Scanning a brain (a separate problem) only produces a connectome file; getting it into a digital substrate means pushing that data down a real, physical channel — and every real channel has a hard capacity limit set by the Shannon–Hartley theorem. This simulator treats the connectome as a fixed payload of bits and streams it from the biological brain (left) to the digital substrate (right) at exactly that limit, so you can see why "we have the data" and "the copy arrived intact" are separated by physics, not just engineering effort.
C = B · log2(1 + SNR) (bit/s, Shannon–Hartley)
payload = N · bits/neuron ÷ compression
fidelity ceiling = compressionTerm(compression) × noiseTerm(SNR)
- Channel bandwidth (B) — the raw frequency width of the transfer link; capacity scales linearly with it.
- SNR — signal power over noise power, in dB. Capacity only grows with log2 of it, so doubling SNR never doubles throughput — a diminishing return baked into the formula itself.
- Lossy compression — shrinks the payload so it crosses the channel faster, but each neuron's state is reconstructed less exactly; the fidelity ceiling drops as compression rises.
- Connectome detail — how many neurons the payload represents; more neurons means more total bits to push through the same channel.
- Each neuron is assigned a fixed random draw at scan time. When its turn in the bitstream arrives, it is copied clean if that draw falls under the current fidelity ceiling, or arrives corrupted (flickering red) otherwise — so a low-SNR or heavily-compressed run can finish "on time" while still delivering a substrate riddled with errors.
Real-world relevance: even with a perfect scan, moving an exabyte-scale connectome through any physical link runs straight into Shannon's limit — bandwidth and noise floor, not ambition, set the ceiling on how faithful a transferred mind can be.