Making Sense of the Hype [Wading Through AI - Episode 2]
Also worth saying: people having luck with a 1500 line script or app that solves one problem on one machine without any consideration to privacy, security, or performance is profoundly different than shipping and supporting a product over time. We’re definitely in a phase where people are gonna miss the difference and that’s unfortunate.
500 lines of python to go archive a website like this isn’t novel. It’s not technically impressive. But it *is* fiddly. And I wouldn’t have bothered without LLM codegen.
My old World of Warcraft guild was using some cursed PHPBB-alike where every URL is a php file with half a dozen query strings. Every page has like 100 valid urls and all of them get used. /viewtopic?id=foo and /forum/viewtopic?id=bar&gid=baz are I shit you not the same page.
Related prediction: security and ops is gonna be a massive fucking mess for a while. Whole classes of scraping, mitm-ing, and drm stripping tasks have gone from “too much of a faff” to reasonably attainable.
So many little 200-5000 line scripts and apps I’ve always had gnawing at the back of my mind that I’ve whipped up in the last month or two. Spinning up a little one off tool to solve your own problem is so much more feasible than it used to be.
Is there really no off the shelf marching ants selection component?
I'm not totally worthless yet.
Louie Mantia, Jr.
