There is a massive chance to disrupt this market right now. But bandwidth costs alone make it difficult to challenge google in this space.
I think this service is fundamentally very hard to produce. For all its faults with monetization, it's not clear it can be done better (beyond toy scale). It's clearly not profitable to have humans match videos with advertisers, and computers are inevitably going to screw up, like in this instance. edit: it seems this case is just a basic policy issue, though that policy is a consequence of trying to avoid attempting to solve this problem for the long tail of small time uploaders. I just assumed it was another false positive based on the title.
Keeping advertisers and uploaders happy seems like an almost impossible task. Google may not be doing very well at it, but that doesn't mean it's easy to do better.
The problem is actually bandwidth costs; unless you're at the scale of Google (where bandwidth and storage is really cheap for you), advertising alone wouldn't even pay for the service itself, let alone pay the uploaders.
I imagine a world where creators pay someone for video distribution, but also plug-in advertising from third party ad agencies that have learned to match ads to content due to advantages in machine learning. Same as text today.
In minds of general population, YouTube is a search engine. The was majority believe that YouTube aggregates the content all around the Web. Why wouldn't they? The content shows up there minutes after it originates somewhere else. It's almost impossible to prevent it from gaining any traction on YouTube. Another company just tried to go against them. You can read how it went here https://vid.me
It wouldn't even have to be perfect like a lot of publishers usually demand, most youtube videos would be fine if the payment would simply authorize the user to stream and download the video as they wish.
An arbitrary youtube video [1] is 56.69 megabytes (7m36s of 720p MP4). AWS outbound bandwidth, which I'm told is outrageously expensive, costs $0.050 per GB, so a single view costs $0.0028 to serve.
Seems to me bandwidth isn't an insurmountable problem - if you can get the mix and volume of viewers, creators and advertisers on board, and make $.10-$.30 per view, and you can keep your overheads for engineering, support and moderation low.
You always forget the huge(?) amount of un/low monetizable content/traffic/countries.
They are still growing exponentially.