- *negative result:* on **Qwen-3.6-35B-A3B**, `2610.20826`'s phase/entropy/CV/Jaccard/top-1 proxy is neutral near 4 bpw and loses at the steep IQ2 floor: uniform KLD `0.034321`, proxy `0.038142–0.038249`, random `0.037728/0.037808` (random marginally better, within about 1 SE of proxy). proxy p99 KLD rises `0.437 -> 0.487–0.492`; mandatory downgrades inject tail damage. this is a coding-domain, layerwise negative result, not a theorem against all mixed precision or a measurement of our IQ2 formats.
- *our boundary:* the K512 header has **144 packed expert tensors**, not 144 independently precision-assigned experts: `68 IQ2_S`, `48 Q2_0`, `24 IQ2_XXS`, `4 IQ3_S`; each tensor packs 512 experts with one type. ukisai's imatrix/GSQ-RCO artifact is not DuoRate's HQQ/GPTQ/MiLo/MoEQuant/BitsMoE baseline. its IQ codes are not prefix-readable StagQ planes; truncating their bytes is corruption, not Stage-2 read-tier selection.
- *rule:* preserve the current quant assignment unless a format-/byte-/arena-matched ablation beats it on mean **and tail** KLD, top-1, persona/JA/ops/tool behavior and measured SSD/GPU time. cache ranks may optimize reuse, but must not silently become precision ranks. cache-conditioned SSD miss weights also differ from distortion weights, so Proposition 1 cannot be copied unchanged into `stream.cpp`.
44. **reading single-token top-1 agreement, or a single-run prefill tok/s, as signal: the teacher-forced decode noise floor** (`audits/opus_prefill_opt.md:82-89`, `call_1381557`, `call_6642651`, `call_1274067`):
- *trap:* treating a `96.88%` vs `98.44%` vs `100.00%` top-1 column, or a `205.28` vs `246.83 tok/s`prefill column, as a property of the knob under test. neither is.
- *decode noise floor, measured:* **base vs base, identical config, run twice** gives `mean KLD 0.0012`, **`top-1 95.3%`**. decode is not run-to-run deterministic even with MoCE off, because residency/timing-dependent read-ahead paths (`EXPERT_SPLIT`, `SPLIT_TURN`) change which slot an expert sits in and when it is read. so a 1-token top-1 flip out of 64 eval tokens (`63/64` vs `64/64`) is exactly one noise event, not a result. `98.44%` vs`100.00%` on a 64-token teacher-forced eval is **one coin toss**.
- *prefill noise floor, measured:* in the 5-arm multi-chunk sweep on`phase4_contrastive_vn_chunk0.txt` (`2,487` tokens), two arms with the **identical**chunk-0 setting (`300.85`) came out at `10,075.62 ms`and `12,115.41 ms`— a **20% spread from ordering alone**, driven by the arena fill read rate (`2.2 s`→ `0.8 s` as the sweep warmed the SSD). every prefill number in a sequential sweep of arms is confounded by position.
- *rule:* (a) quote top-1 only when the delta exceeds one token on 64, i.e. state the raw `n/64`; (b) never rank prefill arms from a sequential sweep — interleave them (A/B/A/B, median of >= 5 pairs) or pin the arena with `STRATA_METAL_STREAM_FAKE=1`; (c) a KLD ordering is trustworthy at `> 0.002` separation, a top-1 ordering is not trustworthy below `2` tokens; (d) `rc != 0`arms contribute nothing — `call_6642651` returned `rc=3` on all four arms (the `gdn_prompt.metal:891` `MTL` 3.2 `thread`-qualifier compile failure) and its "results" are arena-setup log lines, not measurements.
44. **measurement traps found by re-measuring the chunk-0 MoCE default: the noise floor was measured on an arm that does not share the mechanism, "reproducible" was claimed from n=2, position-level KL exonerated a change that flipped a boolean, and prefill counters are invisible without `STRATA_METAL_TRACE=1`** (`call_1381557`, `call_6642651`, `call_1274067`, `sol-moce-chunk0` `d18939749fb6f294fb0c272a61cca5d1`, `audits/opus_prefill_opt.md:82-89,92-120`):
- *trap 1 — a control can be deterministic for a reason its treatments do not share.* `repeat0` (arm `0`) came out **exactly `0.000000`** mean KLD, `64/64`, on both corpora and both cycles. that is structural, not luck: `prompt_moce = 0` unpacks to `p_prot = 0, p_bonus = 0` (`prompt.metal:41-47`), so the bias path at `prompt.metal:51` never fires and **routing never reads `smap`**, making arm 0 residency-independent. every *boosted* arm reads `smap[idx] != 0xFFFFFFFFu` at `:55`, so its routing depends on which experts are instantaneously resident and therefore on I/O timing. **a floor measured on the arm that ignores the mechanism bounds nothing for the arms that depend on it.** the earlier `0.0012 / 95.3%`floor (`audits/opus_prefill_opt.md:83-87`) was a property of that streamed configuration, not a constant, and must not be quoted as one.
