The other thing to consider is that for me to think that forking makes the future of the project questionable, you'd have to show that it's likely that usage(original project) < usage(upstream) + usage(fork). In other words, I'd be interested to know the combined usage of MariaDB and MySQL and see if you could actually predict with confidence that it would significantly lower than original MySQL usage. Not just that one of the forks didn't have as many people.