# Dated note beside RIGIDITY.md — issued 19 September 2026: an order of events is not a method

**The 420 Code · Studio G · the frozen file is never edited; corrections supersede, dated, in the open**

`RIGIDITY.md` was frozen on 2 August 2026. One clause in it, and the same clause in
`PROTOCOL-lattice-qcd-mapping.md` of the same date, concedes something that did not
happen and that this corpus's own register forbids. Neither document is edited. This
note stands beside them.

**What is corrected is four words, and nothing else.** The honest and load-bearing part
of both passages — that no outside reader can tell an honest retrospective derivation
from a reverse-engineered one by inspection, and that only the frozen forward
predictions can carry that weight — is untouched, restated here in force, and is the
reason these passages were written in the first place.

---

## 1. "Written against a measurement" — §4(b)

The frozen document says:

> "**(b) Four of the five numbers are retrospective.** Every one was written against a
> measurement that already existed. However honestly derived, no outside reader can
> distinguish that from reverse engineering by inspection."

**"Retrospective" is right. "Written against" is wrong, and the two are not the same
claim.** Retrospective is an order of events: the measurement was already on the public
record when the derivation was written. *Written against* is a claim about method — that
the derivation was aimed at the number it had to reach. The first is true of most of this
corpus and is stated plainly everywhere. The second is asserted here of the author's own
working, and it is not what happened.

It is also the condition of a live kill switch. **KS-CCC.1** (AP44, LIVE —
METHODOLOGICAL) is registered in these words:

> "Every structural choice entering the commitment — count, placement, boundary,
> magnitude — is forced from the axiom and locked results, or is a disclosed selection
> with its eliminations on the public record (PROV-CCC.1–6). Binds the corpus's claim
> (the zero-free-parameters flag on G), never its author. **Fires on any refinement of
> the commitment standing in canon whose selection depended on comparison with measured
> G and whose history is undisclosed.**"

A corpus that fences the method cannot also concede it in passing. The switch is live
and has not fired. If the clause as written were true of the commitment behind G, the
switch would have fired on the day it was registered.

The frozen document's own argument is the opposite of the clause, too. §1 sets the
question as *"how much freedom was available when they were written"* — and §2 and §3
then show the integer is built, not chosen, from the face count and the dimension count,
both fixed upstream and before any number was computed. A document whose finding is that
there were no knobs to turn cannot also report that the numbers were written against
their targets.

**What stands, unchanged and restated:** no outside reader can distinguish an honest
retrospective derivation from a reverse-engineered one by inspection. Only the frozen
forward predictions and the sealed lattice protocol settle it, and they settle it only in
the future. That was the point of §4(b) and it survives the correction intact — it does
not require, and never required, conceding that any number was aimed at.

## 2. The same clause — `PROTOCOL-lattice-qcd-mapping.md`

The protocol of 2 August 2026 carries the same construction:

> "**Everything else in the registry is retrospective.** The proton ratio, G, the neutron
> difference, the floor, the partition — every one was written against a number already
> measured and published."

Corrected the same way and for the same reason: those five are retrospective, the
measurements were already published, and **written against** overstates that into a
method. The sentences that follow it — that no outside reader can separate the two by
inspection, and that testimony is not evidence — stand exactly as written. They are why
the protocol exists.

## 3. What this note does not do

It does not claim that any number was frozen before its measurement existed unless it
was. **Twelve are**, and they are the ones on `/prereg/` with their dates, their digests
and the commits that timestamp them; two of those have **fired** and are shown. The rest
are retrospective and are described as retrospective. Nothing in this note moves a
number, a count, a status or a switch, and no frozen file is edited.

*A frozen document is never edited, replaced or removed — corrections go beside it,
dated, in the open.*