Code is a human formalism that is only incidentally made executable. The elegance of code represents understanding of the problem to its most minute detail. LLM code being recognizably terrible shows it still doesn't understand what it produces to the full extent, and just as with people, it will inevitably compound to it becoming unable to efficiently work on it. And if the human abdicates that responsibility too, then there is no code, only subtly broken software.
I actually don't see any significant signs of AI use. There's Copilot listed in the contributor list, but I'm not seeing commits listed under it. Did they wipe it off Github?
Some seem to stamp Mostly AI based on weaker circumstantial like large init commits. Maybe it's just an artifact of human sloppiness.
Or maybe it was just the artifact of choosing these by last update, since vibe-coded apps genuinely do have an abnormal number of releases, and thus would be much more likely to show up.
Co-author is so strange. Like, would anyone credit an IDE or even a script generating code as co-author? Even legally, tools don't have the ontological requirement to be credited as author.
I always tell Claude to turn that off. It looks incredibly tacky. I'll leave an agents.md file in the repo if I don't care about people knowing AI helped me.
The rate at which these "labs" are creating these incidents is simply staggering.
Imagine we made nuclear weapons a private industry, had CEOs bragging how they have enough warheads to blow the Earth to smithereens, and then they "accidentally" nuked three cities over a short period of time each, saying they lost control, or rather couldn't contain their semi-autonomous weapon. All somehow managing to turn the PR around from their abject incompetence and towards SciFi visions of mankind hunted by self-replicating bombs.
This is effectively settled. Every client worth using has standardized on certain extensions, and there are server capability testers for it [0].
There are still a LOT of things very much up in the air, such as message deletion, which work inconsistently from client to client, but message history is not it.
call me out if I am wrong, but isn't history dependent on the server of the chatroom creator, and doesn't every server have different history settings and does it not handle encrypted messages fully?
Yes, XMPP server implementations typically give admins full control over data retention. Different use cases demand different settings, so it's quite adaptable to whatever needs you have.
I'm not sure what you mean by not handling encrypted messages fully, it handles encrypted messages fine.
> so it's quite adaptable to whatever needs you have.
as a user you cannot, even if you self host, because you are not the one controlling message storage
> I'm not sure what you mean by not handling encrypted messages fully, it handles encrypted messages fine.
you can load encrypted messages (including keys if they are needed to decrypt them if necessary) using a fresh login such that you can read them again?
Mastodon too has several client (web)apps. The Fediverse is itself a more neutral protocol with more heterogeneous platforms based on it (e.g. Lemmy, Pixelfed, etc), so Mastodon refers more to the platform (schema, UX, wider community) than the individual UI you use to access it, which just happens to usually be whatever your instance admin serves by default.
As for apps, people simply prefer instant messengers purposefully designed for their platform. There is genuinely a lot of platform-specific quirks around things like push notifications, which a non-iPhone developer not have the motivation to figure out. Besides that, a lot of these predate contemporary cross-platform technologies like Flutter.
Re: fediverse my point is - regular people just do not care about any of the fediverse-y parts of it. If I share a post I saw, I don't say "here's a thing I saw on the fediverse" - I just say "here's a thing I saw on Mastodon."
Plus the reality of the fediverse is if you want to use multiple platforms, you wind up needing multiple accounts anyway. Like if I want to upload to Peertube, I need to make a Peertube account, so we're back to where we started of having multiple places to follow people since I can't use my Mastodon account to upload to Peertube. I'm able to have some cross-platform interaction, like commenting on videos from mastodon - but there's issues there like, Pixelfed only showing people posts with images.
When talking to most people I just call the whole thing Mastodon and treat it like its a single app/service, because that's what people are used to. If I start trying to tout how the fediverse is a decentralized platform of multiple applications and you can kinda/sorta interact across platforms - people's eyes glaze right over. They just do not care.
For apps - I've never met a person that picks a messenger because its tweaked for their platform. They just prefer whatever their friends are on. iMessage is something of a odd case since Apple just rolls you into it automatically and presents it via the default app.
Re: Re: Fediverse. My point was more than indeed the Fediverse is more a technical detail than an actual cohesive platform, so it only makes sense to talk about Mastodon, whereas XMPP is more cohesive by virtue of instant messenging being a lot less opinionated.
