Here's how it will work. Offending content will have to be identified by rights holders, either by uploading it themselves or identifying it when someone else has without permission. YouTube will then create a hash file to make sure that that exact same file is not uploaded again. That's the crux of the issue right there. The big rights holders want YouTube to block every instance of their song or video content being uploaded, whether it's the identical file or not.

It's opt-in copyright protection by blocking repeat upload of the same exact media file. That's nowhere near what the big rights holders want. What they really want may be to kick the legs out from under the media record-breaker that is GooTube.

The new service will only block content if blocking the content is what rights holders want, of course. YouTube says it can't assume it knows. Some may want to promote that content once it's discovered, or to monetize it with revenue sharing of ad money. That seems possible, in theory, though the real impetus behind this whole system is to pull content.

If this is the kind of thing you're interested in, the video identification page is an interesting read. A year ago it looked like YouTube was going to license a third party filtering service, like Audible Magic, but all the options there may have proven technically imperfect. Perhaps unable to nail the problem technically, it appears that Google has chosen to take a legalistic approach to the problem.