The one that is actually supported and not abandoned.

I'm not expecting the people who take the vibecode shortcut to have the discipline and commitment and talent to steward their projects in the manner of the successful ones I've come to trust in the past.

You can fix bugs you care about piloting a $20 sub in abandoned projects relatively easily.

Yeah, I would and do, but many people like shiny and just jump around for some reason. And the ones that have a reason and make products popular overnight, have as reason that they need likes and upvotes as they are influencers.

People who have jobs, lives, other stuff to do in other words, keep sending me 'the latest new bla' many times per day, every day. All products that will be gone tomorrow.

I mean, AI made my own canva because my wife uses it and hates their support, bizarre amounts of long standing critical showstopper bugs and attitude to long standing members. So I hope they do get vibecoded to death, however, I would not release such a project until after a year of testing and fixing, which is where our version is about. Canva in the basis, like Miro etc is a one shot with Opus 5.5: it works fine-ish. But you will need months of grind to get it so people will want to use it.

Edit: in the vibecode age, 2000 commits would be the minimum to consider a project maintained. And then I do have a script to analyse the 'effort' of an average commit because some people fake commits with crap to inflate the number.

Which one?

There are many too many new projects (OSS and closed) that have AI in them and existing projects (Linux, etc.) that are getting tainted by AI.

I think this is a lost cause.