- *trap 2 — n=2 agreeing is not reproducibility.* ops incumbent KLD came out `0.055902` then `0.055018` (spread `0.000884`, `1.6%`), and that was written up as "reproducible ... tighter even than its chat value". a third run of the same config gave **`0.076038`** (`+38%`). boosted-arm KLD estimates on ops are not reproducible to better than`~0.02`. **require a third run before writing the word "reproducible".**
- *trap 3 — heavy-tail KL and argmax damage are largely disjoint sets, so position-level KL alone can exonerate a bad change.*the incumbent's highest-KL positions were`The` (`0.873594`) and `plan` (`0.674194`) — prose, **argmax unchanged**— with `49.65%` of KL in the top `4` positions and `72.47%`in the top `8`. the actual argmax flips were elsewhere and small-KL, including ` true -> false`at position 28 (`KL 0.307667`) and backtick -> `()` at position 24 (`0.035575`). one boolean inversion per 64 ops tokens is disqualifying on its own.
- *trap 4 — metric power is corpus-dependent, so a metric that is saturated on one corpus can be the deciding evidence on another.* top-1 was `64/64` for **every** arm on chat (zero discriminating power at that KL magnitude) and `60/64`, `61/64`, `59/64`/ `59/64`, `58/64` on ops. `audits/opus_prefill_opt.md`'s "base vs base top-1 95.3%" was likewise a chat-class result. never declare a metric uninformative from one corpus.
- *trap 5 — 64-token teacher-forced PPL cannot resolve a `~0.05`-nat KL, and no standard error is derivable.* `KL = H(p,q) - H(p)` connects teacher-forced NLL to KL only if the targets are `p`-samples; fixed corpus ids are not, so neither the identity nor a Monte-Carlo SE applies. `cand_ppl 2.4035` vs `ref_ppl 2.5027` at `n=64` therefore corroborates and contradicts nothing. quote KLD (an exact per-position expectation over the vocab) as the resolving statistic and say PPL does not estimate full-vocab KL.
- *trap 6 — prefill byte/read accounting does not exist unless `STRATA_METAL_TRACE=1`.* `strata_window_cli.cpp:1827` calls `window_reset_stats(w)` unconditionally after prefill + the first head step, `window.cpp:4137` prints only under `PKPROF`/`TRACE` and then **always** `reset_stream_stats()`, and `stream.cpp:1177-1180` zeroes `p_lookups_ p_hits_ p_pre_ p_misses_ p_streamed_ p_chunks_ p_bytes_read_ p_reads_`. the shipped `375.2 MB / 258 reads` in `results/round39/*.log:25` reports `480 distinct lookups` = `48 layers x top_k 10` = **one decode step, not a prefill**. also `p_streamed_` is a **count of submissions, not bytes** (`stream.cpp:893,931,1117`); bytes is `p_bytes_read_` (`:1192`).
- *trap 7 — never rank arms from a sequential sweep.* two arms with the **identical** chunk-0 setting landed `10,075.62 ms` and `12,115.41 ms` (**20% apart**) purely from sweep position, via the arena fill read rate going `2.2 s -> 0.8 s` as the SSD warmed. interleave (A/B/A/B), discard each arm's first run, report the median of paired ratios, and keep the paired-with-off comparisons separate from arm-vs-arm block comparisons (different arms are each paired *against off*, not against each other).
- *rule:* a residency-biased routing knob must be quality-tested on **ops**, not just chat; an arm that reads residency is stochastic by construction and an arm that does not read it is bit-exact by construction; state arms in `n/64` and name the corpus; and a rejected optimization is worth more when the report states all three of speed (real), mechanism (real), and quality cost (real, corpus-specific, unaffordable).
### B. ruled out from the original blueprint
details and primary-source checks in `blueprint_review.md`.
- **the blueprint's expert size math and mixed-quant expert tiers:** claimed 224 experts × 48 layers at Q3_K is 14.2 GB (real: **22.7 GB**, or 16.5 GB at Q2_0), and proposed mixing Q3_K hot experts with Q2_K cold experts inside one layer (impossible in GGUF: `ffn_*_exps` is one 3D tensor per layer with one quant type [src: ISTA card]).