The key advantage of AI code assistants is they allow you to increase velocity. Velocity, of course, being the worst possible thing a software project can have, from the user's perspective. https://mastodon.online/@codechimp/117495289268759476
@mcc really love that the github issue appears to just be people summarizing the same problem with their own llm agent over and over again. that should get everything fixed
@chrisw_b i tend to write very detailed bug reports which i try to infuse with both careful testing and an awareness of what types of information a developer needs. depending on how things shake out, this tends to very occasionally be a godsend to an engaged developer, or just be tremendously irritating to someone doing basic triage because they have to dig through excessive detail to get to the bits they care about. i try to fight this tendency but it can be hard, it's how i think. (1/2)
@chrisw_b a thing i've been considering how is that soon, the mere fact that my posts are long will be an "ignore" signal even for receptive devs, because verbiage is a sign of LLM.
@chrisw_b (I also use a lot of emdashes, but for reasons I *cannot* explain, I realized this year I have been using emdashes wrong since high school. There are two emdash standards, you write them—like this—or else, you write them — like this. I somehow got the habit splitting the difference— like this. Not sure if that makes me seem more or less human)
Kinetic energy is also proportional to the square of velocity, which is how much more it's going to hurt when the slop code comes back to bite you 
@mcc
I would suggest it's only increasing speed, as it lacks direction.