clay-borg/tools/design.py

214 lines
9.5 KiB
Python
Raw Normal View History

CB-WP-0022 T05/T06/T07: the register, and what its first run found T05. tools/design.py, make design, and the register in GroundRules.md -- 14 rows, no new file, because ADR-0012 D2 made §Underdetermined the register rather than building one beside it. Backfill was the test and it caught two things the ADR did not have. First, a `role` column. The first report alarmed on U2 and was wrong to: U2's scenario is green BECAUSE the provisional default it documents is implemented, which says nothing about whether ground-game agrees. GR-E01's was a counterexample that went green. Same colour, opposite meaning -- a register that cannot tell them apart either alarms constantly or never. Only a green counterexample alarms. Folded back into GameDesign §1.3. Second, it contradicted the survey. CB-RES-0007 said "six of the ten already have provisional scenarios." Measured -- grep -lE "\bU<n>\b" over scenarios/ground -- exactly ONE U-item names itself. Five provisional scenarios exist and four probably encode U-item defaults, but the mapping is not written down, so it is not checkable. Same defect class as the wrong premises, found inside the survey that proposed the fix. Now a reported debt: open, lacking a reproduction: 9, target 0. design.py carries the control design-baseline.py never had, asserted directly: a row citing a nonexistent file must not count as reproduced, using the exact path 2da19a4 deleted -- which the old tool called green. design-baseline.py is marked superseded rather than deleted; it is the evidence for how a wrong number got into a survey. T06. The report is a FILE in ground-game under GROUND-WP-0002, committed there, with a hub message that only points at it. It asks for no ruling: it carries GR-E01's withdrawal, our reproduction debt, and two notes that are explicitly not findings. And it had to acknowledge something nobody anticipated. GROUND-WP-0002 is finished -- all ten U-items were RULED 2026-08-03, every one confirmed as the default we simulate, plus five of six provisional scenarios. The survey said "0 of 10 ruled" two days later and this register was built saying `reported`. That is the unread-inbox failure running in the opposite direction: they answered and we did not collect it. The instrument's first run surfaced it. They are `ruled`, not `applied` -- lifting the now-settled provisional flags is owed and is not done, and make design shows them open until it is. T07. evidence/CB-EV-0021. Two of six catches in this pass came from execution rather than process (the role distinction from building it, the ten uncollected rulings from running it), which is InnerLoop §Design goal's prediction holding. make self-tests, facts-check, loop-lint: clean. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-05 15:22:33 +02:00
#!/usr/bin/env python3
"""design — report the finding register (CB-WP-0022 T05).
Implements the metrics in `specs/GameDesign.md` §4 over the register in
CB-WP-0026: collect the rulings -- ten answers that arrived and were never applied ground-game ruled all ten U-items on 2026-08-03, every one CONFIRMED as the default clay-borg simulates, and confirmed five of six provisional scenarios. clay-borg never collected the answers: CB-RES-0007 reported "0 of 10 ruled" the same day, and CB-WP-0022 built the finding register two days later still recording them as `reported`. make design's first run is what noticed -- not a human, not the adversarial review that found four other things. That is the unread-inbox failure running in the opposite direction, and it appears nowhere in the declaration, survey, ADR or spec of the pass that was built entirely around the forward version. It is arguably worse: an unread message is visible as silence, while a collected-but-unapplied ruling looks exactly like work in progress. Ten rulings quoted into §Underdetermined (the three conditional ones verbatim -- U1's designer note, U2's End-only trigger, U8's consume-only-if-it-cancels). Five provisional flags lifted, replaced by ruled/ruled_by/ruled_note so the flag went and the provenance stayed. Register queue 9 -> 0. T02's control came back clean: make sim is 26 passed, 59 rules covered, nothing red. Had a scenario gone red it would have meant we described our own behaviour incorrectly to ground-game. I wrote two U-item mappings and both were wrong. gr-a04 -> U1 (it asserts consent is REQUIRED; U1 asks WHEN the target accepts) and gr-d05 -> U5 (it exercises the UNREJECTED Reverse; U5 is the rejected one). Both plausible from covers:, neither survived reading the description. Third and fourth instance of this defect; the first two reached ground-game. So encodes_u_item is now a declaration and design.py asserts the file names what it