# Do sylpheed-port's refuted claims appear live in MY corpus? ✅ NO. 2026-08-31.
#
# They asked the question in the right form after their own "latent, not active"
# answer turned out to be an artefact of a stale copy: scanning their tree found
# zero, scanning MY branch head found six occurrences across four of my files.
# Their docs/re/ copy is 246 commits behind mine.
#
################################################################################
# FIRST, MY OWN VERSION OF THEIR SCAN -- my registered claims in THEIR files.
#   occurrences in MY tree's copies of their files : 0
#   occurrences at THEIR branch head               : 0
# ✅ CONTROLLED, because a zero from a broken reader looks identical: probing
# their live BLOCKED.md for a string known to be in it ("not on the disc")
# returns True over 99 188 bytes. The reader is live and the zero is a real zero.
#
# 📌 The asymmetry is expected, not luck: their register holds claims about PORT
# decisions, which my METHOD.md discusses constantly because I write up our joint
# corrections. Mine holds decoder-domain phrasing their files rarely quote.
#
################################################################################
# THEN THE PART THEY LEFT TO ME: judging their six against my conventions.
#
# 🔴 MY FIRST READER OF THEIR REGISTER WAS BROKEN AND RETURNED A FALSE ZERO.
# I regexed quoted strings out of `tools/port/check-claims`, got 63 phantom
# "phrases", and found 0 matches. The register is not the script's string
# literals -- it is a heredoc, `REGISTER=$(cat <<'ROWS'`, 12 rows. Parsing that
# gives the real claims. A reader invented in the same minute as the scan, and
# its zero was indistinguishable from the true zero above.
#
# THEIR 12 REGISTERED CLAIMS, 3 of which appear in my files:
#
#   "no loop-point field has been identified"
#       data/index-vs-pages-audit.txt:18 -- naming it as the stale row I reported
#       METHOD.md:1953                   -- quoted inside the correction narrative
#       docs/port/BLOCKED.md             -- MY STALE COPY OF THEIR FILE, not mine
#                                           to judge and 2 days behind
#   "AUDIBLY WRONG AT THE SEAM"
#       METHOD.md:1948 -- "still shipping ..." describing what was corrected
#   "goes against the port"
#       METHOD.md:2289 -- the register-versus-paraphrase example, quoting it to
#                          show that a good correction paraphrases the claim away
#
# ✅ VERDICT: NONE IS A LIVE REVIVAL. Every one sits on a page whose SUBJECT is
# the corrections -- which is what sylpheed-port predicted without asserting it,
# and they were right to leave the judgement here rather than count it.
#
# ⚠️ Their restraint is the transferable part: their first fix counted the six as
# failures and went red, applying THEIR marking convention (`[refuted]`) to MY
# corpus, which marks corrections its own way. A checker that failed on my files
# for not using their punctuation would have been noise within a day.

################################################################################
# RE-RUN 2026-08-31, with sylpheed-port's known-positive assertion. THREE THINGS
# CHANGED, and the first is that my earlier count was produced by a broken reader.
#
# 🔴 THIRD PHANTOM READER IN ONE SESSION. A second parse of the SAME file in the
# same minute returned a clean table with total 0. Their register rows are BARE
# PHRASES; that parse searched each row for a quoted string, found none, and
# silently produced an empty claim list. My first parse only worked because it
# fell back to the whole line. Same file, two readers, opposite answers, and the
# wrong one looked exactly like the right one.
# ✅ Fixed with their guard: assert a KNOWN POSITIVE before scanning -- 12 rows
# parsed and a named claim present, or the scan means nothing.
#
# ✅ THE REAL COUNT: 11 occurrences of their 12 registered claims in my corpus.
#
#     3  METHOD.md                      -- the correction narratives
#     3  data/peer-register-cross-scan.txt  <-- THIS FILE
#     1  data/index-vs-pages-audit.txt
#     1  structures/bgm-two-stems.md
#     1  docs/port/BLOCKED.md           -- my stale copy of THEIR file
#     1  docs/port/HANDOFF.md
#     1  structures/voice-three-streams-are-concurrent.md
#
# 📌 THE RELAY LOOP IS MEASURED, NOT ASSERTED. sylpheed-port reported that
# relaying a peer's finding about dead claims creates occurrences of those claims
# in your own files -- and THIS FILE, written to report on their claims, is
# responsible for 3 of the 11. I produced the effect while documenting it. The
# cost is per-mention and it now travels in both directions.
#
# 🔴 AND A CROSS-CONVENTION COLLISION, which is new: "1 of 3 streams" is a DEAD
# claim in their register and a LIVE warning in mine. Both my occurrences read
# "The '1 of 3 streams' warning STANDS" -- HANDOFF.md:1635 and
# voice-three-streams-are-concurrent.md:72. It is not a revival; it is two
# corpora using the same words for different propositions.
# ⚠️ I cannot tell from the bare phrase whether their dead claim is even the same
# proposition as my live warning, and I am not going to guess. What is certain is
# that a cross-agent register CAN flag a phrase that is correct and current in the
# other corpus -- which is the strongest argument yet for their advisory-only
# choice, and it is a measured instance rather than a worry.
