Lab notes
A record of
changing my mind.
Not a raw changelog. Short notes about what I tested, what it taught me, and what changed in the product thinking.
64 checks. Not 64 features.
The beta list is mostly closure: make the existing paths dependable, truthful and boring in the best possible way.
Three tracks in. Many more to map.
Tsukuba, BB Raceway and High Speed Ring have V1 Track Lab maps. The focus is shifting from new features to beta readiness and track coverage.
Track Lab is becoming part of the normal LapMind loop
Saved laps can now open directly from Sessions — and Tsukuba is becoming the first visible example of the richer track-specific map direction.
The lap change is evidence before it is a line
Boundary Foundation v1 has passed its automated gates. Physical GT7 validation comes before learned start/finish geometry.
Why Discord uses your own private bot
The private-bot route is intentional architecture: no LapMind voice backend, no LapMind account, and no per-user voice cloud.
Track Lab, Corner Lab, and the line between evidence and product
Track Lab is experimental. Corner Lab is research-backed and planned. Neither gets promoted by wishful thinking.
The candidate is not the release
A historical checkpoint on why automated proof and hands-on approval are kept separate.
Why Race Finished became a barrier, not just a sentence
Finish is a state transition. Everything queued behind it is history.
What a Discord reconnect should—and should not—remember
The optional voice path can recover without replaying a stale race.
Why a reversible Reject is safer than an immediate delete
Session history should let a driver change their mind.
Why the first complete lap should win over a fragment
Track Lab chooses a complete live lap deterministically.
What kerbs taught us about careful surface evidence
A surface code becomes useful through context, not colour alone.