claims -- and that check's own first version grepped for mentions and went red when two files recorded why they do NOT encode U1 and U5. A mention is not a claim, which is exactly the looseness that let "six of the ten have provisional scenarios" stand. Two positive controls went red for the best possible reason, both broken the same way -- asserting against live repo data instead of constructing their condition. rule-coverage.py required at least one provisional item to EXIST; it now builds a fixture and reports the live count as a diagnostic, because there is no number of provisional items this project should have. design-baseline.py pinned "2 of 6" while recomputing one row from a live glob, so the dated snapshot was never a snapshot; frozen to its 2026-08-03 list and unwired from self-tests, since per ADR-0012 D8 it is no longer a reporting tool. ScenarioFile is deny_unknown_fields and refused the four new fields until declared -- correct: a corpus accepting unknown metadata would let a typo'd encodes_u_iem sit there claiming nothing. DEVIATION: ADR-0012 D2 said "no new file". GroundRules.md crossed the loadability limit, so the register moved to specs/FindingRegister.md. D2's substance holds -- one register, same machinery, nothing competing -- but the literal instruction did not, and it resolves an awkwardness D2 named itself. make all: exit 0. loop-lint clean. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-05 16:13:37 +02:00
`specs/FindingRegister.md`. ADR-0012 D8 retired `design-baseline.py`, which
CB-WP-0022 T05/T06/T07: the register, and what its first run found T05. tools/design.py, make design, and the register in GroundRules.md -- 14 rows, no new file, because ADR-0012 D2 made §Underdetermined the register rather than building one beside it. Backfill was the test and it caught two things the ADR did not have. First, a `role` column. The first report alarmed on U2 and was wrong to: U2's scenario is green BECAUSE the provisional default it documents is implemented, which says nothing about whether ground-game agrees. GR-E01's was a counterexample that went green. Same colour, opposite meaning -- a register that cannot tell them apart either alarms constantly or never. Only a green counterexample alarms. Folded back into GameDesign §1.3. Second, it contradicted the survey. CB-RES-0007 said "six of the ten already have provisional scenarios." Measured -- grep -lE "\bU<n>\b" over scenarios/ground -- exactly ONE U-item names itself. Five provisional scenarios exist and four probably encode U-item defaults, but the mapping is not written down, so it is not checkable. Same defect class as the wrong premises, found inside the survey that proposed the fix. Now a reported debt: open, lacking a reproduction: 9, target 0. design.py carries the control design-baseline.py never had, asserted directly: a row citing a nonexistent file must not count as reproduced, using the exact path 2da19a4 deleted -- which the old tool called green. design-baseline.py is marked superseded rather than deleted; it is the evidence for how a wrong number got into a survey. T06. The report is a FILE in ground-game under GROUND-WP-0002, committed there, with a hub message that only points at it. It asks for no ruling: it carries GR-E01's withdrawal, our reproduction debt, and two notes that are explicitly not findings. And it had to acknowledge something nobody anticipated. GROUND-WP-0002 is finished -- all ten U-items were RULED 2026-08-03, every one confirmed as the default we simulate, plus five of six provisional scenarios. The survey said "0 of 10 ruled" two days later and this register was built saying `reported`. That is the unread-inbox failure running in the opposite direction: they answered and we did not collect it. The instrument's first run surfaced it. They are `ruled`, not `applied` -- lifting the now-settled provisional flags is owed and is not done, and make design shows them open until it is. T07. evidence/CB-EV-0021. Two of six catches in this pass came from execution rather than process (the role distinction from building it, the ten uncollected rulings from running it), which is InnerLoop §Design goal's prediction holding. make self-tests, facts-check, loop-lint: clean. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-05 15:22:33 +02:00
was a hand-maintained dict counting itself; the difference that matters is
that **every number here is computed over rows that name real files, and
the reproduction check stats the file.**
`design-baseline.py`'s `has_reproduction` was `p.startswith("scenarios/")`
and nothing else, so its own positive control was green against a path
that had been deleted. The self-test below asserts the opposite property
directly: a row citing a nonexistent file must NOT count as reproduced.
