You know that guy who submits the 5000-line diffs for review and says "don't worry, it's fine"?
I ran into one of those today at work.
I "returned to queue" five such Christmas Eve diffs from him. I marked each, with increasing directness, with "make this a series of changes that I can understand"
Then it turns out he's gone off on PTO and probably won't be coming back. Gah.
@trochee that's terrible manners. ignore that shit and make him fix it or just say it wasn't adequately analyzed before he left
@hipsterelectron
🎯 that's exactly what I'm gonna do
"technically brilliant but bad at cooperation ritual" is such a huge red flag in the tech world
20% of my job is seeing what needs to be done. Another 20% is *doing* it.
The remaining 60% is "make it make sense to the people around me"
@trochee also the act of coding itself has not been the bottleneck for a long time at most companies. so widening that part of the pipeline has very negative consequences on everything downstream of it
@exchgr indeed. The incentives are properly aligned around the "20% doing it" — sorta
— but they're unaligned and often *perversely* aligned with the other 80%
So "10x coders" pop up all the time who are skipping the research and sense-making, and they're producing the appearance of progress (and making *more* work for everybody else)
@trochee 10x coders, for 10x the pain
@trochee I strongly suspect that most ‘rockstar’ programmers are ones who screw things up so badly that ten people invest time and effort in properly redoing the work that they were supposed to and they take credit for it.
@trochee Nailed the ratio there. And boy howdy, that 60% can be annoying af but, like, that’s the work right there.
@donaldball yeah i came up with that by realizing that it goes like:
Monday: figure out what needs to be done
Tuesday-Thursday: convince the people around you that it's a good idea, and break it into pieces small enough to swallow one at a time
Friday: actually write and ship code
oh, and another thing -- the giant diffs have all kinds of LLM-smell throughout, and the diff description is a summary of the work but with *no* "why" suggestions.
>:[ i cannot wait for the bubble to pop, even if it means I'm living on savings for a while and doing "fix my printer" consulting for non-profits
@trochee I know this, I hate this.
I had a guy put up a 2000 line PR making fundamental architectural changes to the codebase I maintained. The diffs grew as I gave him very pointed feedback on how none of this would fly.
The only way I could solve that problem was by quitting and letting the next poor sob handle that hot potato. RIP to my replacement.
uncanny, how did you know he was making major architectural changes to *exactly the things I was planning to work on in January* ?
we've all met that guy
@counterVariable
'sfunny , my first reaction was to gut-check *myself* because I was asked myself "did he just steal your thunder and you're knee-jerk rejecting because this was *your* problem"
this is probably why I was pretty gentle in the first two or three passes through, but then I started to realize that *nope* this is a plan that is moving forward without consideration of what else he might break, but just to get his own local thing fixed up.
@trochee the clown show kicker was that they didn't want to contend with my feedback until my manager also agreed with me. For they had a fancier title than me you see, that must mean that my thoughts have less weight.
I feel your pain. At least for me I didn't have the LLM nonsense yet. 🙃