It just so happens it splits into well-maintained projects supporting the concensus features, and dead or dying projects that don't, modulo a couple contrarians, with a couple more features floating in limbo, having finalized or draft specs that nobody really implements.
That sort of problem isn't best fixed by centralization on one client/server but just certification, which Conversations/ModernXMPP already leads.
As for apps, yes normal people aren't very conscious of what a native app is, but they do feel when their app lags on their older phone, doesn't recieve push notifications, or just looks a little odd in places, which can happen in a one size fits all app. Some of the most popular messengers, like WhatsApp or Telegram, to my knowledge use platform specific tooling, and normal people genuinely appreciate how snappy they feel.
But the issue is more so about developer time. Yes a cross platform app lets you reuse the same work everywhere, but the lead developer still has personal blind spots, and might simply not consider or be able to test things like bespoke Apple workarounds for non-iCloud/encrypted chat push.
Plus it's a communication protocol, like email; we should be expecting people reimplementing it in the weirdest of places. For software like this, the duplication of effort is what drives you to make an elegant design in the first place.
Sorry but I just fundamentally disagree about the app centralization.
If you want xmpp to gain traction you need an app that exists on both mobile platforms. It can (and should) have per-platform tweaks and everything to be performant. But a person needs to be able to just tell their friend a single app, they need to be able to find it in their app store.
The question is "how does xmpp gain traction?" and the answer is having a single flagship app that runs on everything.
If a person needs to recommend a single app is because most people are tech illiterate and don't understand what protocols are, because no one tells them.
I can expect this from our parents and grandparents, but not from younger generations that were born with PCs and smartphones. They need to learn; we need to educate them. Otherwise they will stay being the perfect fodder for Big Corpos and their monopolistic tactics.
We need to start focusing on the protocols, not the platforms.
This is a very flimsy argument to say a vibe coded one-shot Claude app is equivalent to actually hand-made Django app hosted on a Linux system that runs systemd which used Claude PR reviews.
People don't care about your entire dependency chain, see how little the most valuable FOSS libraries recieve in donations or sponsorships compared to end-user apps. They care about how well put together your own app is, which we know AI use changes for the worst (which some argue is worth it for momentum).
As long as your dependencies aren't outright dysfunctional (see rsync debate), it's concensus to disregard AI use there.
Someone recently rage quit one of the linux distros because the maintainers decided to accept LLM generated code (if it was good).
Other distros ban use of LLMs. Not sure how that's going to go down given Linus seems to be willing to use them for the kernel.
I don't think it's avoidable (unless you stick to a pre-LLM game engine), but it's clearly an error to underestimate quite how much people really loathe this stuff.
Just because techbros keep misreading allegories for societal ills as ads doesn't mean their authors have any influence on the future, beyond the warning or inspiration their works bring.
Yes, but does can author truly predict how they'll misread their work? Should we treat the audience as cattle to be deprived of every emotion that doesn't serve some external political goal? The author might indulge in vividly painting their setting so its more surface level reading might appear desirable, but when its evils are not even subtle but are already just actual text, can you do much more to counteract bad actors trying to sell society than same dystopia brought to life?
It used to be OpenAI was a real research organization that wrote real open-access papers that aren't marketing brochures, and when they did large training runs, they released all artifacts including model weights. Now certainly they are anything but. We haven't learned learned anything meaningful about ML from OpenAI since GPT-3 was released.
Their open-weights competitors like Facebook can at least claim some kind of public benefit, but it's still just running a well-understood algorithm on dubiously obtained data with longer and longer runs, give or take some inconsequential architectural tweaks.
Anthropic's mechanistic interpretability work is the most "lab-like" of these, but it's still just secondary to selling subscriptions and fear-mongering for regulatory capture/investment/publicity.
Is this actually grounded in real experiences and not headlines? Have real projects / people actually experienced this hyperbolic insta-hack you describe? People are getting DDoSed regularly sure, but no, you don't get hacked the instant you forward a port, in my experience hosting a few public services in my community.
This was an issue before watermarking was publically announced. I'm also not an expert, but aren't bits of a hash indistinguishible from uniform randomness?
Depends on the particular hash. Some hashes are designed to produce similar outputs when the input data is similar (for example, SSDeep). I'd imagine hashes designed to survive light copywriting would also not be entirely indistinguishable from random.
reply