@neilhenning That's not a real thing.
And seems to imply you think everyone using slop is inevitable.
@yora I think it depends on the progress this year and next but if you extrapolate out Claude’s 4.5 -> 4.6 progress then it is inevitable.
As someone who takes great pride in my code and being a coder this brings me no joy.
@neilhenning @yora I read that 4.6 uses around three times as many tokens ( i.e, electricity ) as 4.5... I also read elsewhere that claude is *massively* subsidised.
An interesting thread from users :
@neilhenning @yora heh i totally misunderstood. i thought the last stage of grief is you accept that gen ai sucks.
but i guess that would be more like sowing vs reaping.
@neilhenning I haven't been through those stages. Despite all the disruption this will cause, I recently landed on similar thoughts https://ghuntley.com/real
It's both exciting and daunting, but it really is a revolution, and I never thought I'd see this in my career. I'm trying to convince as many people in my company to embrace it. We have a few hardcore devs, but the rest are split between folks slowly adopting it and folks in stages 1-3 still refusing.

Hey folks, the last year I've been pondering about this and doing game theory around the discovery of Ralph, how good the models are getting and how that's going to intersect with society. What follows is a cold, stark write-up of how I think it's going to go down. The
@gustav @neilhenning I really wonder how this whole "create applications by writing a spec" will turn out in the long run, it sounds like a 1970's waterfall enthusiast's wet dream, but there's a reason that idea never quite worked.
I'm firmly on the "spec is code and code is spec" side of things, e.g. if your spec needs to be as detailed as the code implementing the spec you gain exactly nothing, because iterating on the spec will take just as much effort as interating on code.
@gustav @neilhenning it's a cycles thing, if I had 'believed the hype' in the 1990s (and would be 'carreer-driven') I would have gone full in on Java and databases which would have made for a very boring life ;)
Similar I hope that a younger me today would chose a different path than going full-in on LLMs. This sort of "industrialized software development" always sucked and will always suck.
@gustav @neilhenning ...maybe some young underground rebels will figure out absolutely breathtaking new ways to juggle AIs, similar to early punk, hiphop and electronic music used sampling, instead of using LLMs to generate the 1000th cookie-cutter CRUD application.
Not holding my breath though... but such a hypothetical counterculture would actually make me interested in the topic.
@morten_skaaning @neilhenning "Don't copy that floppy!" ;)
...I'm not in the AI camp, but I find it utterly fascinating how the table has turned on copyright and "intellectual property".