Awesome repositories are no longer awesome

I help maintaining one of these awesome lists, awesome-javascript, since 2018. There's been always self-promoting or similar activities, or someone coming up with an idea/forking a repository and doing some small changes, then thinking they're suggesting something groundbreaking. Because of this, we needed to add a few checks over the years to the pull request template, that basically asks if the resource is out there for a while, actively maintained, actively used by people, etc.

Recently, thanks to AI being this cheap and accessible, the amount of shitty PRs quadrupled if not more. And the worst part? Agent doesn't care about the checklist at all and re-writing the PR body with bunch of nonsense that we maintainers need to read, understand, and evaluate. I started keeping the score for the most shittiest PR based on how soon after the first commit made they open the PR. It was 11h, then 10h, and now it is 6h. How on earth a 6h old repository with zero user is considered "awesome" I don't know, but people should stop prompting their agents to create 10s of these PRs every day.

Yes, there's always an alternative: stop maintaining these lists. I don't think it's a viable option either. Sometimes you start using a new piece of tech and want to have access to somewhat curated list of resources that might speed up your understanding. There's still a bit of value to keep them alive and it's dying out bit by bit (literally) every time someone mass-creates these spammy PRs.

Thanks for reading my unsolicited rant.

The problem is digital things don't decay in the same way as physical objects. AI can shit out a webpage for an institute that doesn't really exist and make it sound awesome until you realize there's nothing there. A webpage with a vibrant community behind it that is actively maintained looks, at first glance, the same as a page that's been put out to the pasture. A physical store that's out of business will accumulate cruft, broken windows, vines everywhere. There's not really an equivalent for bitrot. Sure, my website from 1996 looks dated because I used frames and we don't do that no more, but eBay and Yahoo and Myspace are still around. Link rot does set in though. Links to pages that no longer exist because the person running it has passed. Archive.org has such a good purpose.

Thing is though, for software, sometimes things can just be done. If a perfectly good piece of software hasn't been updated in years because it hasn't needed to be updated, then let it sit, ensconced in stone. The problem is it's hard to tell the difference between that, and something that needs maintaining but hasn't been.

They were always junk, full of formulaic, self-promoting, astroturfing trash. In most cases, it would be more appropriate to view them as content marketing (c.f. digital pollution), not actual developer resources.

Every 'awesome' repo I can remember has seemingly been made by someone who hasn't used the things included and only read a project's readme. This dates back years before the LLM craze.

I would rather hear from users of software and their opinions, otherwise I can't take seriously something ostensibly curated for quality. In my experience such lists serve better as a way to discover new projects using eg. some framework you're interested in, if they're maintained (which is still valuable).

I maintain 4 awesome lists and I agree: it takes a lot of time and some are more catalogs than curated lists. Curation takes time and is like gardening especially given the half-life of content on the web. My solution was to encode my editorial process in the form of contribution guidelines: https://github.com/kdeldycke/awesome-iam/blob/main/.github/c...

I found a few lists that I found genuinely useful, like self-hosted. But I also think the problem with them is they were gamed for some reason, maybe because of Github stars and the fact that Github attention is useful for other things. I think it would have been much more useful to have them almost on a different platform. IMBD lists for example seem to work better because there's not much reward in making them except a love of movies.

Also I think another problem is that they were given high recognition on Github, and thus became a formulaic pattern rather that some individual who is really passionate about trying the stuff on the list, which also made most of the later ones generic and much less interesting...

I've personally never found awesome lists great at curation, at best they were always a laundry list of what was out there. But I think they were an excellent jumping point for discovery. I was recently working on a new project with C#, and never having used C# or the .NET ecosystem the lists were a great way to find a lot of options (even if I didn't end up using them).

Some of them probably don't really need to exist, I think I saw one that was fantasy book recommendations which looked like it was just a copy of the top 100 fantasy books from goodreads.

That highlights a bigger point. In the age anyone can create and publish their own software, including people without the most basic knowledge about it, how is one supposed to distinguish good from bad and avoid drowning in a sea of crap?

Also, this has another angle, let's say I've created some tools I've invested quite a bit of time in but what is the point of publishing it, if anyone that needs it can simply make it for themselves (spending a week or two of time and some tokens)?

