Wet and dry: the electrode question

Every argument about EEG hardware eventually lands on one interface: the boundary between metal and skin. It is the least glamorous component in the stack and the one that decides what the entire downstream pipeline gets to work with. The choice between wet and dry electrodes is usually presented as convenience versus quality. The honest version is starker: it is a choice about how many hours of brain you can collect, and hours are the currency the whole model is bought with.

What the gel actually does

Skin's outer layer is a poor conductor, so the electrode must couple to the ionic currents underneath through some bridge. Conductive gel is that bridge. Paired with silver-silver-chloride electrodes, a chemistry chosen because it exchanges charge reversibly and therefore adds almost no noise of its own, gel brings the contact impedance down into the single-digit kilo-ohm range. Low impedance means the tiny neural voltage arrives at the amplifier intact instead of being eaten by contact resistance, motion becomes a small wobble instead of a spike, and the recording session can run for hours. The costs are procedural: skin preparation, syringes of gel, careful placement, twenty minutes of setup, and a wash of the hair afterwards. It is a clinic ritual, and every minute of it is a reason a person does not put the cap on.

The best signal in the world loses to the cap people actually wear.

What dry trades away

Dry electrodes, metal, conductive polymer, or newer materials pressed against skin or through hair, skip the ritual at a price: impedances ten to a hundred times higher and unstable, wobbling with pressure, hair, and small movements. The engineering countermeasure is the active electrode, a tiny amplifier mounted at the contact point that buffers the signal before it travels down the wire, converting a fragile high-impedance pickup into a robust low-impedance output. Done well, dry systems record acceptable EEG in the slow and middle bands and deteriorate most in the fast bands, the speech-relevant ones, precisely where the skull already taxes the signal. Saline-based semi-dry systems sit in between, wicking saltwater from a reservoir, buying hours of decent contact without full gel preparation.

Why this is a strategy question

Recall the arithmetic from note five: representation quality scales with diverse hours, and hours scale with wearability. A cap that takes twenty minutes of clinic ritual will collect the hours of a study; a cap that pulls on like a hat collects the hours of a life. The wet route maximizes the quality of each hour, the dry route maximizes the count of them, and the model at the far end is hungry for both. That is why our hardware decision is genuinely open and genuinely important: it is not a parts choice, it is a statement about whose hours we will have, at what fidelity, and how fast they compound. When the decision lands, it lands here, and we will publish the reasoning with the numbers.