it's interesting that the novel/hobby projects all failed.

the ones that succeeded, are those that can be used for marketing by other companies (whoismyisp) or which already exist (nslookup) so there is a proven demand but you made a cleaner solution.

i was always into this idea of building mvp to test markets. but the main issue is getting an audience. often success is 98% luck and timing and 2% hard work.