"""
import os, re, sys, subprocess, datetime
ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
CB-WP-0026: collect the rulings -- ten answers that arrived and were never applied ground-game ruled all ten U-items on 2026-08-03, every one CONFIRMED as the default clay-borg simulates, and confirmed five of six provisional scenarios. clay-borg never collected the answers: CB-RES-0007 reported "0 of 10 ruled" the same day, and CB-WP-0022 built the finding register two days later still recording them as `reported`. make design's first run is what noticed -- not a human, not the adversarial review that found four other things. That is the unread-inbox failure running in the opposite direction, and it appears nowhere in the declaration, survey, ADR or spec of the pass that was built entirely around the forward version. It is arguably worse: an unread message is visible as silence, while a collected-but-unapplied ruling looks exactly like work in progress. Ten rulings quoted into §Underdetermined (the three conditional ones verbatim -- U1's designer note, U2's End-only trigger, U8's consume-only-if-it-cancels). Five provisional flags lifted, replaced by ruled/ruled_by/ruled_note so the flag went and the provenance stayed. Register queue 9 -> 0. T02's control came back clean: make sim is 26 passed, 59 rules covered, nothing red. Had a scenario gone red it would have meant we described our own behaviour incorrectly to ground-game. I wrote two U-item mappings and both were wrong. gr-a04 -> U1 (it asserts consent is REQUIRED; U1 asks WHEN the target accepts) and gr-d05 -> U5 (it exercises the UNREJECTED Reverse; U5 is the rejected one). Both plausible from covers:, neither survived reading the description. Third and fourth instance of this defect; the first two reached ground-game. So encodes_u_item is now a declaration and design.py asserts the file names what it claims -- and that check's own first version grepped for mentions and went red when two files recorded why they do NOT encode U1 and U5. A mention is not a claim, which is exactly the looseness that let "six of the ten have provisional scenarios" stand. Two positive controls went red for the best possible reason, both broken the same way -- asserting against live repo data instead of constructing their condition. rule-coverage.py required at least one provisional item to EXIST; it now builds a fixture and reports the live count as a diagnostic, because there is no number of provisional items this project should have. design-baseline.py pinned "2 of 6" while recomputing one row from a live glob, so the dated snapshot was never a snapshot; frozen to its 2026-08-03 list and unwired from self-tests, since per ADR-0012 D8 it is no longer a reporting tool. ScenarioFile is deny_unknown_fields and refused the four new fields until declared -- correct: a corpus accepting unknown metadata would let a typo'd encodes_u_iem sit there claiming nothing. DEVIATION: ADR-0012 D2 said "no new file". GroundRules.md crossed the loadability limit, so the register moved to specs/FindingRegister.md. D2's substance holds -- one register, same machinery, nothing competing -- but the literal instruction did not, and it resolves an awkwardness D2 named itself. make all: exit 0. loop-lint clean. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-05 16:13:37 +02:00
REGISTER = os.path.join(ROOT, "specs", "FindingRegister.md")
CB-WP-0022 T05/T06/T07: the register, and what its first run found T05. tools/design.py, make design, and the register in GroundRules.md -- 14 rows, no new file, because ADR-0012 D2 made §Underdetermined the register rather than building one beside it. Backfill was the test and it caught two things the ADR did not have. First, a `role` column. The first report alarmed on U2 and was wrong to: U2's scenario is green BECAUSE the provisional default it documents is implemented, which says nothing about whether ground-game agrees. GR-E01's was a counterexample that went green. Same colour, opposite meaning -- a register that cannot tell them apart either alarms constantly or never. Only a green counterexample alarms. Folded back into GameDesign §1.3. Second, it contradicted the