Experiment log · $BBYA
In one line
input64-electrode activity from Cortical Labs' simulator, 3 windows × 30 sprocesscount spikes → letters (ticker) · raw frames → 64 × 64 picture (logo) · wait for a burst → launchoutput$BBYA + logo, live on PONS · every window hashed (SHA-256 of the raw frames)
Files & links
Watch
The run, replayed from its own data (×8)Rate trace, the three windows, stim pulses, the burst that fired, hashes as they were produced.
The tool itself — terminal recording of a shadow run (×4)Same program, same simulator, recorded from the terminal. A shadow run: different ticker ($KVANF), nothing launched — it shows how a ticker and logo come out.
Population rate · the run
shaded = decision windows (30 s) · ticks = stim bursts · dashed = mean + 2.5σ · T = scheduled time · ● = the burst that fired
Decision windows
Trigger
Output
Event log
What you are looking at
The full run that produced $BBYA, on Cortical Labs' simulator. The simulator plays back a stored electrode recording and runs the same program the real chip runs. Three 30-second windows: the first picked the ticker from the most active electrodes, the second became the logo, the third armed the launch. After the set time (T), the first burst — the moment the activity jumped far above its usual rate — was the launch. Each window's hash is the SHA-256 of its raw frames; the proof is the hash of those three hashes.
How we used neurons for this run
Right now: not living ones. Cortical Labs ships a simulator with their SDK for people waiting on a chip. It plays back a stored recording of electrode data — 64 channels, 25,000 samples a second, the same format the real chip produces — through the same API (
cl.open(), spikes, stimulation calls). Our program cannot tell the difference; it reads spikes, counts them per electrode, hashes the raw frames, sends its pulses. The only things the simulator cannot do are respond to those pulses (it is a recording, not tissue) and produce something new every window (it loops every 60 seconds — that is why window 1 and window 3 have the same hash). So the ticker and the logo came from real electrode data through the real pipeline, but nothing was alive during this run. We say that everywhere.
How this compares to FLY
FLY used the FlyWire connectome: a wiring map of one fruit fly's brain, made by slicing a dead fly into thin sections and reconstructing about 140,000 neurons. It is a map — fixed, finished, the same for everyone who downloads it. Anything "the fly" does is a computer simulation on that map, and no signal ever goes back into a living thing.
CORTICAL runs on a Cortical Labs CL1: about 800,000 living human neurons on a 64-electrode chip. The tissue is alive, its activity changes minute to minute, and the program stimulates it back — a closed loop, which is what Cortical Labs built the chip for. Three differences that matter:
- Alive vs. a map. A connectome is a photograph of wiring. The CL1 is the wiring, firing, now.
- Two-way. We send pulses to the electrodes that fired most and read what comes back. FLY's simulation cannot be poked.
- Checkable. Every window is a recording you can download; the hash of its raw frames is the decision. Re-run the code, get the same ticker and logo. Nothing here depends on trusting us.
Today we are on the simulator, so on point 1 we are honestly level with FLY: a computer playing back data. Points 2 and 3 already hold — the pipeline is the real one. The day the CL1 instance is live, point 1 flips, and this page shows the same experiment on living neurons, streamed as it happens. That is what the launch pays for.
Why it matters
Every memecoin has a name and a picture that somebody typed in. This one doesn't. The ticker and the logo came out of a program reading neural activity, and every step can be checked: take the recording, run the same code, get the same result. The simulator is honest about being a simulator — it repeats itself every 60 seconds, which is why windows 1 and 3 hash the same. The launch exists to pay for the real chip. When it is live, this page shows the same experiment on about 800,000 living human neurons.