Five pieces, each a specific instance from the build. Two are about the multiplier; three
are about its limits. That ratio is the honest proportion, not a rhetorical hedge.
Building the product before the hardware existed
In draft
A simulated hardware layer, and the week the textbook-correct choice finally became the fast one.
The bug that was quietly erasing the recordings
In draft
A dead clock chip, and the difference between tracing a failure and suspecting one.
The click that wasn't where it sounded like it was
In draft
An afternoon spent searching the wrong domain, and the thing that pulled me out of it.
What voltage should be on this pin?
In draft
Bench debugging when the reference layer is instant. Nobody automates the meter.
Tuning, and talking to chips that don't explain themselves
In draft
The unglamorous majority of the work, and the part that compressed the most.
Decide: list the five unpublished pieces as above, or show only what's published. Listing them signals a body of work; it also promises five things. Per 2b-1's shelf test, leaning toward the spine post alone at launch.