Ooh, nice one. That's an even better example than Beethoven.

Even better still: there's a song which consists of 4 minutes and 33 seconds of silence. That's it - silence.

"Your latest video upload contains 5 seconds of stunned silence, which has been identified as an extract of 4'33". This extract is copyrighted. Your video has been deleted."

The creative elements of Cage's work are that it's performed as a piece by an orchestra, and presented as a work (eg a specific length, sold as other works are). Your any-[other]-length-of-silence doesn't infringe. If it did then Cage's piece would lack the artistic distinctiveness needed to make it a work for copyright purposes.

Just having a urinal doesn't infringe on Duchamp's "Fountain", not even if it's the same model, only if it is presented as artwork does it become a copy of Duchamp's "work".

You know that, the lawyers know that, but how do you expect the content filtering mechanism to know that???

True, for 4'33'' there is a simple rule that they probably follow - ignore silence :). But for Fountain (if it ever came up) it's hard to imagine that the difference between a protected copy and a non-protected similar image could really be automatically discovered.

This is it. A human being can identify the context. Taking Fountain as an example -- a photo of a urinal on a building merchant's website is an illustration of the product they're selling, not an infringement.

But the filter doesn't know about context, it just correlates two images... and you get "Comparison with copyrighted work 'Fountain', 75% match".

75% > 0%, so the filter says "non".

Isn't this:

> decisions to disable access to or remove uploaded content shall be subject to human review

intended to handle those cases? I'm not saying that it will be adequate.