Bun's problem may be developing in the open
8 points by av
8 points by av
Do you ever open a link and start to read and immediately begin to notice certain very specific patterns and structures in the text?
Single sentence paragraph.
I sure have. Very short sentence. It's everywhere. I can't stop noticing it. It's everywhere. Another short sentence.
Another one sentence paragraph.
Did everyone really used to write like this? I feel like I'm going crazy. I swear down this wasn't a common writing style outside of maybe LinkedIn five or six years ago.
But now it's everywhere.
It's not something – it's something else.
Of course it is.
And I'm tired of it.
And so on.
I call this the linkedin style because that's most often where I saw it. The style is absolutely detestable. When I see people using it, I immediately begin to question the quality and authenticity of their work. I'm also tired of it.
And in this case, you can see his archives:
2018: https://00f.net/2018/10/18/on-user-authentication/
2021: https://00f.net/2021/03/26/it-doesnt-work/
There are a few “punchy” sentences in the older post, but the overall feel is completely different.
Already the styling of the page is a giveaway... Specifically the light-to-dark gradient over text and generous inner whitespace of boxes. (At least, from what I've seen. I have no idea if Claude et al. actually like this style.)
my friend writes in similar fashion, when I asked him, he said it's because of low attention span of people these days and such styles does well
in this particularly case, the writing may be bad but the actual take is good. I think they captured what was going on in all the bun drama very accurately.
In my mind it reads a lot like "default" LLM text generation: the style of speech an executive uses standing on stage at the global town hall or maybe product launch. Short, punchy sentences that will be accompanied by graphics and maybe a light show. Lots of pauses for dramatic effect.
I highly doubt any human actually writes those and just assume it is a model spitting out what LinkedIn wants to hear.
I couldn't agree with the base sentiment more (as in "developers shouldn't be judged for unfinished experiments"), but I feel like this sort of explanation would work much better either if this was a small project with no expectations or if Bun properly made a statement about it.
As it stands, the optics of this is "after VC-funded company is acquired by big AI company, suddenly a vibecoded Rust-rewrite branch appears out of the blue, owner dismisses it in a random comment on the Orange site as simply an experiment, then suddenly the branch is merged and Bun is now a million lines of Rust bigger."
It might as well be that this really is just a benign experiment or was just made to placate Anthropic, but all this could've been avoided with a blogpost on the project's site with a short explanation of their plans and expectations. Or, you know, just doing this stuff in an internal, private repo, and then asking for feedback once the project is in a presentable shape. That is "research and experimentation," not by making a throwaway comment and then pushing a massive changeset through.
I find it hard to blame anyone for feeling a tad nervous and considering alternatives seeing this.
owner dismisses it in a random comment on the Orange site as simply an experiment, then suddenly the branch is merged and Bun is now a million lines of Rust bigger."
he also posted a semi update kinda thing later, as how it went from experimental to something more
cargo check reported over 16,000 compiler errors when I wrote that message. It could not print a version number or run JavaScript. I didn’t expect it to work this quickly and I also didn’t expect the performance to be as competitive. There’ll be a blog post with more details.
Well, this was published today. They did pull it in earlier in the week.
Could someone who flagged this explain why they flagged it?
I flagged this. It is unambiguous slop. The moderators here have expressed that they do not want to add a "slop" flag, against the overwhelming wishes of the community, and have communicated that the appropriate action is to flag such content as spam, so I have done so.
I disagree with this, I think slop is a special category of spam that can be submitted in ignorance (as was done by ~av here), but... in absence of a slop-reporting mechanism, "spam" has to do.
Bun is part of Anthropic now, and I'd categorize it as "absolutely not representive of open source projects" when it comes to incentives, priorities, community and culture. It's not only a large OSS project anymore.
I don't want to shame people into not using AI, a project with this much eyes(and money) will get a very different treatment than my 0 star dotfile repo. But if anything the value of this move and the sudden-ness of it is not obvious to me, nor do I feel like it was communicated well.
People being quick to judge or hate, is a social problem, but I don't think it absolves a large project from criticism, and framing the PR as some non-corporate(really? Bun and Claude?) hacker doing some experimenting is definitely not how I see it.
Slop should be criticized(constructively!) imo, for me it dehumanizes and devalues an OSS project(even if some big corpo is behind it).
I'm also unsure what we would call a 1mil+ changes pull request that started as an experiment...that got merged not shortly after. A triumph of automating through NLP or slop(or both?)?