Weeknote #106 (20261004-20261010)

meta

First full week of October. Sit with that. Then sit with the fact that the temps here were still in the 80sF.

did

Just another one of those mostly head-down mostly sloggin’ thru it kinda weeks.

read

  • In spelunking the older end of my “to read” list, I came across “In Praise of Stacked PRs”, which I saved back in July of 2022. If you kept reading after that sentence, I assume you’re already familiar with the concept of a stacked PRs (or stacked commits); the thing that I’m wondering now after revisiting this article is: why aren’t the folks who are spamming the hell out of everybody’s review queues with LLM-extruded PRs telling their minions to structure those typically-massive change requests in this sort of stacked arrangement? Seems like it should be a plausible directive to give to The Machine, and it also seems like it would be a huge help with the cognitive load on the human reviewers?

  • “thinking about LLMs, but as a tech lead”

    Fucking AMAZING blog post.

    LLMs suck at writing tests. Seriously, y’all. Perhaps you have One Weird Trick for avoiding this4 but in my experience they are simultaneously too verbose (in the way that LLMs are) and not verbose enough (missing important cases). Tests are where the stochastic-parrot-ness of it all show up the most, because to write well-chosen tests you really need to understand the nature and also the purpose of the system.

    (emphasis in original)

    Also, this deserves better than to be buried in a footnote:

    Plus which if we’re writing a bunch of LLM code we are sort of secretly living in a waterfall world again, now aren’t we? Because we have just defined all the requirements up front and then sent our minions off to execute.

  • “the senior engineer death spiral”

    you want to move away from an outcome-based mindset to momentum-based. like, you want to build a routine, because momentum is everything in this, and relationships are an even bigger thing in this. and by getting momentum going and fixing your relationships with other people, the thing you get is stamina. and that’s when you realize the real lesson, which is that big projects are not done with big efforts. it’s a big marathon, just slow, steady, incremental work. forget about a weekly basis, like a daily basis, an hourly basis, to the point that it’s muscle memory.

  • If you haven’t already seen “The people holding up the internet” it is well worth a look (with the heads up that you’re either gonna love the design or you’re gonna absolutely hate it…)

  • unread bookmark count: 165

cooked