3D fruit fly on macOS with FlyWire connectome
Archived — this story has rotated out of today’s deck. It is kept here in full.
The gist
A 3D fruit fly on macOS uses a 668-neuron FlyWire connectome simulation to walk, groom.
It shows how real neural circuits can drive behavior, making neuroscience interactive on your.
Background
The project, DesktopFly, is an open-source macOS app that simulates a fruit fly's brain using real connectome data from FlyWire (FAFB v783). It runs a 668-neuron leaky integrate-and-fire model with ~19,000 synaptic connections, including looming-detector neurons and the Giant Fiber escape command. The fly's escape is not scripted; it emerges from the neural simulation when your cursor approaches, triggering real synaptic pathways. The code is MIT licensed, with data under CC BY-NC 4.0.
How it unfolded
- recentlyDesktopFly was posted on GitHub, gaining 46 stars and 2 forks.
- Aug 18, 2026The project appeared on Hacker News, reaching 253 points and 94 comments within 10 hours.
- Aug 19, 2026It trended on Hacker News front page at position 27 with 226 points, and was shared on Reddit and other tech news sites.
Who’s saying what
- Public
- A Hacker News commenter appreciated the open source approach over startup's sensational claims, calling it more transparent.
- Expert
- Another commenter noted that with only 668 neurons, the simulation runs super fast, and they plan to extend it to multiple flies and add food.
Still unverified
The claim that the fly's escape behavior exactly matches a real fly's timing (~4 ms) is based on the simulation's design, but not independently verified.