ESP32 30-hour debugging saga

what happened

INMP441 I2S mic on ESP32. spent 30 hours debugging why audio data wasn't reaching the server. tried everything — rewiring, different libraries, different boards. the bug: setFollowRedirects was silently dropping the POST body on a Google redirect, returning 400. fixed ONE LINE after 30 hours.

also: mug warmer circuit took 6 hours — dead transistor AND dead LED AND dead capacitor, all at once. three failed components masquerading as one bug.

why it's a gotcha

when you're debugging hardware + software, the number of possible failure points is enormous. the real killer isn't the bug — it's the wrong assumption that keeps you looking in the wrong place. you assume it's the wiring, so you spend 10 hours on wiring. it's actually an HTTP library silently eating your data. ai-rot can make this worse — when you're deep in an AI loop, you keep prompting the same wrong assumption for hours instead of stepping back.

the fix

lesson from an Orbit mentor: "if nothing is working, you have a wrong assumption. aggressively backtrack and sanity check." when you've been stuck for more than an hour, stop and list every assumption you're making. test each one independently. the bug is almost always hiding behind an assumption you haven't questioned — and that includes environmental assumptions. also: when debugging hardware, test each component individually before assuming the circuit design is wrong.