Brian Armstrong: publicly funded science should work like open source — every experiment a repo
alishbaimran_ · x · 2026-11-09
Coinbase CEO Brian Armstrong argues publicly funded science should work like open source software. Today a result is just a PDF: data isn't shared, code isn't public, and few studies get replicated. He envisions every experiment as a versioned repo — data, code, and protocol — that anyone can fork, rerun, and patch, with funders rewarding replications as much as new papers.
AI makes this urgent: agents are producing more discoveries than ever, shifting the bottleneck to replication and validation; commercialization still requires moving atoms and regulator meetings. Reposter @alishbaimran adds that lack of replication has long held biology back, making reproducibility infrastructure critical in the agent era.
More from AGI Musings
- Not Every Problem Needs Agentic AI: Teams Pay 10-30x More for the Wrong Choice — DavidLinthicum · 2026-11-09
- Cambridge's David Krueger on MIRI: the problems are hard — what matters is the capabilities-to-safety research ratio — DavidSKrueger · 2026-11-09
- Applying the bitter lesson to my LLM workflow: skip agents, just talk to the model — esrtweet · 2026-11-09
- Hyper-ideas may no longer matter to us: the graver question of machine creativity — SoloGen · 2026-11-09
- Machines may reach ideas strictly better than humans', argues researcher on AI creativity — SoloGen · 2026-11-09
- We've been automating idea generation for a long time, researcher notes on singularity — SoloGen · 2026-11-09