Opus 4.5 isn't perfect though: my afternoon was spent checking boneheaded mistakes that an intern shouldn't make. SWEs are coming up with checking and review patterns (my fav so far is @dollspace.gay's VDD), but it's far from a "set it and forget it" technology.
gist.github.com/dollspace-ga...

To me, this is the big insight: lots of things are code

My 3 year old loves Frozen, and in the second one Anna is crushed with grief (Elsa and Olaf are dead, spoiler, but they get better). Anna resolves that her only way forward is to "do the next right thing". It's not my favorite song but I've been thinking about it a lot recently.

I get paid to work with computers (admittedly not PCs), so I'm smaht but there's no way I would have done this on my own. I've never touched WinDbg before, and wouldn't have taken the hours to figure it out. For the right problem space, these tools are magic.

It pointed me to crash dumps, appeared to write parsers for them, walked me through WinDbg Preview steps to narrow things down. When it was time to take action, it gave instructions on disabling the offending module, and answered questions about the performance risks.

But I didn't want to spend hours reloading the machine only for it to be hardware, and Windows machines must keep crash logs, so I went to GPT 5.2 and left it on Auto. It took n messages, an hour, and a couple hundred megs of debug dumps but it found the culprit - some audio enhancement software.

My version has long been "capitalism is a misery machine". In its optimizations, it can't measure how enjoyable something is for worker or user; an insane trade that triples misery for a buck looks good on a spreadsheet. I'm also a capitalist, but you've got to know that it'll fall into that trap.

He also said something like "a good metric should be like a lighthouse. A North star for a time but if you keep going that way you'll end up in the rocks". I'm going to repeat that one.

It was an imperfect but interesting metric. It showed a clear negative trend over time, likely because of growing complexity of the code base and lagging tools. Then they made it an individual and team target, and it became worse than useless and killed morale.

Alex cited Goodhart's Law, which I'd never heard of but have seen firsthand. "When a measure becomes a target, it ceases to be a good measure". Uber used lines of code per engineer as a proxy for productivity (ugh, but okay).

en.wikipedia.org/wiki/Goodhar...

I like Mike's podcast with @komorama.bsky.social too, maybe more. One of the few podcasts that I wish was 3 hrs instead of 45 minutes.

www.techdirt.com/2025/12/05/t...

I can't vouch for the slides-over-rollerbag bit though

I have a Timbuk2, which I think should hit all those needs. Mine is kind of old, but this looks about right. Lots of organization, my only complaint is that at this point I want two water bottle holders. Definitely holds a Macbook Pro + iPad comfortably.
www.timbuk2.com/products/182...

How is this not "I will never \"explain my posts\""?

They get pretty mad if you try to park there overnight after you quit though

I'm partial to "I Shall Exterminate Everything Around Me That Restricts Me From Being the Master"

en.m.wikipedia.org/wiki/I_Shall...