That would be pretty awesome actually. All of the noted features, but particularly syntax highlighting and some sort of versioning. Key use case would be "multiplayer programming," rather than just one person driving while the other person stares at the screen over their shoulder.
TeX and Markdown would be good, but they strike me as different enough uses that maybe they'd be a separate app -- with TeX or any other markup language, you'd want to be able to plug it into a publication system, or at least generate nice output (whether DVI, PDF, XML, HTML). Lot of users there as well, but they might be different users.
I don't see why it's worrying. Half the reason to do something like 'make an etherpad competitor' would/should be to learn how it works. Once you have that, there's a ton of different products you could make/morph into.
I would personally be inclined to join the open source development.
But to 'compete', tailored support for any non-technical niche. The equivalent of the "Basecamp for X" phenomenon, where X equals churches, fashion, teachers, politicians, etc.
Something for co-authoring & proofing academic papers would be nice, but I suspect neither a professor or a student would pay for it.
Focus on collaborative editing for things like school reports, round robin novels, etc. Support TeX eventually, images sooner. Then hook it up to print on demand and allow people to sell beautifully bound copies of their stories/theses/whatever right from the site.
I'd like a real-time-updating version of instacalc. Great tool for communicating about calculations, but it sucks having to send the URL back and forth all the time.
Can HN name any niches that they feel should be filled with a similar product?