Not only it has became essentially impossible to find if a given thing is quality (for example I recently looked for a pi plugin that does tool use classification for AI that I could extend with my ideas. I found one, then I discovered the author copied leaked cloude code source and "released it under MIT license" - which made my work on top of it basically useless and I had to throw it all out and start from scratch, as I wanted to publish the result- the final thing is called adaptive-acl).

And second, is there any point in publishing code? In the "old days" when software was worth something the motivation to publish open source, in addition to feel good factor, was that people would send you patches with functionality you may find useful.

Both that and the feel good time has ended. I still publish some of my stuff that I think may be useful to someone at some point, but I'm not sure there is any actual point in this.

In a way AI has made each of us much more self sufficient. A computer with a good local model and half a dozen GPUs, few TB of software and one can have all of their computing needs fulfilled forever. No Internet or other people required. There is certain power in this. My young pre-internet self would have felt like living in a star trek universe with this :-)

But something is definitely being lost in the process as well.

Even before AI they hardly provided any value.

Are you going to click them one by one and figure out what they do? When you need a library/existing tool, do you use Google or come to such a project and do a search?

I am old enough to know Yahoo Directory, and "awesome" projects are exactly that.

You can't really do much other than bookmarking/"starring" and then completely forgetting about it.

Never understood those repositories. Always viewed them as a collection of junk.

I’m maintaining awesome-multiplayer-games[0], and was also surprised about a recent surge of attempts to add AI-made games (or simply other unrelated games), despite not being very popular or so. I suppose LLMs are helping to find these kinds of repos and can quickly create PRs, in attempt at promoting projects. I’ve now added official contributing guidelines that I can point at when closing such PRs.

0: https://github.com/herrherrmann/awesome-multiplayer-games

I get the impression that we need social mechanisms to segment ourselves away from AI.

I'm broadly pro AI, I like datacentres etc, but its clear we cannot let our entire culture be penetrated by it.

At work we'll use AI aggressively, at home we'll socially punish anyone who blurs this line.

I've heleped maintain awesome-bash for years and I see a similar pattern. Almost all of the recent PRs are links to AI-generated projects. And somehow, within days or weeks of their creation, they get hundreds of stars.

There's always been self-promotion and spam, but AI takes it to another level. On many issues, I don't think I'm responding to a real human.

It would be nice if forking and modifying these awesome lists were easier, since many are out of date. I have my own personal catalog, roughly based on many awesome lists, but it was a bit of a hassle to setup.

I think that my site (https://relatedrepos.com/) is a more useful and modern approach to discovering open source projects within a neighborhood. It is updated daily and therefore does not go out of date.

The related repositories are more relevant to what the user is looking at. Also, the semantic search functionality helps you to explore according to meaning rather than just keyword matching.

I used to love these lists. In some ways I think they might be more needed than ever, good quality web resources are in an even smaller minority than ever.

I maintain awesome-bugbounty-tools [0] and have never seen that many PRs before.

I had to add a CI check to fail repos with fewer than 50 stars. In the past two months, I closed more PRs than I have merged in the past 6 years or so.

Honestly, it doesn't add that much value anymore, so at this point archiving the repo will be awesome.

[0]: https://github.com/vavkamil/awesome-bugbounty-tools

The problem maybe can be solve by have a better and higher standard rules, like only repository that have more than 2000 stars, active maintainance, and actually working can be listing. And we need to verify the PR follow that rules using AI agent, not manual, and it will make the life of awesome list's maintainer become easier.

I take my list to 1001 entries and only then start pruning. I'm at half that now so I know I currently have non-awesome entries. However, during this ramp-up I collect the information needed to programmatically prune some day. A repo without recent commits might be perfect and therefore awesome. Taking some time to add more and more signal to each entry will eventually yield the metrics needed for cutting off the tail.

Expecting awesome to mean curated manually is unrealistic now. But there are plenty of opportunities to make an awesome list compelling via automation.

One of many examples is WebMCP-support. How often did you find a truly awesome, perfectly maintained list and felt overwhelmed? Then you bookmark it and get a coffee. A list with proper agent-support will allow me to cut through the non-noise. Finally the list will be actually helping me instead of being a bookmark.