survey. CB-RES-0007 said "six of the ten already have provisional scenarios." Measured -- grep -lE "\bU<n>\b" over scenarios/ground -- exactly ONE U-item names itself. Five provisional scenarios exist and four probably encode U-item defaults, but the mapping is not written down, so it is not checkable. Same defect class as the wrong premises, found inside the survey that proposed the fix. Now a reported debt: open, lacking a reproduction: 9, target 0. design.py carries the control design-baseline.py never had, asserted directly: a row citing a nonexistent file must not count as reproduced, using the exact path 2da19a4 deleted -- which the old tool called green. design-baseline.py is marked superseded rather than deleted; it is the evidence for how a wrong number got into a survey. T06. The report is a FILE in ground-game under GROUND-WP-0002, committed there, with a hub message that only points at it. It asks for no ruling: it carries GR-E01's withdrawal, our reproduction debt, and two notes that are explicitly not findings. And it had to acknowledge something nobody anticipated. GROUND-WP-0002 is finished -- all ten U-items were RULED 2026-08-03, every one confirmed as the default we simulate, plus five of six provisional scenarios. The survey said "0 of 10 ruled" two days later and this register was built saying `reported`. That is the unread-inbox failure running in the opposite direction: they answered and we did not collect it. The instrument's first run surfaced it. They are `ruled`, not `applied` -- lifting the now-settled provisional flags is owed and is not done, and make design shows them open until it is. T07. evidence/CB-EV-0021. Two of six catches in this pass came from execution rather than process (the role distinction from building it, the ten uncollected rulings from running it), which is InnerLoop §Design goal's prediction holding. make self-tests, facts-check, loop-lint: clean. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-05 15:22:33 +02:00
BEGIN = "<!-- design-register:begin -->"
END = "<!-- design-register:end -->"
KINDS = {"underdetermined", "inconsistent", "inert", "degenerate", "unplayed"}
OPEN_STATES = {"raised", "reported", "ruled"} # the queue: not yet applied
CLOSED_STATES = {"applied", "withdrawn"} # the log
NOTE = "note"
NOTE_EXPIRY_DAYS = 30 # GameDesign §3.1, same figure as rule-coverage.py
def parse(text):
"""Rows between the register markers. Raises if the block is absent —
a register that silently reports zero findings is worse than one that
fails."""
try:
block = text.split(BEGIN)[1].split(END)[0]
except IndexError:
raise SystemExit(f"no design-register block in {REGISTER}")
rows = []
for line in block.splitlines():
line = line.strip()
if not line.startswith("|") or line.startswith("|---"):
continue
cells = [c.strip() for c in line.strip("|").split("|")]
if len(cells) != 7 or cells[0] in ("id",):
continue
rows.append(dict(zip(
("id", "kind", "state", "repro", "role", "raised", "owner"), cells)))
return rows
def reproduced(row, root=ROOT):
"""GameDesign §1.1: the artifact must resolve. A named test is admitted
by its `crate::module::name` shape; anything else must be a real path
on disk, and that is checked by stat, not by prefix."""
p = row["repro"]
if p in ("", "", "-"):
return False
if "::" in p: # a named test
return True
return os.path.exists(os.path.join(root, p))
_SIM = None
def sim_passes(root=ROOT):
"""`make sim` is the authority on whether scenarios pass. Run once and
cache asking per row costs a full suite per finding."""
global _SIM
if _SIM is None:
r = subprocess.run(["make", "sim"], cwd=root, capture_output=True, text=True)
_SIM = r.returncode == 0
return _SIM
def alarming(row, root=ROOT):
"""§1.3, and the distinction the backfill discovered: **only a
counterexample alarms when green.**
A `default` reproduction encodes a provisional choice and is SUPPOSED
to pass U2's scenario is green because the default it documents is
implemented, which says nothing about whether ground-game agrees. A
`counterexample` is the model-checker shape: it demonstrates the
defect, so it must be red while the finding is open. GR-E01's went
green when the edition landed and nothing noticed for four days.
