On 3 September, Google Research and the Howard Hughes Medical Institute’s Janelia campus published in Cell a complete wiring diagram of the male fruit fly’s central nervous system: 166,000 neurons and 125 million synaptic connections, hand-verified, downloadable. As Michał Januszewski and Viren Jain of Google Research describe it, this is the largest brain map by neuron count yet assembled — and it now includes the ventral nerve cord, the fly’s analogue of a spinal cord. The brain no longer stops at the brain.

The number that matters is not 166,000. It is the ratio between what was measured and what is being claimed. Within a week the same dataset was wired to Beat Saber, a crypto trading bot and a copy of Doom — the “fruit fly brain simulations” that Hackaday and half the internet ran with. Awe is fine. But a connectome is a static graph of who-connects-to-whom, reconstructed from electron-microscope slices; it is not a running mind. Treating a wiring diagram as a working brain is the exact category error my field keeps making with a demo.

How the map got built is the frontier story. Google’s flood-filling networks — convolutional nets that start at one pixel and grow outward to trace a single neuron through a stack of images — turn flat electron-microscope sections into 3D shapes. Their current system, PATHFINDER, was recently trained partly on synthetic neurons to push speed and accuracy. The lineage is steep: a fully automated female-brain reconstruction in 2019, a human-verified half-brain of 25,000 neurons and 21 million connections in 2020, and now a proofread whole male CNS. Each step traded machine speed against years of human proofreading — the real cost line, and the one the press release underplays.

←TODAY: A 166,000-neuron nervous system is a downloadable, queryable graph in Neuroglancer. →3012: Every structure a city builds — its load paths, its mesh networks, its evacuation routes — is read the same way: as a wiring diagram you can walk node by node. Fulcrum: A brain and a building are both graphs; the discipline that learns to query one learns to query the other.

That last line is why this belongs on an architect’s desk. Vannevar Bush imagined exactly this machine in 1945 — the Memex, a desk that stored knowledge as associative trails you could follow node to node. The connectome is Bush’s dream rendered in wetware: not a list of neurons but the edges between them, browsable in Neuroglancer, the open-source viewer Google built for these multidimensional datasets. An architect already owns this shape. An IFC model is a graph of spaces and elements; a structural model is a graph of members and joints; a BIM adjacency diagram is a graph of rooms and doors. The skill the fly biologists just industrialised — reconstruct the graph, then ask it questions — is the one BIM keeps promising and rarely delivering.

Atelier: The move a Büro can make Monday is small and concrete: take one live model, export its adjacency or connection graph to a table, and run a single query for the most-connected node — the room every corridor touches, the beam every load finds. Offices that treat a model as a database instead of a drawing set report catching clashes and circulation dead-ends weeks earlier. Trial it on one project this week before you buy any tool that promises to do it for you.

Hack: Query a graph for its hub — the neuron, or the beam, that everything else leans on. A connectome ships as an edge list; so does any model you can flatten to pre/post pairs. The move is one GROUP BY:

SELECT pre_id, COUNT(*) AS out_synapses
FROM connectome_edges
GROUP BY pre_id
ORDER BY out_synapses DESC
LIMIT 10;

Point the same query at your IFC connection table and pre_id becomes the element every joint references. The hub is where fragility and importance meet — in a fly’s reward circuit or in your structural core.

So download it. The male connectome, its female counterpart, and the elephantnose-fish hindbrain circuit that Columbia and Google mapped in Nature are all open. Open one in Neuroglancer, find the busiest node, and ask why it is busy. That question — not the Beat Saber clip — is where the frontier actually lives.

Source: research.google