I've always wanted to try a media content site (obviously based on open standards) where to avoid the whole DMCA mess you'd require all content to be relicensed under a CC license (of your choice).
That means if you don't own licensing rights, which is very obvious if you even include third party material, you can't upload it.
That means personal video is entirely safe, and anything you personally record is always safe, but you never get into the fair use mess that made youtube, vimeo, etc so bad in the first place.
I'd think such a project would work best if you were trying to promote original works, though - and serving video is not really that complex - the infrastructure around it is applicable to almost all media anyway. But that is some serious scope creep, but as someone who consumes most forms of digital media from freelancers and amateur content producers, it has to be incredibly annoying as someone who uses multiple services to have their art on deviantart, their music on soundcloud and bandcamp, their videos on youtube, their blog on tumblr, and their web games on newgrounds or armor games.
One major roadblock in content hosting, as the founder of imgur has came foward about on many occasions, is the raw size of the content database, plus the bandwidth costs. I think popcorntime has soundly demonstrated the capabilities of the torrent protocol to handle real time streaming - the startup time has significant overhead, but given sufficient health in the network can stream effectively. I'd definitely look into distributing any larger videos, audio, games, and photo galleries via torrents since they would all be CC licensed. Of course then you aren't talking about a website anymore since you can't integrate functionality like that in the browser, so it is kind of a tangential idea, but I primarily develop in Qt and think it would be a real realistic cross platform deployable application on every platform if wanted to go the app route. I could see there still being a website.
And then you get into monetization, which was youtubes problem. I think patreon and kickstarter are soundly proving that patronage is a perfectly viable business model. I find it baffling that the only service I know that does patronage or subscriptions is Twitch, but even that is just a very cut and dry $5 a month.
I guess in part that is because all the payment mechanisms out there are antiquated pains in the ass. And for a startup to try to integrate payments like that would be a huge legal mess. I'd rather just give users blockchain.io wallets and let them put money in on their own, like bitcointip or dogecointip do on reddit - that model has proven hugely successful. And as cryptocurrencies enter the mainstream, I think that kind of business would be prime to explode.
Instead of fixed payments, it would be integrated in all the content pages with every feasible way to pay - buying swag, patreon style per content gifts, kickstarter esque projects where you propose a project and get it funded preemptively, monthly subscriptions like Twitch, and one time quick donations. I would imagine users might want to still monetize via adds, but I would rather such a service be funded through, say, 5% of all patronage transactions than through ad networks, because that makes you beholden to their demands for how to run your service. It gives you immediate monetization metrics so as long as you can get some audience and if people actually funded the artists it would probably be profitable quickly.
But hey, that is just me. All my colleagues and mailing lists think I'm crazy. I keep throwing it around and if I ever met a few folks interested in trying it I'd definitely be up for giving it a major go - I think its the inevitable "maximal" solution to intellectual content.
Wow, that was quite the rant. Sorry HN.