"""
if row["role"] != "counterexample":
return False
if not row["repro"].startswith("scenarios/"):
return False
return sim_passes(root)
def report(root=ROOT, today=None):
today = today or datetime.date.today()
CB-WP-0026: collect the rulings -- ten answers that arrived and were never applied ground-game ruled all ten U-items on 2026-08-03, every one CONFIRMED as the default clay-borg simulates, and confirmed five of six provisional scenarios. clay-borg never collected the answers: CB-RES-0007 reported "0 of 10 ruled" the same day, and CB-WP-0022 built the finding register two days later still recording them as `reported`. make design's first run is what noticed -- not a human, not the adversarial review that found four other things. That is the unread-inbox failure running in the opposite direction, and it appears nowhere in the declaration, survey, ADR or spec of the pass that was built entirely around the forward version. It is arguably worse: an unread message is visible as silence, while a collected-but-unapplied ruling looks exactly like work in progress. Ten rulings quoted into §Underdetermined (the three conditional ones verbatim -- U1's designer note, U2's End-only trigger, U8's consume-only-if-it-cancels). Five provisional flags lifted, replaced by ruled/ruled_by/ruled_note so the flag went and the provenance stayed. Register queue 9 -> 0. T02's control came back clean: make sim is 26 passed, 59 rules covered, nothing red. Had a scenario gone red it would have meant we described our own behaviour incorrectly to ground-game. I wrote two U-item mappings and both were wrong. gr-a04 -> U1 (it asserts consent is REQUIRED; U1 asks WHEN the target accepts) and gr-d05 -> U5 (it exercises the UNREJECTED Reverse; U5 is the rejected one). Both plausible from covers:, neither survived reading the description. Third and fourth instance of this defect; the first two reached ground-game. So encodes_u_item is now a declaration and design.py asserts the file names what it claims -- and that check's own first version grepped for mentions and went red when two files recorded why they do NOT encode U1 and U5. A mention is not a claim, which is exactly the looseness that let "six of the ten have provisional scenarios" stand. Two positive controls went red for the best possible reason, both broken the same way -- asserting against live repo data instead of constructing their condition. rule-coverage.py required at least one provisional item to EXIST; it now builds a fixture and reports the live count as a diagnostic, because there is no number of provisional items this project should have. design-baseline.py pinned "2 of 6" while recomputing one row from a live glob, so the dated snapshot was never a snapshot; frozen to its 2026-08-03 list and unwired from self-tests, since per ADR-0012 D8 it is no longer a reporting tool. ScenarioFile is deny_unknown_fields and refused the four new fields until declared -- correct: a corpus accepting unknown metadata would let a typo'd encodes_u_iem sit there claiming nothing. DEVIATION: ADR-0012 D2 said "no new file". GroundRules.md crossed the loadability limit, so the register moved to specs/FindingRegister.md. D2's substance holds -- one register, same machinery, nothing competing -- but the literal instruction did not, and it resolves an awkwardness D2 named itself. make all: exit 0. loop-lint clean. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-05 16:13:37 +02:00
rows = parse(open(os.path.join(root, "specs", "FindingRegister.md")).read())
CB-WP-0022 T05/T06/T07: the register, and what its first run found T05. tools/design.py, make design, and the register in GroundRules.md -- 14 rows, no new file, because ADR-0012 D2 made §Underdetermined the register rather than building one beside it. Backfill was the test and it caught two things the ADR did not have. First, a `role` column. The first report alarmed on U2 and was wrong to: U2's scenario is green BECAUSE the provisional default it documents is implemented, which says nothing about whether ground-game agrees. GR-E01's was a counterexample that went green. Same colour, opposite meaning -- a register that cannot tell them apart either alarms constantly or never. Only a green counterexample alarms. Folded back into GameDesign §1.3. Second, it contradicted the survey. CB-RES-0007 said "six of the ten already have provisional scenarios." Measured -- grep -lE "\bU<n>\b" over scenarios/ground -- exactly ONE U-item names itself. Five provisional scenarios exist and four probably encode U-item defaults, but the mapping is not written down, so it is not checkable. Same defect class as the wrong premises, found inside the survey that proposed the fix. Now a reported debt: open, lacking a reproduction: 9, target 0. design.py carries the control design-baseline.py never had, asserted directly: a row citing a nonexistent file must not count as reproduced, using the exact path 2da19a4 deleted -- which the old tool called green. design-baseline.py is marked superseded rather than deleted; it is the evidence for how a wrong number got into a survey. T06. The report is a FILE in ground-game under GROUND-WP-0002, committed there, with a hub message that only points at it. It asks for no ruling: it carries GR-E01's withdrawal, our reproduction debt, and two notes that are explicitly not findings. And it had to acknowledge something nobody anticipated. GROUND-WP-0002 is finished -- all ten U-items were RULED 2026-08-03, every one confirmed as the default we simulate, plus five of six provisional scenarios. The survey said "0 of 10 ruled" two days later and this register was built saying `reported`. That is the unread-inbox failure running in the opposite direction: they answered and we did not collect it. The instrument's first run surfaced it. They are `ruled`, not `applied` -- lifting the now-settled provisional flags is owed and is not done, and make design shows them open until it is. T07. evidence/CB-EV-0021. Two of six catches in this pass came from execution rather than process (the role distinction from building it, the ten uncollected rulings from running it), which is InnerLoop §Design goal's prediction holding. make self-tests, facts-check, loop-lint: clean. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-05 15:22:33 +02:00
findings = [r for r in rows if r["state"] != NOTE]
notes = [r for r in rows if r["state"] == NOTE]
queue = [r for r in findings if r["state"] in OPEN_STATES]
closed = [r for r in findings if r["state"] in CLOSED_STATES]
print("design — the finding register\n")
print(" QUEUE (open findings)")
for r in sorted(queue, key=lambda r: r["raised"]):
age = (today - datetime.date.fromisoformat(r["raised"])).days
mark = "repro" if reproduced(r, root) else " - "
print(f" {mark} {r['id']:<4} {r['kind']:<16} {r['state']:<9} {age:>4}d {r['owner']}")
if notes:
print("\n NOTES (not reportable — GameDesign §3.1)")
for r in sorted(notes, key=lambda r: r["raised"]):
age = (today - datetime.date.fromisoformat(r["raised"])).days
flag = " EXPIRED" if age > NOTE_EXPIRY_DAYS else ""
print(f" {r['id']:<4} {r['kind']:<16} {age:>4}d{flag}")
# ---- §4 metrics. The log is reported on request, not by default,
# because a default view mixing open and closed loses the queue
# property (ADR-0012 D5).
n = len(findings)
repro = sum(1 for r in findings if reproduced(r, root))
debt = [r for r in queue if not reproduced(r, root)]
expired = [r for r in notes
if (today - datetime.date.fromisoformat(r["raised"])).days > NOTE_EXPIRY_DAYS]
unresolved_green = [r for r in queue if reproduced(r, root) and alarming(r, root)]
print(f"\n findings {n} (+{len(notes)} note(s))")
print(f" with a resolving reproduction {repro}/{n}"
f"{'' if not n else f' = {100*repro//n}%'} target 100%")
print(f" open, lacking a reproduction {len(debt)} target 0"
+ (f" [{', '.join(r['id'] for r in debt)}]" if debt else ""))
print(f" reproductions green while open {len(unresolved_green)} target 0"
+ (" <-- ALARM, GameDesign §1.3" if unresolved_green else ""))
print(f" notes past {NOTE_EXPIRY_DAYS} days {len(expired)} target 0")
print(f" closed (log) {len(closed)}"
f" [{', '.join(r['id'] for r in closed)}]" if closed else "")
bad = [r for r in rows if r["kind"] not in KINDS]
if bad:
print(f"\n UNKNOWN KIND: {', '.join(r['id'] for r in bad)}"
" — a sixth kind means the taxonomy was invented (ADR-0012 D4)")
return 0
def self_test():
ok = True
def check(name, cond, detail=""):
nonlocal ok
ok = ok and bool(cond)
print(f" {'ok ' if cond else 'FAIL'} {name}{' ' + detail if detail else ''}")
rows = parse(open(REGISTER).read())
check("the register parses", len(rows) >= 14, f"{len(rows)} row(s)")
check("every kind is one of the five",
all(r["kind"] in KINDS for r in rows),
"a sixth kind means the taxonomy was invented")
check("every state is known",
all(r["state"] in OPEN_STATES | CLOSED_STATES | {NOTE} for r in rows))
