Oracle isolated it: "standard mode still having heat and leaks only occurred in
steam not in solo. Sauron and Lynx did not have the issue but Conn Man, Rajel
and myself did." The night-9 logs close the case:
affected (Oracle/ConnMan/Rajel): [exp] experience=2 heatModelOn=1
clean (Sauron's std sessions): [exp] experience=1 heatModelOn=0
CAUSE: the glass front end's JOIN layout trimmed the menu to "the mech list +
the JOIN button; everything else is the operator's call" -- and the experience
selector's hidden default is 2 (veteran). So HOSTS got whatever they picked
(Sauron hosted the lobbies all night -> his standard landed), while every
JOINER launched as veteran regardless of anything they did -- which is why
Oracle's cycling-the-settings experiment changed nothing, and why solo (full
menu) never showed the problem. heat >= veteran, hence "standard but heating".
DESIGN CONFIRMED BY THE USER before fixing: experience is PER-PLAYER by the
original design -- the sysop set each user's tier and mixed-experience matches
were legal. Architecture already supports it end to end: each node's master
player reads its OWN egg's experience (btMission+0xE4 -> BTPlayer @004c0bc8),
so per-node choice IS per-player choice.
FIX: the join layout gains the GroupExperience selector (the selection->egg
write path is the pre-existing host path, field-proven). Also: the front end
now prints "[fe] pilot experience=<x> (join|host/solo)" at launch, so every
future field log answers this class of report without asking anyone.
VERIFICATION BOUNDARY, stated plainly: builds both configs; the write path is
shared with the host flow which the field already exercises; the visible
selector + the [fe] line need one join-mode launch to eyeball, and the field
re-test is Oracle running a standard steam match as a JOINER and seeing no
heat.
Tooling note recorded in test-harness.md: bash-heredoc python collapses one
backslash level even single-quoted -- a "\n" arrives as a real newline and
replaces silently no-op. Build backslashes from bytes([92]); verify replaces
by length delta, not by the script saying "fixed". (This burned four edit
rounds tonight and several earlier C2001 hunts.)
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
User mandate (2026-08-02): "do tests like this from now on" -- after the #110
grind bench, where the field composition (peer fire destroying an arm over the
wire) replaced the constructed proxy that had let #86 be called fixed while
players kept hitting it.
The topic carries two halves on purpose:
DOCTRINE -- what counts as VERIFIED:
1. reproduce the REPORTED scenario, not a convenient adjacent one
2. scalpel hooks (BT_SELF_DAMAGE_ZONE / BT_KILL_SUBSYS / BT_FORCE_*) locate
defects; they support a "fixed" claim only with proven path-identity to
the field composition -- and the field composition still gets run
3. MP symptom -> two-node proof (master-side correctness says nothing about
what a peer sees)
4. visual symptom -> pixel proof (gotcha 23)
5. coverage claims need the axis enumerated and measured (all gates, all
chassis), because per-chassis behaviour lives in authored data
6. an unexplained extra effect in a passing run means the run has not passed
MACHINERY -- the bench_common.sh contract (summarized, file = source of truth),
single-node and two-node skeletons (relay, ports, affinity, fire cadence,
GOTO_STOP standoff), process hygiene (stale-node taskkill first, never
double-background, teardown kill order, stale-exe tells), and log-reading rules
(capped diagnostics are not evidence of absence; alarm lines are not trends;
name the actor at every refusal; field logs have no gates set -- spawn-time
summaries ungated, per-frame traces gated).
Routed: Quick Lookup row, CLAUDE.md reasoning step 4, build-and-run parity
section, reconstruction-method Key Relationships. checkctx CLEAN.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
