One thought I've had about AI and programing is that you'll run in to a similar problem. Code is a bit special because it's a language that's understandable by both humans and computers, and when you're programing you're essentially writing for both audiences at once.
Voice is maybe not as important when writing code (although you still want to keep the coding style consistent) but even so I think that writing code by hand has the advantage of you being able to express your thoughts in a more coherent way than the output of a handful prompts will. The problem isn't just with however powerful the AI model is, but that prompting is a kind of vague and indirect way of interacting with the system and it necessarily introduces another layer between the author and whoever ends up reading it.
Ok I understand that you don't like bluesky for whatever reason, but could you actually formulate why so that it's possible to have a discussion instead?
It uses ATProto (the bluesky protocol) and allows you to self host the git part and/or your personal data (e.g. comments that you leave on other repos). It's still very much in development as is the ATProto itself, so it doesn't seem mature enough for serious use yet. ATProto does for example not handle private accounts/posts yet which means that all your tangled repos have to be public.
If you find yourself writing regexes often enough that speeding up that process would increase your productivity by "a lot", then you should get good at writing them yourself which means practicing without an LLM. If its something that you don't do often enough to warrant getting good at then the productivity increase is negligible.
I think the main benefit here isn't really productivity but developer comfort by saving them from having to step out of their comfort zone.
One of Zed's focuses is AI integration. Other than that, I suppose if you prefer having a dedicated application with its own UI instead of a terminal based ones.
I try to write comments whenever what the code isn't obvious on its own. A "never write comments" proponent might argue that you should never write code that isn't obvious on its own, but that doesn't always work in practice
Sometimes you have to write cryptic code for performance reasons
Sometimes you have to deal with unintuitive edge cases
Sometimes you have to work around bugs in 3rd party code
Sometimes you are dealing with a problem that is inherently complex or unintuitive, no matter how you put it in to code
One of my most controversial gaming takes is that I like the first witcher game the most of the trilogy. There is a lot of jank and some cringeworthy parts but overall it feels like a much tighter experience than the later games, notwithstanding some clearly undercooked parts. It takes a lot more cues from older rpgs in how it's structured and I suppose I might just have a weak spot for that.
To be fair, I've never gotten that far in the third witcher so maybe I'd like more it if I played it enough to properly get in to it. I just got kinda bored after a dozen or so hours which is not a problem I had with the first witcher.
Yeah I in general think that Graber is very good at giving good answers in interviews!
Honestly even if bluesky does become enshittyfied, which is a very real possibility, the work they've done on AT proto so far will probably be extremely useful for whoever takes a crack at a more decentralized internet next. There are a lot of clearly smart and passionate people who are given space to research and experiment with different ways of doing things and I think that's both very valuable and interesting
I agree that the interviewer shouldn't have implied that they are decentralized today! I don't know if bluesky even say that they are decentralized themselves, on their website it says that they're "building an open foundation for the social internet" which is more accurate but maybe they mischaracterize themselves somewhere else.
So first of, the part of my comment that you quoted doesn't make sense because what I'm saying is that bluesky theoretically allows for decentralized relays but it's impractical in practice. Your analogy doesn't really apply to that.
I do think that it's misleading to call bluesky decentralized today (at least without any caveats). The goal of the project however is to eventually create a more meaningfully decentralized social network and they have tangible plans for moving in that direction so I think it's unfair to dismiss this aspect of bluesky completely.
I think that it's fair to want the interviewer to ask more critical questions and in general be more precise with their phrasing but
repeat that PR talking point
is a very cynical and uncharitable take on bluesky and decentralization. Cynical takes aren't necessarily wrong but they're not necessarily correct either.
The AT protocol is by its own account an ongoing project with problems that still need be solved before it is able to provide a social network with all the properties that they're interested in.
I don't think that it's accurate to say that bluesky is "completely" centralized (it is less centralized than most social media) as much as it's de-facto centralized. One reason for this is that it's prohibitively expensive to self-host relays. This is something that the AT protocol devs have plans for addressing, so it's possible that this de-facto centralization is a temporary stage in the evolution of bluesky and AT proto.
It is of course possible that they are lying or that they will be unsuccessful despite best intentions but taking for granted that it's just a "PR talking point" is, once again, very cynical in a way that I don't think is completely motivated.
What are the odds that you're actually going to get a bounty out of it? Seems unlikely that an AI would hallucinate an actually correct bug.
Maybe the people doing this are much more optimistic about how useful LLMs are for this than I am but it's possible that there's some more malicious idea behind it.
They are open to drop some features apparently, but maybe not "90%"
The developers are taking a “less is more” approach. This means that some features of the original sudo may not be reimplemented if they serve only niche, or more recently considered “outdated” practices.
Well it's not like everyone who uses Chat GPT is going to become delusional but if you start going down the path Chat GPT is going to make it a lot worse
One thought I've had about AI and programing is that you'll run in to a similar problem. Code is a bit special because it's a language that's understandable by both humans and computers, and when you're programing you're essentially writing for both audiences at once.
Voice is maybe not as important when writing code (although you still want to keep the coding style consistent) but even so I think that writing code by hand has the advantage of you being able to express your thoughts in a more coherent way than the output of a handful prompts will. The problem isn't just with however powerful the AI model is, but that prompting is a kind of vague and indirect way of interacting with the system and it necessarily introduces another layer between the author and whoever ends up reading it.