# THE control design-baseline.py did not have. Its YES-control passed a
# path that had been deleted and still returned True.
check("a nonexistent reproduction does NOT count",
not reproduced({"repro": "scenarios/ground/gr-e01-threshold-unreachable-2p.yaml"}),
"this exact path was deleted by 2da19a4 and the old tool said yes")
check("a real reproduction DOES count",
reproduced({"repro": "scenarios/ground/gr-p05-solve-legality.yaml"}))
check("a named test counts", reproduced({"repro": "games_ground::view::tests::a_spectator_sees_no_hands"}))
check("an em-dash does not count", not reproduced({"repro": ""}))
# The distinction the backfill discovered: a green DEFAULT is expected,
# a green COUNTEREXAMPLE is the alarm. Without this the report cried
# wolf over U2, whose scenario is green precisely because the
# provisional default it documents is implemented.
check("a green default does not alarm",
not alarming({"role": "default", "repro": "scenarios/ground/gr-d01-darvo-trigger.yaml"}))
check("role is recorded for every reproduced row",
all(r["role"] in ("counterexample", "default")
for r in rows if r["repro"] not in ("", "-", "")))
CB-WP-0026: collect the rulings -- ten answers that arrived and were never applied ground-game ruled all ten U-items on 2026-08-03, every one CONFIRMED as the default clay-borg simulates, and confirmed five of six provisional scenarios. clay-borg never collected the answers: CB-RES-0007 reported "0 of 10 ruled" the same day, and CB-WP-0022 built the finding register two days later still recording them as `reported`. make design's first run is what noticed -- not a human, not the adversarial review that found four other things. That is the unread-inbox failure running in the opposite direction, and it appears nowhere in the declaration, survey, ADR or spec of the pass that was built entirely around the forward version. It is arguably worse: an unread message is visible as silence, while a collected-but-unapplied ruling looks exactly like work in progress. Ten rulings quoted into §Underdetermined (the three conditional ones verbatim -- U1's designer note, U2's End-only trigger, U8's consume-only-if-it-cancels). Five provisional flags lifted, replaced by ruled/ruled_by/ruled_note so the flag went and the provenance stayed. Register queue 9 -> 0. T02's control came back clean: make sim is 26 passed, 59 rules covered, nothing red. Had a scenario gone red it would have meant we described our own behaviour incorrectly to ground-game. I wrote two U-item mappings and both were wrong. gr-a04 -> U1 (it asserts consent is REQUIRED; U1 asks WHEN the target accepts) and gr-d05 -> U5 (it exercises the UNREJECTED Reverse; U5 is the rejected one). Both plausible from covers:, neither survived reading the description. Third and fourth instance of this defect; the first two reached ground-game. So encodes_u_item is now a declaration and design.py asserts the file names what it claims -- and that check's own first version grepped for mentions and went red when two files recorded why they do NOT encode U1 and U5. A mention is not a claim, which is exactly the looseness that let "six of the ten have provisional scenarios" stand. Two positive controls went red for the best possible reason, both broken the same way -- asserting against live repo data instead of constructing their condition. rule-coverage.py required at least one provisional item to EXIST; it now builds a fixture and reports the live count as a diagnostic, because there is no number of provisional items this project should have. design-baseline.py pinned "2 of 6" while recomputing one row from a live glob, so the dated snapshot was never a snapshot; frozen to its 2026-08-03 list and unwired from self-tests, since per ADR-0012 D8 it is no longer a reporting tool. ScenarioFile is deny_unknown_fields and refused the four new fields until declared -- correct: a corpus accepting unknown metadata would let a typo'd encodes_u_iem sit there claiming nothing. DEVIATION: ADR-0012 D2 said "no new file". GroundRules.md crossed the loadability limit, so the register moved to specs/FindingRegister.md. D2's substance holds -- one register, same machinery, nothing competing -- but the literal instruction did not, and it resolves an awkwardness D2 named itself. make all: exit 0. loop-lint clean. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-05 16:13:37 +02:00
