“You’re six months behind! You’re missing out!” No, buddy, I’m a cybersecurity consultant and not a moron

“You can’t have a take on this if you weren’t using the latest tools last week!” Buddy I _HAVE A DAY JOB_

I find that all of this makes me incredibly angry. Like, what the fuck are we doing here. Why are we doing this? Who are we doing it for?

I like Thomas but I’m really fucking sick of this

Also i dunno but it’s just really hard for me to be interested in trying the fancy expensive, high-investment versions of the tools when the free versions are all unfit for purpose. Like not just bad but wrong. There’s a bare minimum of purpose alignment that something has to meet for me to consider investing more deeply in it and ALL of the free stuff I encounter is so unfit as to be actively offensive

If LLMs really are great for coding, and only coding, the coding LLM people had better shoot the rest of the LLM community or something and take down their stuff because they are making y’all look evil

@kevinriggle I have a very hard time believing code is the only and unique place LLMs don't slop.

@quinn If they’ve finally integrated LLMs with linting and type checking and so on and so on I could see there being less slop at the word-by-word level… which still leaves lots of room for more subtle forms of slop at the method, subsystem, and interaction (e.g. not writing security vulnerabilities) levels!

@kevinriggle there's an potential paradox here, the better LLM code gets, the more closely you have to look at it, because the bugs and insanity get harder to find.

@quinn YUP! And the more practiced and experienced you need to be without the LLM tool in order to be able to see that!! Also unlike a compiler the LLM is nondeterministic so we can’t really build reliable abstractions on top of it