AI definitely loves to over-engineer things but I’ve had lots of success with the latest models just telling them not to. Also I do simplification passes afterwards telling them to find anything that can be simplified or YAGNI.

I guess the tend to over-engineer is probably because the benchmarks don’t care about simplicity and maintainability? Who knows.

I watched a bit of it but it seems like just a random discussion of memory use intermixed with abstruse Haskell mentions.

He doesn’t even say what his thesis is at the start. What is this video trying to tell me?

Maybe a year ago. Not today.

And yet, the latest generation of models is surprisingly underwhelming.

Hard disagree. The latest generation (Astra 6/Opus 5.5) is the first one that doesn’t produce slop IMO. They basically fixed Claudish, the code quality (at least in Rust) is close to what I would write most of the time, they don’t spew a gazillion useless comments through the code, etc.

I don’t think you can talk about agentic coding without saying which models you’re talking about - there’s a huge difference between the latest frontier models and those from 6 months ago.

I totally agree. Whenever I’ve criticised Ruby, e.g. for it’s insane level of hard to follow “magic”, people just say “that’s just Ruby on Rails… not all Ruby is like that”. I have no idea why it’s popular or why we should listen to anyone that thinks Ruby is amazing and Rust is terrible. It shows an astounding level of poor taste.

What if we combined taxis with bloom filters? Doing this would allow users to request taxis as needed. Using a bloom filter as a phone number.

It is not known how much AI can be improved

I didn’t say it was. But at this point it would be a pretty wild prediction to think it’s not going to continue to rapidly improve.

You said

It’s better than you right now, but every mid-level tech worker will tell you it sucks. You have significantly higher capacity than it, you just have to learn to get there and that isn’t easy or quick.

I was just pointing out that this view completely depends on AI improving slower than he does. I would not put money on that.

Yeah the problem is will he learn faster than AI improves? I wouldn’t want to be in his shoes.