Three phases, one input. A star system carries four per-player masks and three of
them agree on nearly every system of every save the corpus holds, so a model built
on the wrong one looks right until it does not. The gate is the DERIVED
active-presence mask -- fleet-here OR gate-here OR owner, recomputed on every
arrival and departure -- not the sticky one and not the explored one.
S29 SystemObservedStamp the system's own last-observed turn (whole function)
T17 RebuildPlayerViewTree the per-(system, player) observation record: who saw
the system, on what turn, and what encounter was there
T21 UpdateSurveyAndStats the explored sweep: seen this turn implies surveyed
game/sim/visibility is pure and knows nothing about save shapes; app/visibility_phase
wires it to them. The mask is READ FROM THE SAVE and never rebuilt: neither reference
pair moves a mask leaf, so the loaded value is the value these phases would see, and
rebuilding it from an unmodelled movement pass would be a change with no evidence.
Measured, closed and regressed reported separately and never netted:
turn1-state -> turn2-state 209 -> 158 closed 51, regressed 0
turn2-state -> turn3-state 108 -> 87 closed 21, regressed 0
of which this lane closed 46 and 16 (the rest were already closed at main). The 46
are the brief's 32-leaf target in full -- 8 record counts, 8 player ids, 8 turn
stamps, 8 encounter ids -- plus 8 system stamps and 6 explored masks.
Three further pairs the model was never fitted to, all zero regressions:
human-turn2 -> human-turn3 353 -> 311 closed 42 (a different game, 21 systems)
zuul15 -> zuul16 276 -> 264 closed 12
zuul16 -> zuul17 341 -> 329 closed 12
The corpus's one discriminating row is a host test rather than a comment: a system
whose last visiting fleet has gone carries the sticky and explored bits set, the
active bit clear, and a stamp frozen a turn behind. The test asserts the freeze AND
asserts what the wrong gate would have produced, so a future edit that swaps the
mask fails loudly instead of quietly agreeing with five saves.
Labelled hypothesis, with the workload named in the header: the encounter id is
recovered from the encounter fleet at the system, because the field the original
reads is set once at map generation and is not on the wire. It agrees on all six
encounter fleets in the corpus and no save can separate it -- none kills an
encounter while leaving its system visible.
Not written, deliberately: the colony-ownership stamp that moves beside these.
Its gate is demonstrably NOT the active mask (one system in the corpus has a zero
mask and moves it anyway), the formula is not held, so it is reported, not written.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ARBgSooAfokKUy6wKUKEyZ
Reads the whole colony output chain off the instruction stream (every range
disassembled to the next function start) and compares two of its functions
against the running game.
The population -> output law is linear and is carried by the executable:
output points per head are typeOutputModifier x 1.8 / 500000, and the
three-row population-type table is built in code rather than loaded, so the
imperial (1.0) and civilian (0.33f) modifiers are facts about the binary.
A system's total output is a SUM of three terms, not one multiplicative
chain. The station bonus scales only the imperial term and morale only the
civilian one, so OutputModifiers no longer carries either; they belong to
GroupOutputInputs. The function previously described as the base-output term
is the over-harvest RESOURCE demand, and it is corrected in place.
Live on VM140, both hooks in compare mode over two species and two workloads:
GroupOutput 13,105 calls / 0 divergences; ComputeTotalOutput 11,252 calls /
1 divergence of one ulp, in a value its caller rounds to an integer. Both
functions declare a whole-object Guard: 0 undeclared writes in 24,357 calls,
which is what makes the side-effect-free claim a measurement.
sim::Narrow forces the double rounding a 32-bit x87 build otherwise skips;
without it every civilian row came out one ulp low.
Also fixes ComputeBankruptcyLimits' elimination divisor, which was the
decimal -0.15 rather than the image's widened float -0.15000000596046448.
The two disagree for every maximum income divisible by 3 and for essentially
every empire above ~3,000,000.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ARBgSooAfokKUy6wKUKEyZ
Pure module game/sim/techgraph: PrereqsMet (AND of ORs, empty group fails),
SetResearched (stamps, child-cost sweep, sticky turnAvailable, zero-cost
recursion) and the newly-available collector, all read out of 0x00581e10,
0x0057d8e0 and 0x00587cc3.
Wired into the B3 hook in compare mode only, over the scratch node copies:
four more node write-backs, the EVENT_TECHS_UNLOCKED list (still an input,
still nullptr when it could not be computed), the de-duplicating observed-tech
append and the one RNG word RollResearchEvent draws.
docs/U-unlock.md section 4 is the prediction, written before the build was staged.
The behavioural compare found 8 of 45 StrategyServer::MoveFleet calls diverging by
one ULP on a position component. Read off the instruction stream, the cause is that
the engine's vector normalise narrows to float32 four separate times and we kept
everything in double:
delta.c = f32(dest.c - pos.c) stored back to a float32 slot before normalising
sumsq = f32(x*x + y*y + z*z) products/adds in 53-bit regs, only the SUM stored
len = f32(sqrt(sumsq))
inv = f32(1.0 / len) a reciprocal, MULTIPLIED through, not three divides
dir.c = f32(delta.c * inv)
and the same call returns the leg distance, so it is never recomputed in a wider
precision either. The position tail was already right, which is why the error was a
constant absolute ~1.2e-7 (half an ULP of the inputs) rather than a formula error.
Adds NormalizeVec3 / StraightLeg / StraightLegDistance / AdvanceAlongUnitDirection
and rebuilds AdvanceAlongDirection on them; the movement hook now takes both the
direction and the distance from one StraightLeg call, as the original does. The
arrival test is an exact float compare, so the distance has to be that same float32.
Tests pin float32 BIT PATTERNS, not tolerances: one case per narrowing plus four
independent legs component by component. A CHECK_NEAR would pass against the old
arithmetic.
sim::Distance is left in double on purpose and flagged at its declaration: it now
serves only the node-line/stutter geometry, which very likely needs the same
treatment but has zero behavioural coverage to correct it against.
Live, same VM/save/workload, run twice by this lane:
control recap-7584bad-20260908T0615Z 45 calls, 45 compared, 8 diverged, exit 1
fixed mf-45bdf7d-dirty-20260908T0721Z 45 calls, 45 compared, 0 diverged, exit 0
with identical arguments, identical pos.before and identical ORIGINAL pos.after on
all 45 calls. The control reproduced the eight divergent call_ids exactly.
Coverage unchanged and still thin: all 15 moving calls are the same straight-run
waypoint type; types 2-5 were attempted and could not be reached (the only player
that would travel a node line has no ships on this save). See docs/M-movefleet.md.
ctest 32/32; tools/clean_room_check.sh OK.
