Because maintaining is not as fun as starting? Because other people can maintain a well structured project but not as many people can start something from scratch? Because his skills can be better used elsewhere?
That sounds like the Google philosophy though, where smart people come up with ideas and write the initial implementation (then get promoted) and other less smart people take over, and you end up with a mess that had great potential like Bazel?
My understanding is that this is not a philosophy at all, it is an incidental result of a bad set of misaligned incentives, where to be promoted you need to start something new or drive growth and post launch maintenance/growth is weighted much less
This could also end up like Fabrice Bellard's projects: yes he is no longer maintaining tinycc, but as a result we are now getting ffmpeg, qemu, and more.
