Whitelisting

Following last week’s post about content verification, here is a second possible solution to the problem of imperfect copyright infringement detectors: whitelisting. For those of you who do not know, whitelisting is the practice of approving a person to use material they would otherwise not be allowed to. It is the opposite of blacklisting: restricting people from using material which is normally public. Essentially, when you blacklist, you give a list of people who can NOT use/access the material; everyone else is allowed. Consider it like a ban list on a website. Conversely, when you whitelist, you give a list of people who CAN use/access the material; everyone else is not allowed. Consider it like with Google Docs, where you can set it so only specific people can view or edit a document. The question is, should a copyright infringement detection system use whitelisting or blacklisting? There is really only one thing to consider.

Will the majority of people be allowed or disallowed to use the material? If most people are allowed, there should be a blacklist. If most people are disallowed, there should be a whitelist. Whichever the case, whitelisting or blacklisting must be a manual process. It grants exceptions, which means there should be few. Still, the fact of the matter is that manual processes are more work. It’s why Content ID cannot be manual: there is too much material to go through by hand. In order to make whitelisting/blacklisting feasible, users should only be manually considering a few users for approval/denial.

While US law is generally “innocent until proven guilty,” copyright law is the opposite. No one is allowed to use someone else’s material without their permission. Without their permission. Sound familiar? Yes, copyright is essentially a whitelist already. In fact, whitelisting is a possible answer to the problem presented last week. Who should be the content holder when multiple people have the rights to material? Trick question: all of them.

So, the solution seems simple enough, right? Combine content verification with whitelisting, and the problems are fixed. Only people who own material can be content holders and stop people from wrongfully using their content. If they give someone the right to use their material, they just add them to the whitelist and never have to worry about having to manually approve each and every video they approve. It’s a simple fix, and everyone’s happy! …well, it would be if both content verification and whitelisting were perfect.

The truth is, verification isn’t perfect. Once someone is approved as a content holder, what they upload isn’t necessarily monitored to confirm that it’s theirs. They can claim someone else’s material, whether intentionally or unintentionally, and they would be the content holder of that material until it’s contested.

The truth is, whitelisting isn’t perfect. Say you produced a song and let a game use it (think Guitar Hero, Rock Band, Dance Dance Revolution; even games like Grand Theft Auto have radio stations with real songs). If that game company allows people to upload videos of the gameplay, it would include in-game music, which is copyrighted music. Since it is part of the game, the game company determines whether it can be used or not. However, the automated system would flag it as the music, and the whitelist wouldn’t transfer… Add onto that the fact that sometimes you don’t need permission to use copyrighted material (i.e. Fair Use), and there’s quite a few odd cases.

Now, does that mean it’s not a good idea? No. A combination of content verification and whitelisting is a very good first step. If done correctly, it will make the system near-perfect, with only a few problems which can be manually caught. Can it be improved? Perhaps. Perhaps with difficulty. Perhaps easily. Regardless, the minimum requirements of a good copyright infringement detection system is a mix of content verification and whitelisting. Without it, you get situations where people flag their own content.

Looking forward, how can we counter some of these problems? Is there a good way to monitor claimed content? Should whitelisting be transferable, and how so? How do you account for Fair Use in copyright?