Field notes from the edge where human intuition meets machine augmentation. Every post is a star; the lines are the ideas they share. Wander the constellation and let the adjacent possible find you.
A cross-section to begin with, including the interactive pieces — the shapes of scientific discovery, the arithmetic of exploration, and the senses beyond the famous five.
A discovery is not one kind of event. It helps to place any breakthrough on two axes at once: which stage of inquiry it belongs to, and what it does to the surrounding framework. The stage runs from noticing a…
One caveat up front that matters for reading the table: "scout fraction" is not measured the same way across studies — some report a dedicated scout caste, others a trail-lapse rate (foragers that ignore an…
School teaches five senses: sight, hearing, smell, taste, and touch. The body actually runs many more sensory systems than that, monitoring both the outside world and its own internal state at all times.…
On Its Head You Can Only Manage What You Can’t Measure Inside the box thinking Why Optimization Is Often Sub-Optimal Simplicity Is Overrated Right-Brain Analytics The New LBO: Left-Brain Outsourcing The Return…
In a recent substack post ( the always insightful Rishad Tobaccowala presents "10 thoughts about AI, Humans and Work in 10 minutes". They are all worth pondering (although I would not invoke Suleyman in # 3)…
In March 2000, Guy Theraulaz and I made the cover of Scientific American for "Swarm Smarts", a sort of companion article to our book with Marco Dorigo, "Swarm Intelligence" ( We were elated, not just for…
The original “fitness beats truth” (FBT) theorems do not require an explicit assumption that “seeking truth is costly.” The key requirement, instead, is that evolution favors any perceptual or cognitive…
Open-endedness represents perhaps the most profound challenge in computational science: how do systems transcend their initial constraints to generate genuinely novel, complex, and meaningful structures? This…
I saw a post by @juergen schmidhuber recently ( that gave me pause. About convolutional neural networks and the neocognitron, a 1979-80 neural network architecture by the Japanese scientist Kunihiko Fukushima…
With all the healthy chatter about "verifiability" ignited by Andrej Karpathy, I want to point out that even if "verifying" (a program) can be hard, it is still easier than coming up with a good idea for a…
That's the title of a wonderful post by Rafael Irizarry ( The TL;DR is the post's byline: "UMAP is a powerful tool for exploratory data analysis, but without a clear understanding of how it works, it can…
Eli Lilly's Chorus unit, established in 2002, emerged as one of the most successful innovations in pharmaceutical R&D productivity, achieving 3-10x productivity improvements over traditional development models…
These are not silos. Pick a field, then follow the lines out of it into neighboring territory.
Machines that learn
29Working alongside AI
11Search, selection, design
25How living systems change
7Never running out of new
36Many parts, emergent wholes
28Where new things come from
154The machinery of the cell
73Turning science into medicine
16Minds, memory, and bias
15The bandwidth of experience
8How discovery actually works
The whole corpus, drawn as a force-directed map where an edge means two pieces share ideas. A note on foraging ants sits a short hop from one on search algorithms.
Open the atlas →One irregular dispatch when a new piece lands: notes on AI, evolution, complexity, and the biology of discovery. No noise.
Delivered with Buttondown · unsubscribe anytime