There are more examples like this. Awesome Lists have a golden future as one of the knowledge resources that are not gatekept.

I maintain a very very niche one of these (an ‘awesome’ list) and for the past few months I’ve received several non-awesome PRs per day.

I'd like to better understand the reason behind these pull requests. Its not like Claude is goofing off on the Internet and trolling humans. People are prompting AI to create pull requests against a bunch of repositories.

But why? Are people trying to advertise their tools in the "awesome" lists?

What about the many PRs against popular tools? Are they trying to build a resume through OSS contributions?

Downfall of modern society if things don't change course.

I maintain an awesome list, and I guess it just isn't popular enough to be barraged with low-quality PRs? I got a PR recently that I suspected was AI created, but it was just corrections and I verified all of the corrections to be... correct. So I merged it.

- https://github.com/dannyfritz/awesome-ray-tracing

There is a lot of value in good curation but its also extremely time consuming to curate well. Its very difficult to do it as a group activity especially a highly disconnected one via git/opensource and where a lot of these lists started out good the collective effort has ruined their value.

I think we need curation more than ever on the internet on a range of topics where social media of various forms don't work very well. I also think there is a real place for index's as well because search often does adjacency quite poorly since its not really optimised for that. Short curated lists are great for beginning, long ones not so much to anyone.

I used to reach for an awesome list the first thing as i ventured in something new. Nowadays, i almost kinda forgot about them. Still I hold a fond memory of it, they usually contained some real gems (especially when they were relatively new).

Third, many of them are maintained by people with vested interests - for example they're an "awesome list of marketing automation tools" and the author is the owner of one such tool.

They forgot a third reason...

Rather than distract myself picking through awesome lists to solve minor problems, fill minor gaps, AI just implements it as part of the task

Which is great cause using the web is a context switch I don't want.

It's all dynamic function generators all the way down. Anything else is fundamentally bike shedding syntax and semantics of a spoon that doesn't exist.

Anything else is making Mona Lisa out of code, objectifying it with formal OOP dogma.

The intention is really important here. If you're applying good curation and editorial guidelines, it makes a world of difference.

What I want are boring lists. Tell me the one or two most boring answers to my problem. I don't want you to lump the incumbent project in with then ten opinionated reactions to it as if they are all equally mature.

Simple: create an awesome list of awesome lists.

> They usually contain curated knowledge and resources related to a certain domain

No, almost never curated.

I just use these lists as plagiarism. I have never selected software from them. Either in the before times or after. But now I sometimes point the agent at it to tell me things and steal ideas.

For R there are CRAN Views, which are well maintained: https://cran.r-project.org/web/views/

I still find a lot of interesting resources and ideas on Awesome List trackers like this: https://www.trackawesomelist.com.

content curation > content generation

I like how Github users independently re-invented the early Yahoo.

Favorite awesome list memory: maintainer of a certain awesome list accepting my project (which made a nice QoL improvement), copying my approach for his own project, and then kicking me off the list. It's his list, which is fine, it's just so petty.

Hope the Github stars are treating you well!

> You are in most cases better off simply asking a decent AI model for inspiration or the material you need.

I wonder where AI model get their lists from...

Downloaded the top 1000 repos locally

Point my agent at them and tell it to find useful code - total size is around 100gb

There's lots of junk to waste time on out there get to work!

Could have said “awesome repositories considered harmful”

Easy to solve: we just need to start creating curated awesome lists of awesome repos!

Implicitly credible sources attract those who benefit from exploiting credibility. This weakens credibility until audiences move on.

I'd like to add a third reason: SEO hacking

I made an extension to show me github stars/last update/number of commits, and a few more bits, so I can quickly sift through the BS.

https://github.com/testy-cool/repoaura https://chromewebstore.google.com/detail/repoaura/ephpinbdbo...

Also I had been wanting to curate an awesome list as well, and made myself an annotation dashboard. For example I wanted one to curate Claude Code Mods.

It's a pretty simple idea you can expand on https://imgur.com/a/XjRCbL2 I have not had enough mental space to curate them though. I wouldn't let an AI handle the curation, because I notice the curated content I like has effort put into the curation part.

Like YT vids, anything titled with AWESOME, leaves me suspicious.

Disagree. Some material is just timeless or even just having a name is good to go on compared to knowing nothing before.

