the sensor build log.

This is a build log for something that doesn't exist yet. At our in-person nights, the lens will get an optional companion instrument — one that reads something other than light. I'm publishing the plan before I've bought a single part, for the same reason the math is public: so you can check the work before it ships, not after.

Everything in the other lab notes reads light off a camera. This one is different. It's the piece I'm most tempted to over-promise, so it's the piece I most want on the record early.

The instrument.

The whole rig is deliberately boring, cheap, and honest:

GSR is galvanic skin response — a measure of electrodermal activity. Your sweat glands are wired directly to the sympathetic nervous system, and they respond to arousal a beat before you're consciously aware of it. More activity lowers your skin's electrical resistance; the sensor reads that as a changing voltage. It's real physiology, in real units. The same signal a polygraph leans on, minus the pseudoscience about what it "proves."

The read.

The plan is a short capture, tightly scoped. While the camera takes your portrait, the kiosk samples a few seconds — call it 3.5 — of conductance. It normalizes that against the session's own range rather than some absolute scale, because baseline skin conductance varies enormously between people and even between hours of the same day. Then it maps the normalized signal across the seven bands: activated leans warm, calm leans cool.

The most important line in the whole design is the fallback. No sensor connected? The read quietly returns to the optics. The camera engine is the source of truth; the sensor is a garnish. Nothing about the experience breaks if the hardware is missing, flaky, or left at home. That constraint keeps me honest — the sensor has to earn its place, it can't become load-bearing.

"It's a biometric signal translated to color. A playful instrument, not a biofield reader, and not a diagnosis."

The boundary.

Here's the part I want to say plainly, in writing, before anyone's standing at a table hearing me say it out loud. This sensor measures electrodermal activity and turns it into a color. That's it. It is not reading an energy field, an aura in the metaphysical sense, or your health. If I wanted to, I could slap "biofield reader" on the kiosk and no one at a party would fact-check me. It isn't one, so I won't.

There's also a hard technical limit worth stating: the browser API that reads the sensor, Web Serial, is Chromium-only. So the kiosk is a laptop or an Android tablet. An iPhone physically cannot read it. Rather than pretend otherwise, we'll just say so at the table.

Why not a ring or a watch?

Everyone asks. The honest answer is that the wearables people already own mostly can't do this, and I'd rather explain why than fake a smooth integration:

Apple Watch — its heart data lives behind HealthKit, which is native-app-only. A website can't touch it. No amount of wanting changes that.

Smart rings — the data is real, but it syncs to the cloud on a delay of minutes to hours. Useless for a read that happens in three seconds at a table.

The one that works — a Bluetooth heart-rate chest strap, over Web Bluetooth, gives you live beat-to-beat data in the browser today. If we ever add a wearable path, it'll be that one — and it'll be written up here first, before it appears at a single event.

What "done" looks like.

Buy the parts. Get raw serial data onto the screen. Watch what my own conductance actually does across a calm minute and a startled one. Tune the band mapping against that, not against a guess. Run one night with it as a pure add-on while the camera stays the real read. Then write the follow-up — including whatever I got wrong in this post.

That's the deal with the lab-notes track: the plan goes up before the build, so the gap between what I said and what happened is visible. If you build hardware and you can already see a flaw in this — the normalization, the 3.5-second window, the band mapping — tell me before I spend the $45.

— Ashley