Why do you feel that the site's cut could be as low as 10%? It's very expensive to host and serve high-quality video, and you've not described anything that would lower the site's costs relative to something like YouTube (which I don't believe is stunningly profitable).
But I think the answer you're looking for is that the publisher gets to make money as a facilitator, rather than a program scheduler. The model doesn't make users the customers, it makes content creators AND advertisers (together; doesn't work without both) the customers. If advertisers want to work with a content creator on your network, and they want that content viewed with your network, then they will have to make a deal to join the network. That's passive revenue for the site, on top of the percentage of the ad buys. Essentially, you put the "no free rides" subscription model on the advertisers, and then let content creators decide who gets free rides for THEIR subscription models.
On top of that, there's a whole other revenue stream in developing "networks" and selling "social" functionality to advertisers and content creators that get value out of interactivity (ex: monetizing the inclusion of advertisers in "best value" ad-buy deals that target a "content creator type" like "these ads go well if you provide cooking content") . And we haven't even talked about "whales" and monetizing the customer base with twitch-like engagement mechanisms. I cringe at the thought of some of it, but if the creators can make money off of it, more power to them! And a little percentage of their windfall to the site! ;)
Of course, there's always a new way to make a buck, and there's some very deep pockets in advertising, so I don't fear for the innovation that could go on, in a site run as described.
Once the network is established as a good place for creators to be able to actually make money, you attract the good creators. They aren't getting pennies on an ad (or, if they are, that's what they were happy with; we all start somewhere/traffic volume is important); they're getting the full ad buy minus some percentage for the site (more like 10% -20%, less like 50%-90%).
As the site, acting as a publisher, you get to dictate the ad network, and set boundaries for keeping things profitable (neo-hayes-code "cost of doing business" gatekeeping). As content creators, they get freedom of association and deal-making (which includes refusing to serve ads that they disagree with).
Then, when you have a solid network of solid content, you layer in a subscription model which could allow you to sensibly bundle content. While each content creator could set their own subscription tiers (including free ones, for the purposes of advertising their own content or giving things away for free because they are such absolute champions), they would be able to opt-in certain tiers to be "bundled", which would allow users to pick and choose content creators that they want their subscription bundles to apply to. This is where you get to the first rumblings of reasonable payment for journalism work. Because, as far as I know, it's almost impossible to get people to pay for journalism, no matter how good it is. But if you can add it in to a bundle, while still letting people choose their outlets/sources, you can subsidize the journalism with the media profit (like how we do it on TV).
Of course, to facilitate that, the site would need some robust utilities for writing new articles, such that it might be indistinguishable from a WYSIWYG website creator, but that's not something that's insurmountable. Especially if the site is profitable.
Anyway, that's my swing. I've been working on it, myself, but it's not like I ever intended to run a social media app. I certainly don't have the funds to run it, myself, and it's an untested model so investment would be very surprising. So just POC stuff and collecting feedback from content creators. It all seems very promising, but I still haven't answered the question of 'will it make money?'