AI is nice but never gives the same answer, will tunnel you if you have zero idea on the topic, and can be outright wrong or refuse to answer.

They always felt like the Webrings of the 21st century.

I don't think this is down to AI as the author suggests. I feel these repos were always about farming stars (or whatever else people are trying to get out of GitHub outside of code contribution).

I remember the first ones were genuinely cool and interesting, and you could tell someone made them because they actually liked the tools. But it quickly turned into dumping in whatever was popular in a given community, often tools the author hadn't tried even once. When a list has 100+ links, do you really think the author checked them one by one?

A list of a sensible size looks different, e.g. this one: https://github.com/ibraheemdev/modern-unix

awesome-selfhosted was a favourite, but has become outdated. Other awesome repos have become a target for self promotion and junk.

It's kind of inevitable that they stale because of the growth. Let's say it is about programming language - if it's good language then there're more good materials about it, that means more content to maintain in an "awesome repo", and then again and again, with or without AI it would go fully stale or unmaintainable.

I guess there's no solution, these things should go to history and we need to find a better way to find "best of the best" individually while avoiding SEO-hacked resources with some bs. I actually think there's future in individual learning, so everybody will have their own learning plan, their own learning materials and all that adapted to the way they think and consume information.

Were they ever awesome? By design, they attract projects looking for visibility etc.

Most of those so-called awesome lists have zero to very little no human judgement when new projects are included. One example I found in the wild is [0], which has ca. 50 awesome projects already while libghostty isn't even stable. I doubt that even 10% of those projects are anywhere good, let alone _awesome_.

I, for one, would like to have an opposite — awful-<something>. Just so we know which things we should avoid. With AI, it's getting even more important to know what to use.

[0]: https://github.com/Uzaaft/awesome-libghostty

In the long run we need to find viable alternatives to github.

I don't mean gitlab either. I think we kind of need repositories for everyone, as a human right - sounds strange but IMO we kind of need to build a new web with a lasting impact on the world. We should not need web-archive; things should be documentable. (Github is not responsible for disappearing web-pages as such, mind you; I just mean that we need to look at the whole problem here, not just on github only as gatekeeper.)

I don't like Google, but google-docs is great. LibreOffice should have something like that too. Every state as well. I really think this all needs to become a human right. Why is a US company offering this and just about everyone else is not, including governments? I don't understand it. And that is also just one example of many more things. We need to change the web to a better variant. In many ways the 1990s was better than what we have today. Everyone built websites. Today this is a bit ... well. Gone. People may at best put their "website" on facebook. That's a walled garden. It's not the same as having separate websites. I don't want the world wide web to become a walled garden gatekept by a few companies. But I don't see alternatives emerging - and nobody has time anymore, which is another problem.

Most of these were as slop as they are now.

The difference today is just that AI has supercharged the sloppening, making it obvious even to the most oblivious of observers, that the format has issues.

Which is a common theme in many things that have been "ruined by AI"

Were they ever?

I mean, even at the time of creation, when it was still timely, a link to a top 50 downloaded packages list is a pretty shitty link to include.

Even before slop, I found a lot of those awesome lists and CS degrees repos useless.

So, according to this, we should be using LLMs for recommendations of materials we need, but we shouldn't trust LLMs when they publish a list of materials we might need. Got it.

So your problem is that these lists are filled with AI slop or are out of date, and your solution is to...ask the slop engine that doesn't understand numbers.

"As AI gets better at finding and creating resources (and creating pull requests for said resources), these awesome repositories will become less relevant and less useful."

And how did AI get good at finding awesome resources, according to you?

What’s the LLM equivalent of SEO?

I’m wondering if there are techniques people are using to get their content featured by LLMs. Of course you won’t get clicks or attribution so maybe there’s no point.

They never where.

tbh, the first one (whichever it was) was certainly awesome. everything after that, not so much. this is just the world we live in.

The real tragedy here is the asymmetric warfare AI enables.It costs zero effort for a spammer to script 100 PRs linking to their AI-generated wrappers,but it takes genuine human energy for maintaners to vet and reject It's a fast-track to maintainer burnout.

[flagged]

[flagged]

[dead]

[flagged]

[dead]

[flagged]