Anthropic May Know A Lot About Dice, But Do They Understand The Game?

Anthropic, the AI company behind Claude, has demonstrated a lack of understanding of software engineering, as evidenced by issues on its Claude status page and Claude Code GitHub issues page. Despite its expertise in large language models, the company is repeating familiar product trajectory mistakes, prioritizing intelligence over fundamentals and discipline, which may threaten its long-term viability.

You’d naturally be skeptical of advice on how to throw dice from a leading casino dice manufacturer, even though they obviously know a lot about dice. You’d be perhaps even more skeptical if it turned out that their track record at the craps table was worse than average. And yet you listen to advice on software engineering with AI from Anthropic? Do they know a lot about LLMs? Of course. Does that make them better at using them in developing software? Evidently not. What a craps player needs to know about dice to improve their odds is likely very little – a bit of high school maths. What they know about craps is probably the differentiator. Anthropic have demonstrated quite unambiguously – check out the Claude status page https://status.claude.com/ and the Claude Code issues page https://github.com/anthropics/claude-code/issues for the receipts – that they don’t really understand software engineering. They’re speed-running product trajectories that are very familiar to some of us. Like so many software companies staffed by very smart people who do very smart stuff, in the long term they’ll have to do their growing up in public and start focusing more on fundamentals and discipline than smarts – if there’s going to be a long term. Anthropic may understand the dice better than most, but they don’t seem to understand the game.