# CB-WP-0026 T03. A scenario may claim a U-item only if it names it,
# because two plausible mappings were written and both were wrong on
# reading what the scenario actually exercises (gr-a04 asserts consent
# is required, not U1's timing; gr-d05 is the unrejected REVERSE, not
# U5's rejected one). A claim nobody can check is how three wrong
# premises reached ground-game.
import glob
for path in glob.glob(os.path.join(ROOT, "scenarios", "ground", "*.yaml")):
body = open(path).read()
m = re.search(r"^encodes_u_item:\s*(\S+)", body, re.M)
if m:
check(f"{os.path.basename(path)} names the U-item it claims",
re.search(rf"\b{m.group(1)}\b", body) is not None, m.group(1))
# The measured claim: exactly one U-item has a scenario encoding it.
#
# **Asserted on the DECLARATION, not on a mention.** The first version
# grepped for `\bU<n>\b` and went red the moment two scenarios recorded
# *why they do not* encode U1 and U5 — a mention is not a claim, and a
# measurement that cannot tell them apart is the loose proxy that let
# "six of the ten have provisional scenarios" stand unchallenged.
claimed = sorted({re.search(r"^encodes_u_item:\s*(\S+)", open(p).read(), re.M).group(1)
for p in glob.glob(os.path.join(ROOT, "scenarios", "ground", "*.yaml"))
if re.search(r"^encodes_u_item:\s*\S+", open(p).read(), re.M)})
check("exactly one U-item is encoded by a scenario", claimed == ["U2"], f"{claimed}")
CB-WP-0022 T05/T06/T07: the register, and what its first run found T05. tools/design.py, make design, and the register in GroundRules.md -- 14 rows, no new file, because ADR-0012 D2 made §Underdetermined the register rather than building one beside it. Backfill was the test and it caught two things the ADR did not have. First, a `role` column. The first report alarmed on U2 and was wrong to: U2's scenario is green BECAUSE the provisional default it documents is implemented, which says nothing about whether ground-game agrees. GR-E01's was a counterexample that went green. Same colour, opposite meaning -- a register that cannot tell them apart either alarms constantly or never. Only a green counterexample alarms. Folded back into GameDesign §1.3. Second, it contradicted the survey. CB-RES-0007 said "six of the ten already have provisional scenarios." Measured -- grep -lE "\bU<n>\b" over scenarios/ground -- exactly ONE U-item names itself. Five provisional scenarios exist and four probably encode U-item defaults, but the mapping is not written down, so it is not checkable. Same defect class as the wrong premises, found inside the survey that proposed the fix. Now a reported debt: open, lacking a reproduction: 9, target 0. design.py carries the control design-baseline.py never had, asserted directly: a row citing a nonexistent file must not count as reproduced, using the exact path 2da19a4 deleted -- which the old tool called green. design-baseline.py is marked superseded rather than deleted; it is the evidence for how a wrong number got into a survey. T06. The report is a FILE in ground-game under GROUND-WP-0002, committed there, with a hub message that only points at it. It asks for no ruling: it carries GR-E01's withdrawal, our reproduction debt, and two notes that are explicitly not findings. And it had to acknowledge something nobody anticipated. GROUND-WP-0002 is finished -- all ten U-items were RULED 2026-08-03, every one confirmed as the default we simulate, plus five of six provisional scenarios. The survey said "0 of 10 ruled" two days later and this register was built saying `reported`. That is the unread-inbox failure running in the opposite direction: they answered and we did not collect it. The instrument's first run surfaced it. They are `ruled`, not `applied` -- lifting the now-settled provisional flags is owed and is not done, and make design shows them open until it is. T07. evidence/CB-EV-0021. Two of six catches in this pass came from execution rather than process (the role distinction from building it, the ten uncollected rulings from running it), which is InnerLoop §Design goal's prediction holding. make self-tests, facts-check, loop-lint: clean. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-05 15:22:33 +02:00
print("design self-test (positive control)")
return 0 if ok else 1
if __name__ == "__main__":
sys.exit(self_test() if "--self-test" in sys.argv else report())