Rulez777--FramePicoTracker/docs/SUPERSESSION.md

Corrections that must survive the handoff

Historical claim	Evidence-backed correction / unresolved limit
October8 archive is current	Later diagnostic10 activation, mode proof and optical work supersede its current-state summary; keep its earlier experiments.
All tracker IDs were permanently known	Earlier stock capture245/249 observed three identities. Later protected installed branch admits only the positively bound first tracker; do not transplant historical slots/bonds into fresh admission.
FW189 product version proves stock/custom image	Product version remains189 in modifications; require actual fresh capability/schema and image provenance.
Mode0/1 or native25 IR OFF/ON are the same state	Operating mode and IR mode are separate; distinguish native storage and measured effects.
Any native10 ACK means a dark LED frame	Recorded native10 brackets disproved it; transport acceptance is not applied phase or physical darkness.
Only negative native10 corrections invoke the common adjustment helper	Later signed-path review shows both signs outside the small correction window converge; distinguish the specific skipped-pulse path from common helper reachability.
Native command16 is missing	Native configuration path exists; [N,8] and host[N,2] have distinct apply-countdown semantics.
TIMER0 ticks are0.5us	Verified BITMODE/PRESCALER interpretation gives1us for the audited build.
RF gaps uniquely prove tracker listen duty cycle or interference	Station grants, native sample selection and host retention also matter. Correlation is not unique causality.
6000us grid or controller-safe gap alone proves IMU delivery	Native producer state/upstream gates, sample selection and sensor admission remain separate.
Empty queue/new nonce/reconnect establishes fresh OTA state	Stale pending-worker ACK races were reproduced; fresh startup and exact source attribution are independent.
OTA transfer completion proves activation	189 transfer and191 activation are separate evidence. Diagnostic10 is later installed; do not repeat old OTA because an old manifest says not activated.
Isolated generated manifest booleans override later measurement	Preserve the original manifest's offline approval flags but label later bounded hardware evidence separately.
Current copied release C is executed original210	Use the exact210 source and executed273 ELF hash; copied three-peer C differs.
Candidate11 short producer is the installed image	It remains an offline candidate; diagnostic10 remains protected.
All native16 ACKs must wait for finalN verification	This deadlocked mode preparation;392 repairs the ACK/query ordering and393-v2 supplies bounded hardware proof.
391 checking-mode replies already prove applied mode	Nonce correlation was not persisted;393-v2 is the completed fresh proof.
No real IMU or optical measurements were ever obtained	Historical235/236 obtained actual sparse IMU/optical pilot data. Latest radio-only393-v2 intentionally had none. Neither proves stable current fusion/FBT.
A6byte suffix is another gyro sample	It is the native magnetometer auxiliary field.
Default sample period4.976480ms fits the admitted diagnostic10 stream	Recorded counter fit measured4.93997760834ms for the specific target/mode. This does not establish physical time offset or axes.
Low covariance resolves all root/correspondence ambiguity	Competing roots and association errors can remain; do not smooth them away.
A model fit is tracker identity or published6DoF	Requires causal target binding, current clocks/head pose, finite age/covariance and publication admission.
Requested60fps or nominal37us describes actual profile3 capture	Actual retained profile3 upper exposure was about782.267us at30fps; verify metadata on each invocation.
Passive probe is still pending	900 saved pairs completed, zero fits, short exposure and insufficient detected markers; profile3 follow-up remains prepared only.
Short-exposure zero markers proves no photons	It proves sampled detector counts; contrast approximation is not native contour/identity proof.
Fast triangulation helper timing is full pipeline latency	Helper-only and full decode/detection/association/world/publication timings have different denominators.
Every phase number is a unique implementation or successful experiment	Some are reviews, prepared candidates, repeated metadata or abandoned paths. Use actual source/hash/results.

The historical ledger deliberately retains outdated assertions as historical evidence. This table and CURRENT_STATE determine current disposition, not the most confident historical prose or highest folder number.