A developer wired a full connectome simulation of a male fruit fly brain called MALECNS V1.0 to live Coinbase trading: 166,700 simulated neurons, about 25.6 million directed connections and 124.2 million synaptic contacts receive Bitcoin BTC/USDC candlestick charts converted into RGB inputs. A fixed decoding layer interprets the resulting spike patterns as buy, sell or hold signals and an automated Coinbase AgentKit places orders using $100 of real funds. Engineered reinforcement pulses stimulate designated dopamine neurons when trades register profit, and the code and dataset are published under an MIT license on GitHub.
The project, named STONKFLY, follows prior experiments that used fly connectomes for game-playing and simulated bodies; underlying mapping data comes from large-scale efforts including FlyWire and collaborators linked to Google. Documentation explicitly states no profitable learning or live-funded performance has been demonstrated and recommends baseline comparisons. The experiment exposes how connectome-based models are migrating from labs into non-biological, consumer-facing domains - complete with a public website and a speculative STONKFLY token - raising legal, ethical and regulatory questions that current frameworks do not yet address.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.