Vibecoding will lock in today's popular libraries.
There's already a preferential attachment effect for libraries that are popular today.
All else equal, it makes sense to use the thing that others are already using.
There's more likely to be documentation, other compatible libraries, bugs are more likely to have been discovered and fixed, etc.
LLMs are already more likely to recommend popular libraries.
With vibecoding, you just accept whatever code the LLM gives you.
That allows creating lots of new stuff quickly… but also means that it will be harder than ever before for new libraries to break onto the scene.
Maybe that's fine; the javascript industrial complex arguably already has orders of magnitude too much innovation/churn.
Perhaps front end development will now just be frozen as whatever was popular at the end of 2024.
As that layer gets less innovation, maybe innovation will turn to higher layers.