Comment by kshri24
3 hours ago
What worries me even more is tens of thousands (or even magnitudes higher) half-baked, over-hyped, vibe-coded spaghetti "open-source projects" released publicly for clout or to attract investment.
It is like all the garbage papers you find in academia that you need to sift through until you find that one good paper. Needle in a haystack.
2026 will be the year of vibe-code driven enshittification. Github will be the casualty.
I'm gonna blow your mind a bit here, but this isn't just the fault of the people making the software, it's also the fault of the vast majority of the people here and on the internet in general. Quality doesn't get your attention.
The truth is building a project is like a lottery ticket, and there's hard diminishing returns on time invested in quality in terms of payoff. If I told you you could spend 10x more time for a 2x increase in probability of success, if you were trying to make a living from your creativity, you would be stupid to spend the extra time, it's a horrible investment.
The people spamming half baked projects that they quickly abandon if they don't get traction are being rational. People like me that grind on unsexy process bottlenecks and try to keep refining into something really nice are the irrational ones.
In the last 6 months we've seen no fewer then a dozen vibe coded/AI assisted open source, self hosted projects launch that complete against ours. So far all but one has fizzled out, with the same pattern each time: announcement, repo with 1 giant commit, 2-4 months of feature releases, loss of interest from the author, and finally abandonment.
I expect once users get burnt enough time, they'll stop adopting the new cool thing until it's been out long enough with consistent releases.
It will be interesting to watch how they decide what new data to train on if most of it is low quality.