December 19, 2025
Follow the money, not the megaphone
Believe the Checkbook
AI brags it can code; the checkbook screams we need humans
TLDR: Anthropic bought Bun’s human team even though its AI was the top code contributor, showing companies value judgment over raw code. The community is torn between ‘jobs will shrink’ and ‘humans still matter,’ roasting execs for hype and memeing ‘revealed preferences’ as the real truth.
The internet spit out its coffee when Anthropic bought the Bun team—even though Anthropic’s own AI agent was the top coder on Bun’s open-source project. Translation for non-nerds: the code was free to copy, yet Anthropic still paid for the people. Commenters instantly dubbed it “believe the checkbook” energy. One wag, hapless, dropped the meme-y economist zinger: “revealed preferences.” In other words, ignore the keynote; watch where the money goes.
The strongest mood: judgment beats raw code volume. RandallBrown: if AI wrote all his code tomorrow, his output wouldn’t jump, because choosing what to build is the hard part. That set up the biggest clash of the thread: the “AI will shrink engineering jobs” crowd versus the “humans still run the show” camp. conductr played it cool with the “buggy whip” analogy—engineers won’t vanish, but demand may shrink enough to feel like it. Meanwhile, neilv called the “AI as force multiplier for high-judgment people” angle a clever pitch that flatters both AI boosters and craft-first vets.
And the jokes? jollyllama roasted execs: “Believe the checkbook? Why do that when I can get pump-faked into strip-mining my engineering org?” The drama is delicious: public hype says “engineering is over,” while Anthropic and Bun quietly prove the opposite—judgment is scarce, and cash follows scarcity.
Key Points
- •Anthropic’s AI agent submitted more merged pull requests to Bun’s GitHub repository than any human developer.
- •Despite Bun’s MIT-licensed code, Anthropic acquired the human team behind Bun rather than forking the code.
- •The article states the bottleneck in engineering with AI is human judgment, not code production volume.
- •Anthropic’s announcement emphasized the Bun team’s ability to rethink the JavaScript toolchain from first principles.
- •The article advises leaders to use AI to amplify high-judgment engineers, invest in judgment-centric skills, and maintain junior pipelines.