METHOD
How a wiring diagram becomes a weight.
WHAT THIS IS
NEMIQ runs the published connectome of Caenorhabditis elegans, 302 neurons and their measured synapses, as a network on screen. Market data is fed in as sensory input. The network fires. Its firing pattern proposes how a USDG treasury is allocated across tokenized stock tokens, inside hard limits the network cannot touch.
No living animal is involved. What you see is the published wiring diagram, not a live specimen and not calcium imaging. The worm is a graph. Nomi is a name for the graph.
The network does not trade and it does not predict. A measured connectome, driven by market input, proposes weights within fixed limits. That is the whole claim.
SIGNAL → WEIGHT
Each binding below was chosen by a human because the biology reads well. None of it has been shown to beat a naive baseline. Treat the mapping as an argument, not a result.
| circuit | what it does in the worm | what it does here |
|---|---|---|
| VB01 → VB11 | forward locomotion: the undulation wave travels head to tail | each segment owns one basket asset. As the wave passes a segment, a fixed packet of capital moves into that asset and on to the next. |
| AVA / AVD / AVE | backward escape command interneurons | escape state. Proposals during escape are refused. The packet returns to the USDG reserve and the wave runs in reverse. |
| ASH | polymodal nociceptor: noxious touch, osmotic and chemical stimuli | driven by realised volatility. Above threshold it excites AVA/AVD; the avoidance reflex is real wiring, not a rule we wrote. |
| HSN | hermaphrodite specific neuron, egg-laying circuit | a distribution epoch. When HSN fires, the treasury pays out. Who is owed what is computed by the contract, not by the network. |
| DD / VD | GABAergic inhibitory motor neurons, contralateral inhibition | position-size brake. When they fire, every proposed shift is halved before the fence sees it. |
| AVB / PVC | forward command interneurons | gate the wave. When forward command is silent the allocation holds. |
THE FENCE
These limits are fixed in the contract. The network cannot read, write, or vote on them. Every proposal passes through the same clamp function the site runs in your browser; a proposal that had to be clamped is recorded as a rejection in the journal.
ONE TICK
- The client asks /api/tick every two seconds. The tick number is derived from wall-clock time since genesis, so every viewer sees the same tick.
- Sensory input for the tick is a volatility scalar. In this build it is a deterministic pseudo-simulation seeded by the tick number; the real feed is a TODO in the route.
- ASH fires if volatility crosses its threshold. If ASH fires, AVA/AVD/AVE fire with it and the tick is an escape tick.
- Otherwise the segment under the wave proposes a shift of a fixed packet into its asset from the neighbouring segment. DD/VD may halve it.
- The proposal is clamped. If any limit was hit, the journal shows the proposal in red with the limit that refused it. If it fit, the allocation moves.
CONNECTOME DATA
Neuron identities, classes and body positions follow White, Southgate, Thomson & Brenner (1986), The structure of the nervous system of the nematode Caenorhabditis elegans, Phil. Trans. R. Soc. B 314. Synaptic counts follow Cook et al. (2019), Whole-animal connectomes of both Caenorhabditis elegans sexes, Nature 571, as distributed by the OpenWorm project.
This build ships a committed sample: the 302 neuron identities are complete, the command, locomotion, nociceptive, egg-laying and inhibitory circuits are wired from the published tables, and the remaining edges are a seeded structural fill so the site builds without the raw dataset. The fill is marked in the data file. Replace it with the full Cook 2019 edge list via scripts/build-connectome.ts.
Layout is a deterministic force-directed pass computed once at build time with the ventral cord motor neurons pinned along the body axis. No simulation runs in the browser.
WHAT WE DO NOT CLAIM
No yield is promised. No return is implied. The mapping from firing pattern to weight was designed by people and has not been tested against a naive equal-weight or hold baseline. It may be worse.
The network computes nothing about who is owed what. It proposes weights inside a fenced range. Nothing more.