Post

Less than 9 months ago. So there's maybe two major frontier releases before we reach 12? And we are pretty much there already.

I never take these predictions to be applicable to the existing public releases and workflows. I always assume when these guys stay stuff like this, it's about what they are seeing on the horizon from their own internal models and testing, stuff that we won't see until next year.

This is a trick question, Tim😉

only software that runs trains, obvs

We have automated the looms many times over in the past. The codon table the first time...

This was hilarious! Good choices

He hedged by saying “most, maybe all”. It’s at most still, but definitely not all. The proof is that the number of engineers really hasn’t shrunk much at all.

If that was true I would have to repeat myself.. but hey Grifting Works.

Definitely most, and definitely not all. I think the definition of "Software Engineer" is going to keep shifting so we never hit that "all" state, at least until we have fully independent autonomous machine intelligence participating in every day society, which is probably at least 10 years away.

I’d agree with most being correct, but that last 10-20% is the part that invokes 2+ humans talking.

Yes and also I still do all of the parts of the job I like

yes he absolutely is, depending on how you define software engineering. software engineering is dead, long live software engineering

To be fair he never said "velocity will go up" or "software will get better"

Id lay that on the feet of MS, a place where they probably right now figure out for the first time that they could probably use agents to reduce the memory footprint of Teams.

More right than wrong. It’s crazy out here

Yes... But only in a way that is very very much no. It can do all of the parts of the profession. I don't think it can do the profession itself. Yet. Very much yet. Heh. Maybe to like... half depth, if that makes sense?

yes

Mostly right imo. Anyone I talk to who either develops or manages developers has completely changed their workflow and writes little to no code by hand. Shifted in March / April and focus now is on figuring out their optimal stack and workflow.

Nah, I’m definitely still doing the engineering, and Opus 5.5 still makes errors. *My role isn’t “swe” but we still hire plenty of swes , and i don’t see that stopping in 2 months

Oh yea, for sure. I mean in replacement of an experienced senior engineer not someone off the street. Even then that gap has closed a ton with Opus 5.5

yeah, its just that a lot of people haven't actually automated it. it definitely is possible though

for my small company, yes. we haven't gone that far (and won't, for security reasons) but only by choice. if we were building a less serious product, absolutely true

What's happened in the past two years is maybe a little more complicated than "engineers have become editors" but I'd say the ability to read LLM-generated code and have an opinion about whether it's ready to ship is definitely a core skill now.

I use Claude at work but I still write most code by hand and still do A TON of other things by hand. Our systems are very old and very complicated (30+ systems interacting). Starting a project in a fresh or modern environement is not the same as an old one.

yes but you need to have unlimited access to mythos, fable, and opus

Ah, if only. 🌈

Pretty fucking close I'd say