June 21, 2026
Curve drama goes flat
The Flat Curve Society
AI’s future gets locked up — and the comments immediately choose violence
TLDR: The article says the best AI may soon be restricted by governments, leaving the public with weaker tools while big players keep the strongest systems. Commenters were split between alarm and mockery, with critics calling the piece overhyped and open-source supporters insisting the underdogs will catch up anyway.
Steve Yegge’s big thesis in “The Flat Curve Society” is basically this: the smartest artificial intelligence may soon be treated like a nuclear secret, locked behind government controls, with the rest of us stuck using today’s “good enough” tools. In his telling, a model called Fable getting briefly shut down was the flashing red warning sign. Translation for normal humans: the wild era of ever-smarter public AI could be ending, and the next leap may happen behind closed doors.
But the real fireworks were in the reactions. One camp rolled its eyes so hard you could hear it through the screen. The meanest, funniest jab came from a commenter who dismissed the whole thing as a “long-winded screed” and mocked the supposed evidence with the brutal line: “The proof? His react client.” Ouch. Another reader said they were simply confused, accusing the article of treating “AI literacy” like it just means using AI a lot, and dunking on a Netflix example that seemed to show hype more than real results.
Then came the civil war over whether scary, powerful AI is actually new. One person deadpanned, “GPT-2 was already extremely dangerous,” which reads like either a warning or elite sarcasm depending on your mood. Meanwhile, open-source fans rushed in swinging, saying writing off community-built AI today sounds exactly like the smug old claims that Linux would never matter. The vibe? Half apocalypse, half eye-roll, with a side of “we’ve seen this movie before.”
Key Points
- •The article argues that advanced AI models have reached a dangerous stage, citing the temporary shutdown of Fable by the U.S. government as a visible sign.
- •It predicts that within two or three model generations, access to superintelligent AI will be controlled similarly to nuclear weapons.
- •The article says most companies, including many Fortune 500 firms, may have little or tightly supervised access to frontier AI systems.
- •It forecasts that access to the most capable models will be sold as a remote service on vendor-controlled infrastructure rather than broadly deployed inside companies.
- •The article contends that open-source models may reach Fable-class capability but are unlikely to surpass it significantly under future compute and supply-chain restrictions.