I believe the top performers are overwhelmingly using LLMs

[dysfun@social.treehouse.systems]

I believe the top performers are overwhelmingly using LLMs

ah, i see you're still measuring performance by lines of code. we knew that was bad literally decades ago.

[dysfun@social.treehouse.systems]

90% of working developers I know personally now use LLMs

i'm sorry, get better friends.

[megmac@social.treehouse.systems]

@dysfun also most of us who aren't are now stuck reviewing and/or cleaning up their crap, and getting demoralized for it.

[davidgerard@circumstances.run]

@megmac @dysfun 90% of traders i know are *mindful* drug hoovers, we assure you

[nfgusedautoparts@en.osm.town]

@dysfun @davidgerard I recall a place that tried to measure productivity by check ins. That did not go well.

[oytis@mastodon.social]

@dysfun I don't think we have a good measure for performance though.

One person in my LinkedIn feed used merged PRs as a metric, which is only slightly better than lines of code. An ex-coworker of mine said AI tools closed n% tickets from their backlog, which is closer to something useful, but most projects live with ever-growing backlogs just fine.

What could be a universal unit of achievement that would allow us to compare development with and without AI tools (or any other tools)

[cap_ybarra@beige.party]

@dysfun we know it's bad but there are no other metrics the code go fast machine can improve, thus we cook the ol books