Code was our medium for thought
wattenberger.com · 1,073 words · saved by 7 readers
Code was not just the output—it was where we explored problems, made decisions, and figured out what we wanted.
What should we build? I'll build a beeswarm of Notable AI Models releases over time. First I'll pull the Epoch AI dataset details and inspect the existing chart patterns. Explored 2 files · 4 searches There's already a related release chart. I'll rebuild the opening around publication date and use Epoch's Notable AI Models CSV as the source of truth. Fetched notable_ai_models.csv Recent releases are dense, so I'll keep publication date exact on the x-axis and use vertical position only to pack overlapping marks. Ran collision-safe beeswarm layout I'll keep the result quiet: restrained…
saved by
related reading
- Shipping at Inference-Speed | Peter Steinbergersteipete.me
- claude code is not making your product betterethanding.substack.com
- Notes on the Software Factorybenedict.dev
- Writing Code vs. Shipping Code: Productivity Effects Across Generations of AI Coding Toolsmuratbuffalo.blogspot.com
- Thoughts on slowing the fuck downmariozechner.at
- Why We Built Our Own Background Agentbuilders.ramp.com
- Don’t Outsource Your Thinkingteltam.github.io
- Now – Updates from the Linear teamlinear.app
- You Still Need to Thinkcalv.info
- How I'm using coding agents in September, 2025 — Massively Parallel Procrastinationblog.fsck.com
- SWE-chat: Coding Agent Interactions From Real Users in the Wildarxiv.org
- After Automation | Everyevery.to