Benjamin Bouvier šŸ„

Wow. One of the most relevant blog posts I've read about AI coding for a long time: If you thought the speed of writing code was your problem, you have bigger problems.

https://andrewmurphy.io/blog/if-you-thought-the-speed-of-writing-code-was-your-problem-you-have-bigger-problems

Rings quite a few bells here.

#engineering #aicoding

If you thought the speed of writing code was your problem - you have bigger problems | Debugging Leadership

AI coding tools are optimising the wrong thing and nobody wants to hear it. Writing code was already fast. The bottleneck is everything else: unclear requirements, review queues, terrified deploy cultures, and an org chart that needs six meetings to decide what colour the button should be.

Debugging Leadership
Apr 11 at 9:33amWeb
Show threadEl Duvelle1d ago
@bnjbvr great read - thanks for sharing!
Show threadFrank Reiff1d ago

@bnjbvr @sinbad

Itā€™s undeniably accurate, but doing things that have no value to signal busyness to bosses that donā€™t understand, in order to give ammunition to CEOs to convince markets that they should speculate on shares in your company rather than that other equally dysfunctional one over there is what ā€œcorporate capitalismā€ is all about.

Show threadRobin HĆ¼bner1d ago
@bnjbvr I think even ignoring the AI slander bit (which obviously is what got me to read it) this is a great writeup and relevant even pre the slopcode era, everyone has seen some version of all of these pains I reckon
Show threadbartosz šŸš²šŸŒ³šŸšŸ¦€1d ago
@bnjbvr it is so true. It's something that I experience all the time at work but I couldnt quite put my finger on it
Show threadmako23h ago
@bnjbvr @mastodonmigration Iā€™ve said this probably 100x the last month alone: ā€œAI is not replacing engineers; it is exposing how much of modern software engineering was optimized around process and time management rather than engineering.ā€
Show threadMastodon Migration23h ago

@mako @bnjbvr

Feel like AI is only moderately effective because we don't build real things anymore. It might be OK at building spyware and data mining applications, but it is worse than useless at producing a 400 kph mag-lev passenger train.

Show threadMre. Dartigen [maker mode]23h ago

@bnjbvr It actually sounds like this is describing a long term issue in software that AI is just the most recent attempt at a quick fix for, because the actual solutions are, as the author admits, kind of boring and unglamorous.

I've never worked in software, but the case of 'the bottleneck is at station C but management just did something to make station A more productive instead even though it seems pretty obvious that C needs the help' is very familiar. Not sure if it would predate AI in the software or tech industries, but in others it definitely does.

Show threadAnnie G22h ago
@bnjbvr I'm sad how much sense this makes! Since I retired at the start of the pandemic from a 20-year career building Web applications for an academic research division, I missed having to wrangle with AI - prompts, code, or sycophancy. But having been insulated throughout my career due to the non-existent managerial layer (PMs were on to the next research job by the time I had data to work with, and I generally had to invent requirements) I had no idea that the screwed-up chain from no specifications to no user testing or feedback is the same when actual profit is involved! This makes me feel a bit better about my career, and a WHOLE lot more solid about all my reservations concerning AI.
Show threadā‚ Fish Id Wardrobe20h ago
@bnjbvr i'm here to say that the section "so where is the real bottleneck" is, 100%, accurate.