For example, I'm a big fan of DOSBox. And while the core project still has some code check-in's from time to time, its last release was 7ya. Now I use and follow forks because they're consistently releasing useful features and fixes.
I've been following Haiku since it was announced. In that time a couple of personal friends have passed on, and a number of family members. In that time children have been born, gone to school, grown into adults, and had children of their own.
Projects that drag on forever make for an interesting artistic statement, but if their goals in any way include utility then velocity is important because time is important.
Many other niche projects have enjoyed enough velocity to have meaningful utility, in my lifetime. Linux was still niche when I started using it in '94.
Some people enjoy building things. It's not performance art, it's a hobby. Can Haiku not be a hobby for people who enjoy hacking away at BeOS? (You might think I'm splitting hairs here, but performance art is precisely the wrong analogy to use, since performance art is for the benefit of others. Performance art is inherently public and assessable in terms of how it reaches others, unlike hobbies. Performance art is env=PROD.)
Was the purpose of Haiku ever utility? What do you define as utility? If I spend a day building a shelf, does that have more or less utility than buying a shelf from IKEA? What if I like making shelves? What if I like making OSes? Does the shelf (/OS) that I build have less utility than the one from IKEA (/MS)?
In economics, utility is a measure of individual satisfaction. The person who likes tinkering with Haiku is deriving cold hard economic utility from Haiku right now. What should it matter to this person whether you find Haiku web scale and production ready? John Q Public is never going to run Haiku, no matter how much polish you throw at it. Why rush to some goal line that doesn't exist in the first place?
Art has utility, of course; but we're discussing primary goals.
I'm not at all trying to be overly critical or pissy; I'm genuinely curious as to why this is always the gain that AI brings as though it matters (or, rather, this is how it's always stated in one way or another). So I come from a place of confusion when asking this. Why should I care if you can write 100000 or 1000000 lines of code and push out 10 billion commits per minute? (Yes, I am intentionally exaggerating here to prove a point.) If anything, I would argue that increased velocity just brings about recklessness and unnecessary churn instead of accomplishing anything useful.