry for "trials-deflation,"
"deflated score," "overfit probability," "plateau (over peak)," "cross-instrument
generalization," "worst-case floor," or "sign-agreement" — the entire vocabulary
the milestone's JSON output uses (`deflated_score`, `overfit_probability`,
`neighbourhood_score`, `worst_case`, `sign_agreement`).
- Why spec_gap: a downstream researcher reading `overfit_probability` off a winner
manifest, or `worst_case` off a generalize run, has no public definition to look
up — the terms are load-bearing for interpreting the output but undocumented on
the public surface. The ledger's open-thread prose explains them, but the
glossary (the consumer-facing term index) is silent, so the reading of each term
is left to the consumer. I read the numbers as "deflated = luck-penalised,
overfit_probability = P(this winner is noise), worst_case = the floor over
instruments" — plausible, but the public surface does not pin it.
- Recommended downstream action: tighten the design ledger / glossary — add the
five-to-seven milestone terms to `docs/glossary.md` (a docwriter task), so the
output is self-explanatory from the public interface alone.
## Recommendation summary
| Finding | Class | Action |
|---|---|---|
| All three pieces reach output, read sensibly | working | carry-on |
| `… \| head` panics with Broken pipe on stdout | bug | debug |
| Two generalize refusals dump bare usage, no reason | friction | plan |
| `--select` plateau objective absent from `--help` | friction | plan |
| Sweep vs generalize 1-ULP metric mismatch | spec_gap | ratify |
| Milestone vocabulary absent from the glossary | spec_gap | tighten the design ledger |