Have C1, S2 or P6 grown by patching the way M2 had, and should they be rethought?

[derek73]

Why. M2 had grown one stop or release check at a time across #424, #533, #535, #538 and #549, until a guard had to model three other stages' joins to stay correct. #601 rethought it from the top and deleted that machinery. docs/design/AGENTS.md now names the signs: the rule works by scanning and stopping, each fix added a stop or an exception, and a guard models what another stage would do. That last sign is the decisive one. mechanisms.md now records the two patterns the rethink used, READ-WITHOUT-THEN-BIND and LICENSED-BY-NEIGHBOR.

Rule length alone isn't the signal. Measured 2026-10-04:

Rule	decisions.md entries	decisions.md words	Rule prose words
S2	35	18,637	2,408
R4	35	14,296	1,451
C1	15	12,308	2,445
P3	11	7,052	1,351
P6	10	6,024	1,510

R4's entries are mostly vocabulary and rendering cases rather than added stops, so it's out of scope here.

The question. For each of C1, S2 and P6, does it show the decisive sign? If so, what would the READ-WITHOUT-THEN-BIND or LICENSED-BY-NEIGHBOR version look like, and roughly what would it move?

• C1 (the suffix-comma decision) first, because Should a comma part opening with a credential be a suffix comma? John Smith, PhD Jones reads family John Smith #603 is about to reopen it. A rethink would ride along with a change already being made rather than churn a settled rule.
• S2. Maiden marker counts only behind a surname; a credential starts a suffix run (#601, #602) #609's simplify pass found two hand copies of its machinery in other stages (Two stages hand-copy each other's trailing-word logic after #601/#602; share one predicate each #610). Its words-to-spare count, the company clause, the anchors, the capitals lean, the dotted shape and now Should a name word after a credential join the suffix? John Smith PhD Jones reads middle Smith PhD #602's run all interact.
• P6. Assign now re-derives P6's tail so the two stages agree, which is the decisive sign in miniature (Two stages hand-copy each other's trailing-word logic after #601/#602; share one predicate each #610, part 1).

Shape of the answer, per rule.

• Which dated entries added a stop or an exception, and which guard (if any) models another stage.
• A one-paragraph sketch of the rethought statement.
• If it looks worthwhile, a throwaway prototype behind an environment switch, measured the way Rethink the maiden-marker rule (M2): a marker stands after the current name and takes words up to the trailing suffix run #601's was: the invariant grids, the differential corpora and the case table.

Then a decision per rule: leave it, or file a #601-style issue in rule shape.

Not in scope: changing behavior here. This issue decides what to rethink, not how anything parses.

Related: #601, #603, #610.