From ea6e4dbd0faa3d05e4d7c8891fa8b28b259baed2 Mon Sep 17 00:00:00 2001 From: kyuhwan Date: Wed, 16 Sep 2026 13:06:48 -0700 Subject: [PATCH] kb/literature/README.md still described G10, and now has a real case instead MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Four claims in this file stopped being true, three of them when G10 was removed on 2026-09-09 and one when the folder got its first entry: * *"No entries yet."* There is one -- `lifetime-ns-tritc.md`. The schema and rules were written first so the first entry would not have to invent them, and they held: nothing in the template needed changing to file it. * The live case was `bleach_photons` under **G10**, with a hypothetical output block showing a `perturbation.photobleaching` margin. That gate is gone. * *"`bleach_photons` per dye is the top example"*, cited to 07 Phase 0 as literature-or-decay-curve. **No gate consumes it now**, and `docs/07-roadmap.md` no longer names it. Still empty, and no longer a number anything waits on -- the difference between a gap and a hole. The live case is now real rather than illustrated, and both blocks are output from this repository at `c556db0`: TRITC @ 555 nm 153.0 W/cm^2 -> BLOCKED [FAIL] missing.lifetime TRITC @ 555 nm 153.0 W/cm^2 -> PASS 10.00 perturbation.saturation evidence: assumed advances: NO **And it makes the tier's point better than the version it replaces.** The old example showed a literature value producing a margin that said *no*, which is easy to accept. This one produces a margin that says *yes, comfortably* -- and still cannot advance a verdict. A passing verdict is exactly where a reader is tempted to treat an assumed value as a measured one. The photobleaching arithmetic the old block described is kept in 04 §6, and the replaced paragraph says so rather than dropping it silently. `lifetime_ns` is the worked example in its place: G20 reads it, five of seventeen dyes are still missing it, and no instrument here can measure it directly -- that needs time-resolved detection this setup does not have, which is what makes it literature rather than a calibration. Suite: 1221 passed, 11 skipped. Documentation only; no registry or code change. --- kb/literature/README.md | 52 ++++++++++++++++++++++++++++++----------- 1 file changed, 39 insertions(+), 13 deletions(-) diff --git a/kb/literature/README.md b/kb/literature/README.md index 356a31e..fc418f0 100644 --- a/kb/literature/README.md +++ b/kb/literature/README.md @@ -19,8 +19,10 @@ One file per quantity per subject. `bleach-photons-alexafluor488.md`, not it came out of. One paper supplying three quantities becomes three files, each citing it. Files beginning with `_` are not entries. -**No entries yet.** The schema and the rules are here first, so the first entry -does not have to invent them. +**One entry**, [`lifetime-ns-tritc.md`](lifetime-ns-tritc.md) — TRITC's +fluorescence lifetime, which `G20` was refusing for want of. The schema and the +rules were written first so that it did not have to invent them, and they held: +nothing in the template needed changing to file it. --- @@ -35,20 +37,37 @@ This repository has two evidence tiers, `measured` and `assumed`, and worth having: a gate that was returning `BLOCKED` for want of an input can instead *compute*, and report a margin, a bottleneck, and a difficulty grade. -The photo-perturbation lens is the live case. `bleach_photons` is empty for every -dye in `data/fluorophores.yaml`, so G10 has nothing to count against and the lens -refuses. Supply the number and it computes — and still says no: +The photo-perturbation lens is still the live case, and it is now a real one +rather than an illustration. `lifetime_ns` was `null` for TRITC, so `G20` could +not tell whether emission was linear in power and the lens refused: ``` -488 (AlexaFluor488) @ 470 nm 6.2 W/cm^2 -> PASS_WITH_CHANGES -feasibility: MARGINAL evidence: assumed confidence: low advances: NO +TRITC @ 555 nm 153.0 W/cm^2 -> BLOCKED +feasibility: UNKNOWN evidence: assumed confidence: none advances: NO + [FAIL] missing.lifetime +``` + +Supply the number — 2.2 ns, free dye in PBS — and it computes: - 0.20 perturbation.photobleaching ## ``` +TRITC @ 555 nm 153.0 W/cm^2 -> PASS +feasibility: ROUTINE evidence: assumed confidence: low advances: NO + + 10.00 perturbation.saturation ############################## +``` + +**`advances: NO` beside a margin of 10.00 is the point**, and it is sharper than +the version that used to sit here. That one showed a literature value producing +a margin that said *no*, which is easy to accept. This one produces a margin +that says *yes, comfortably* — and still cannot advance a verdict. A passing +verdict is exactly where a reader is tempted to treat an assumed value as a +measured one. -That 0.20 margin is the point. It says ~100% of the label bleaches over 7200 -frames, which is actionable now, and it says so without claiming anyone measured -this dye in this buffer under this illumination. +> **This paragraph illustrated `G10` and `bleach_photons` until 2026-09-16.** +> G10 was removed on 2026-09-09 ([kb/decisions/2026-09-09-g10-photobleaching-removed.md](../decisions/2026-09-09-g10-photobleaching-removed.md)) +> and the output above it was hypothetical. The photobleaching arithmetic it +> described is kept in [04 §6](../../docs/04-decision-engine.md), which also +> says why the gate went. If an entry here ever appears to make a verdict advance, that is a bug in the gate, not a licence. @@ -57,8 +76,15 @@ gate, not a licence. **Belongs.** A number a gate consumes, which has no local measurement, whose published value is specific enough to transfer with stated caveats. -`bleach_photons` per dye is the top example — [07 Phase 0](../../docs/07-roadmap.md) -lists it as literature-or-decay-curve, and it is empty across the whole registry. +`lifetime_ns` per dye is the worked example: `G20` reads it, five of seventeen +dyes are still missing it, and no instrument here can measure it directly — +that needs time-resolved detection this setup does not have. + +> This said `bleach_photons` was the top example, *"empty across the whole +> registry"* and listed by 07 Phase 0 as literature-or-decay-curve. **No gate +> consumes it now** — it went with G10 — and `docs/07-roadmap.md` no longer +> names it. Still empty, and no longer a number anything is waiting on, which +> is the difference between a gap and a hole. **Does not belong.**