Project Fucking Sucks

I love those paragraph so much. Thank you.

“In this article, I’m going to be making an argument for gatekeeping. Not people, but culture, which is always worth defending. I’m not advocating for elitism, credentialism, or hostility towards beginners.

Instead, I say we should band together and defend the norms, values, quality standards, and our shared understanding of what open–source is for.”

I think this is very important distinction that should be understood broader.

I dont see many projects like this. Maybe because I often find out about projects through HN or naturally (e.g. someone gives a talk about it at local meetup) so I have curation bias.

What does tick me off is license changes and unclear commercial/free boundaries (and if they stay that way). I say this is worse as such companies obviously have $ to market these and so the impact is higher.

I would also add to his red flag list anything about gaming for Github stars.

This long predates AI codegen and I believe discussing it here is a distraction. Projects written to grab mindshare rather than to be directly used by those building them have been around for years. These projects exist to get attention.

I would name some, but I notice the author has decided to stay vague rather than call out examples, so I will too. There is something real here but this site is not really helping get to it.

Author makes a lot of good points.

In recent years, I’ve seen some open source projects “sell” their “product” harder than most commercial companies. Extremely polished websites...

Can’t remember the last one that really creeped me out but “Foreman” is a decent example. Feels like they are trying to sell me a monthly membership, and yet they aren’t even asking for financial donations.

Frankly, it makes me even question their motives. The Cups website doesn’t even list its features and Samba is pretty mild in is description. A host of widely used tools are also this way.

In contrast “Ventoy”’s website is more organic (and yet its lack of a documented build process and irreproducible binaries has certainly fed other speculation).

I certainly miss the old-school websites where it felt like I was interacting with something produced as hobby.

It’s not really possible to gate keep. What are you going to do? Forbid people from publishing things you think are dumb?

What this author and anyone who agrees could do is create curated directories. There’s already some of these in the form of the “awesome ____” lists.

I’ve been thinking for years that the time might be right to resurrect a curated directory site modeled after the old school Yahoo. Back then the advent of good search quickly rendered that obsolete, but today we’ve come full circle and there’s a need for signal to be plucked out of the noise.

> You know exactly which one I’m talking about.

No, I don't. Please elaborate (in great detail) :D

The article would have been fine with a theme/title closer to "things that annoy me". I agree with a lot of points raised, but not with gatekeeping. I'll die on the hill that gatekeeping is just wrong, anywhere, in any circumstance, and especially in a field that has open in its name. Open source means open first, source secondary.

So I strongly disagree with the "do. not. publish." point. Let the kids publish. Let them cringe 10-20 years from now when they see their first projects. We did with ours. My first geocities site had all the "slop" markers (triangle warning gifs, page under construction, etc), but I still laughed when I found some print-screens I had saved from that era, years later.

Writing slop and publishing slop is a rite of passage. When I see it, I don't have to like it, and I can avoid it. But I would never tell anyone "don't publish". Go ahead, there's plenty of bits on them github servers. Go nuts.

There's two axes here: Quality-Slop, Human-AI.

Q1: Quality Content (No AI)

Q2: Quality Content (AI)

Q3: Slop (No AI)

Q4: Slop (AI)

More of a spectrum, really, since most posts will in the middle of the quality axis, and most code makes some use of AI now (without being slop).

I think the problem with Q2 is that when you do it right, it passes. You think you're looking at Q1.

It's like the requirement for good AI is to be indistinguishable from human work.

(As soon as I see a ChatGPT-ism, I recoil, even if the article was good!)

Not sure how relevant the Q3/Q4 distinction is, but obviously banning AI completely would eliminate most of the Slop end of the axis too.

I think we'll either see a polaization where some communities become explicitly pro-AI or anti-AI, or perhaps a "hide all AI" toggle and mandatory disclosure (basically unenforceable, except for egregious slop).

"I’m sure there exists actually good AI tooling, but I’ll be honest, if I see a project whose description involves “LLM” or “MCP” literally anywhere, my immediate assumption is that the whole thing is vibe–coded garbage. And frankly, so far, that impulse has been correct."

And yet a good, fast, problem solving local CLI llm interface is missing. Either they're proprietary (claude, codex, gemini-cli), or they're just bad, or missing (AWS ...) or both. Ollama is better than even Claude imho for just text processing but doesn't seem to have anything that can actually act on a system.

Writing a bash script to do some ML task over 100 textfiles is ... pretty damn hard.