all 13 comments

sorted by: hot top controversial new old
[–] 7 points 22 hours ago (3 children)

but who asked for this? doesn't every software have their own preferred place for suggestions? why do we need to centralize this?

  • source
  • hideshow 3 child comments
  • [–] [S] 5 points 22 hours ago (2 children)

    I needed this for Pixelfed and Loops, and figured that the existing auth system in FediDB would be easier and more frictionless for people to use their existing fediverse accounts, rather than make new accounts for this. The fact any other project can also use it is another benefit.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 3 points 20 hours ago

    I see you publishing improvements to fedidb but i doesn't seem to me the source code is being update. Are you publishing these changes as open source?

    Also misskey is reporting incorrect numbers. i reported it here. tbh it makes it kinda harder to trust the reliability of fedidb.

    Also any reason why it shows a graph of active users for software like lemmy but not for piefed?

    Thanks for working on this!

  • source
  • [–] 4 points 23 hours ago (7 children)

    This blog post reads like AI slop.

  • source
  • hideshow 7 child comments
  • [–] 3 points 23 hours ago (6 children)

    Dansup does rely on LLMs, so it wouldn’t surprise me.

  • source
  • parent
  • hideshow 6 child comments
  • [–] -1 points 22 hours ago (5 children)

    @taco_shale032@lemmy.ml @jonathan@piefed.social provide us proof for your maga tactics, don't just shout fake news

  • source
  • parent
  • hideshow 5 child comments
  • [–] 5 points 21 hours ago* (last edited 20 hours ago) (4 children)

    My apologies, this is what I gathered from other users in combination with Pixelfed (or loops?) allowing LLM contributions at some point. I should’ve done some research to confirm before assuming it was true.

    Now that I am doing some research about it, could someone explain what this commit is about? Skill markdown files are for LLMs.


    Relevant blog post

  • source
  • parent
  • hideshow 4 child comments
  • From the blog:

    Needing so this many commits to undo the LLM slop means the developer does not have a clear mental model of what the earlier commits consist of.

    That's a big leap in logic without supporting evidence. The author goes on to question dansup's competence as a developer based on that, which is the dumbest metric you could have. You can measure devs by their output: quality, speed, correctness, beauty of their algorithms, success of their software (for many different ways of measuring success), etc.

    Back to LLM usage: it's entirely possible he was dabbling in their use and this reaction led him to reevaluate his stance. I remember many public reactions specifically targeted at instructing him on why LLM usage is bad, with a respectful tone. It's entirely likely he had a change of heart and sought to purge LLM presence from the project. Lots of people make mistakes, learn from them, and take steps to change. And that's something to be praised, not vilified.

  • source
  • parent
  • [–] 1 point 20 hours ago* (2 children)

    The existence of the skills doesn't necessarily mean the app is vibe coded or slop. We have to judge code quality and UX to make that determination.

    Also, he removed those files 3 days later according to this commit. Maybe he only uses them locally and they got added to the repo by mistake? It would at least imply that he doesn't want everyone else to use AI, though why is anybody's guess.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 2 points 20 hours ago (1 child)

    I’m aware it was removed, that’s why I said "at some point". I never claimed that the app is vibe coded or slop, I simply pointed out that dansup relies on LLMs which could explain why the article reads like AI slop.

  • source
  • parent
  • hideshow 1 child comment