How We Almost Gamified Copyright Infringement Detection on YouTube (& Ideas for Fake News)

Over the past decade YouTube has constructed one of the most efficient and useful copyright management systems [dramatic tone] EVER CREATED BY MAN. The company works closely with rights holders of all sizes to identify fan-uploaded clips which may contain third party video or audio assets, and presents a set of simple business choices – take the infringing clip down (or “mute” the music, if it’s just audio infringement); monetize the clip on behalf of the rights owner; or just track the clip but take no action. The path of least resistance would have been to build a DMCA Takedown Engine but instead the talented team at YouTube solved an enormous challenge in a much more productive manner.

While this system was in its infancy, a group of us were brainstorming creative approaches to copyright management outside of this more complex design. One engineering lead started riffing on an idea that has always stuck with me: build a version of Hollywood Stock Exchange (HSX), but for copyright violating content. HSX is a site where you can trade movies (and actors & actresses) like stocks — the value is related to the box office dollars. So, for example, if you think the next Star Wars movie is going to be a huge hit, you’d “buy” virtual shares in the movie and sell at a higher price once it breaks all attendance records. At scale HSX may also function as a prediction market if you believe the participants as a whole are creating an efficient view into the enthusiasm (or lack thereof) the viewing public will have for a film.

Here’s what we back-of-the-napkin’ed for YouTube:

  1. When a video was uploaded into the system, its “IPO” price would be determined by a number of algorithmic factors such as performance of other videos from that account, the type of content it contained, the mix of referral sources and so on.
  2. Over time, a video’s “price” directly correlated to its view count. So if you saw a video with a low viewcount that you thought was going to go “viral,” you’d want to “buy” it in hopes that its value would skyrocket once it became popular.
  3. However, if a video was eventually removed from the system because it was infringing on someone else’s copyright, the “price” would go to zero. And in the process, you’d lose all the virtual currency you had invested in that video.
  4. Participants (registered YouTube users) would start with a slug of virtual currency with which to start their video portfolio and you’d go from there.

Using this mechanic, a video with a viewcount that was increasing quickly but did not have many buyers *should* signal there was a risk associated with the video being removed. Just as YouTube relied upon its community to proactively flag content which violated the site’s ToS around nudity, violence, and so on, perhaps this game mechanic could assist in the signaling of what *may* be copyright infringement. Basically we talked about this once or twice, usually with alcohol involved, but never implemented.

[note: there are a number of reasons we didn’t want to be in the business of proactively reviewing content for copyright determination. YouTube’s practices around the DMCA were upheld by courts. Also hold aside questions around whether this game would result in incentives to spam the viewcount of videos you “owned,” etc.]

Sooooo, Facebook and Fake News.

There are lots of smart discussions online these days about whether (and how) Facebook should evolve to help counter fake news from spreading on its platform. Most ideas involve some combination of URL blacklists, machine learning as to what fake content looks like and user flagging. It’ll be interesting to see what Zuck and team choose to do – and I think they have a responsibility because it’s directly linked to the trust many users have in Newsfeed.

But how would you turn identifying Fake News on Facebook into a game? Would there be a way to gamify fact-checking? Would be much more difficult in some ways than the YouTube scenario because Fake News seems to get stuck within echo chambers where its consumers *want* it to be real and may otherwise be disincentivized to play the adjacent metagame.

Any ideas?