You both have very good points here, but once I get finished with both of the 90% programming times, and everything seems to finally work with no more bugs (and it's true), then for my heavy industry work I look forward to spending 10X as much effort testing compared to coding.

Oh yeah, especially if the domain is complex, trying to envision how it can fail is as fun of a puzzle as trying to make it correct.

> including writing them all by hand, and push the “this isn’t fun, I’d rather do something else” bits to a minion.

That’s not really the part I’m talking about. My gut says that if tests are a blocker for weekend projects, people just don’t bother writing them. I certainly wouldn’t imagine them taking much longer to code than the core functionality.

In my experience, which seems to resonate with a lot of people, AI quickly stands up really useful boilerplate and very convenient purpose-built scaffolding… but is a lot less useful helping you solve actual problems in a way that makes sense to people that have those problems. Especially if you’re using a less-mainstream language or some other component.

I wasn't thinking about tests, but yes those help for iteration. Deployment, version management, debugging things like compilation errors or package version mismatch, etc.

I've done enough of this in my career that I can do it, but I don't want to. It's not fun. Having an LLM iterate on that stuff is really nice. Especially to get a scaffolding that I can then read and learn from.

For example, I recently was using an eInk screen for the first time and wanting to stream the currently playing album art to it. Claude made it so, so, so much easier to troubleshoot.

> Deployment, version management, debugging things like compilation errors or package version mismatch, etc.

I was using tests as an example. The tasks you refer to, as well as tests, are big time sucks in professional environments, or for established ongoing projects. I really, really, really doubt debugging, ops, and housekeeping/administrivia keep people from finishing weekend “theoretically should be simple!” projects.

My experience is that the AI is surprisingly good in providing solutions which fall in the last 20%. Examples are annoying ui challenges like proper placements or flow challenges. Or caching strategies. Or certain algorithm optimizations.

Having a design background, I find it to be a hindrance to quickly creating work that meets industry standards for things like user flow, or communicating state or purpose of UI elements through placement. I’d likely find it quite useful doing low-level bare metal stuff because I don’t have much experience doing that, but my output would probably be substandard from an expert’s standpoint.

It’s really great at making us think we’re better and more productive than we really are.

i dunno. the last 80% keeps being 80% and it just eats through tokens