HNHacker News
TopNewBestAskShowJobs

momiforgot

2 karma · joined March 28, 2025

submissionscomments
momiforgot··on How can England possibly be running out of water?
England’s water crisis: decades of underinvestment, privatised firms extracting rents, and a state too broke to fix pipes. Fortunately, the establishment has a brilliant plan: escalate with Russia, trigger a limited nuclear exchange, and get a neat lattice of new reservoirs in the form of fresh craters - conveniently clustered around the Home Counties. As an act of war, NIMBY rules won’t apply.

So next time you think war spending is fiscally insane, remember: it’s not waste, it’s water strategy. World-class ingenuity from the people who once ran an empire.

momiforgot··on Where's the shovelware? Why AI coding claims don't add up
2000 developers mocked Dreamviewer: "Ability to generate 10k LOC of crap HTML in 30 minutes is not productivity".
momiforgot··on Where's the shovelware? Why AI coding claims don't add up
LLM-powered shovelware sits in the same box as coke-induced business ideas. Both give you the dopamine rush of being “on top of it” until the magic wears off and you’re scrubbing your apartment floor with a toothbrush at 4 AM, or stuck debugging a DB migration that Claude Code has been mangling for five hours straight.
momiforgot··on I genuinely don't understand why some people are still bullish about LLMs
It's a common and fair criticism. LLM-based products promise to save time, but for many complex, day-to-day tasks - adding a feature to a 50M LOC codebase, writing a magazine-quality article, properly summarizing 5 SEC filings - they often don't. They require careful re-validation, and once you find a few obvious issues, trust erodes and the whole thing gets tossed.

This isn't a technology problem, it's a product problem - and one that may not be solvable with better models alone.

Another issue: people communicate uncertainty naturally. We say "maybe", "it seems", "I'm not sure, but...". LLMs suppress that entirely, for structural reasons. The output sounds confident and polished, which warps perception - especially when the content is wrong.