Engineers are expensive. If you want to even launch, you need money to pay people (if even just yourself.) Boostrapping does work (see Basecamp,) but that business evolved out of a consulting business and wasn’t started from scratch, so they had advantages of paying clients subsidizing Basecamp development. A new kind of “micro-VC” where 10x returns weren’t the goal might be a start, but then it would be tough to raise money without substantial validation first. So there is a chicken-egg problem.
They also came out of the golden era of blogging where you could develop a significant audience subscribed directly through RSS without requiring some middleman distribution or aggregator taking your eyeballs. I guess it can still be done today with social media, but there’s just so much content out there it’s much harder to develop a loyal user base like 37signals did back in the day.