On Thursday the third of September, a team at Janelia published the complete wiring diagram of an adult male fruit fly's central nervous system in Cell. Janelia's FlyEM group, the University of Cambridge, the MRC Laboratory of Molecular Biology and Google Research. Just over 166,000 neurons. Around 11,700 cell types. Released under a licence that lets anybody download the lot.

It is the kind of thing people spend careers on. Decades of imaging, tracing, proofreading and annotation, to produce the most complete map of a working nervous system anybody has ever made.

By Friday teatime somebody had it running inside Minecraft.

the fly brain, in minecraft

— evnsnclr (@evnsnclr) September 4, 2026

Thirteen point nine million views. His own note says he built it with Astra.

Then it got properly out of hand.

The week in fly

Saturday night, a Coinbase engineer called Alex Wormuth wired the connectome into a Doom arena.

DOOMFLY

— Alex Wormuth (@nftechie_) September 5, 2026

Sunday morning, at two thirty five in the morning, somebody rendered Bad Apple on it. Twenty million views for a black and white animation played back through a dead insect's wiring diagram.

bad apple!! on the fly brain

— Kevin (@linguinelabs) September 6, 2026

Monday, a compiler engineer called Jessica Paquette got it playing Super Mario 64. Her GitHub note says the code is, and I am quoting, one hundred percent vibe coded with GPT Astra.

super mario 64, on a fly brain

— Jessica Paquette (@barrelshifter) September 7, 2026

Tuesday night, the one that set the format. Twenty two and a half million views, eighty five thousand likes.

the fly brain can play beat saber

— lyra (@_lyraaaa_) September 8, 2026

By Thursday it had a hundred dollars on Coinbase and a car to park.

STONKFLY

— Alex Wormuth (@nftechie_) September 10, 2026

the fly brain can parallel park*

— Breg Grockman (@alright_mark) September 10, 2026

The asterisk in that last caption is doing an enormous amount of work.

The fly, at time of writing, is up one dollar on Bitcoin. Which puts it ahead of several funds I could name.

The thing I actually love about this

The people building the stunts are more careful than the people reporting them.

The Doom project ships its own neuroscience review, in its own repository, and that document is blunt: the claims that a fly understands Doom, or that the fly is learning to play, are not supported. The author wrote the debunk of his own viral post and committed it next to the code.

The Beat Saber clip did twenty two million views. Its author corrected it herself, within hours, saying it was so far just a motor pattern overfitted to a replay.

Meanwhile several outlets ran the Doom story with the wrong date, and a YouTube video with three hundred thousand views calls it a fly's conscious brain, which is the single claim every scientist anywhere near this work explicitly denies.

So the hobbyists published their limitations and the media published the hype. Sit with that for a second.

What Janelia actually released

This is the bit worth being precise about, because it is where nearly all the confusion lives.

What came out on the third of September is a map. Images, annotations, synapses, skeletons, predicted neurotransmitters. A graph of what connects to what.

There are no weights. There is no simulator. Nothing in the download runs.

Every fly-brain-plays-something you saw this week is somebody's own simulation, built on top of a wiring table, with the missing parts filled in by hand. And the missing parts are not small. No plasticity, so it cannot learn. No neuromodulation. Neurotransmitters predicted rather than measured. Synaptic strengths approximated from counts. The mapping between brain and body chosen by whoever was building the demo, on a Saturday, for a laugh.

A connectome is anatomy. Getting behaviour out of anatomy is still an open research problem, and the people who wrote the map say so.

There is also a runnable fly brain, and it is older than any of this. A team at Berkeley published one in Nature in October 2024, built on the female fly connectome, 139,255 neurons and around fifty million synapses. It correctly predicted how a fly responds to taste and how it grooms itself. That is a genuinely astonishing result and it got a fraction of the attention that a dead fly parking a virtual car got this week.

My favourite bit of pedantry: the male map is the second complete central nervous system connectome this year. A Harvard and Princeton team published the female one in June. The fly memes are three months late and nobody noticed.

Why I think this is the most human thing that happened this week

I have been banging on for a while that the real problem now is not capability, it is digestion.

Our species takes years to absorb what a new capability means. We are getting a dozen a month. So things land, and we nod, and we do not have time to work out what we think, and the feeling arrives months after the fact if it arrives at all.

And I think the memes are what that looks like from the inside.

Confronted with the most complete map of a nervous system ever made, a thing that genuinely raises the question of what a mind is and whether one could be copied, the internet's response was to plug it into Doom within twenty four hours. Not out of stupidity. I think it is the opposite. It is what people do with something too big to hold. You poke it. You make it silly. You find out what it does when you give it a steering wheel.

It is the same instinct as a kid taking a radio apart. Play is how we metabolise things we cannot yet think about.

And then, right on cue, on Wednesday morning somebody announced fruit fly heaven, because if you run a thousand copies of something's nervous system and then close the laptop, a joke is the only available response to a question nobody wants to ask properly.

announcing fruit fly heaven

— Macroblock (@sainimatic) September 9, 2026

The serious version of that question is being asked by about four people. One of them runs a whole-brain emulation company, and his answer is the honest one: we do not know what its experience is. Nobody does.

I do not think a wiring diagram of a fly with no plasticity and guessed-at chemistry is experiencing anything. I am fairly confident about that. What I notice is that "fairly confident" is doing a lot of load-bearing work in that sentence, and that in about four years the same sentence is going to be much harder to say.

So

A landmark piece of science came out. A very large number of people made it play video games. The stunt builders were rigorous about what they had not proved and the coverage was not. And underneath all of it is a question about minds that we are collectively answering with a shrug and a Doom mod.

Honestly, I find the whole thing quite cheering. Not the accuracy of the reporting. The instinct to play with it.

Also the fly is up a dollar.

Anyway. That's where I'm sitting on this.

Si