About this experiment

The question

Bitcoin changes the world around one persistent computational fruit fly. The fly’s wiring is taken from the published adult Drosophila connectome (FlyWire FAFB, materialization 783). We watch whether the resulting behaviour sits toward BUY, SELL or NEUTRAL. Nothing is bought or sold.

What is real?

Neuron identities, FlyWire root IDs, connectivity, synaptic counts, anatomical annotations, cell classes/types, neurotransmitter predictions, sensory populations and descending/behavioural pathways come from published connectome data (Dorkenwald et al. 2024; Schlegel et al. 2024). Every neuron in the running experiment is a real root ID. Every edge is a counted connection from the 783 edgelist.

What is simulated?

Membrane / activity dynamics, thresholds, leak, adaptation, noise, Bitcoin-to-environment encoding, the virtual terrarium, motor decoding, fly movement, and the BUY/SELL/NEUTRAL interpretation are computational. This is not a perfect electrical simulation of a living fly brain. It is a computational neural-activity model operating over real Drosophila connectome connectivity.

What does BUY / SELL mean?

After the fly moves, we look at where it spends time in the enclosure. Persistent motion toward the BUY side is labelled BUY; toward SELL, SELL; local indecision remains NEUTRAL. That label is an interpretation of simulated behaviour. It is not a financial recommendation and not an executed trade. The fly never receives the words BUY or SELL.

Whole brain?

The public FAFB 783 release contains 139,248 annotated neurons and 3,532,517 directed connections at the ≥5-synapse threshold used here. Existing whole-brain LIF simulators (WebGPU / CUDA / Metal) need a GPU and hundreds of megabytes. This experiment runs continuously in the browser on CPU, so it uses a biologically selected sensory→descending subgraph: the actual neuron and connection counts shown in the live panel. The primary graph is never shuffled; a shuffled control exists only on the Control page.

What this project refuses to do

• No paper trading, portfolios, wallets, orders, leverage or P&L.
• No conventional trading algorithm sitting in front of the fly.
• No invented neuron IDs and no substitute artificial neural network.
• No fitting the model to future Bitcoin prices.
• No resetting the fly every candle.

Pipeline

Raw BTC stream → rolling FAST+SLOW aggregator → virtual environment → real sensory neurons → real connectome → computational activity → real descending pathways → motor decoder → fly movement → BUY / SELL / NEUTRAL observation