> I wonder if there's a similar overlooked-but-universal problem that someone in a garage is working on now, that'll spark a new wave similar to the dot-com boom.

I've got one for you. I've found a simpler universal grammar for programming languages. All modern programming languages use BNF grammars for syntax, and these have been researched and developed for about 60 years. But sitting unexplored for 60 years, has been another type of grammar, that uses only geometry to notate structure (achieved in practice via whitespace). I believe this new branch of languages will prove far superior to BNF languages in time, once tooling catches up.

I'm self-funding research on this. Like your description of the early web, it's really unimpressive now (http://ohayo.computer, https://github.com/breck7/treeprogram), but think it will catch on and be big. It seems so obvious to me when you improve something so low level that it will have big effects. But it's been hard to convince other research groups, and of course I could be wrong.