Irrespective of this particular case, this whole area (not just YouTube) of automated content flagging needs serious attention from a number of standpoints. Here's an example of what has happened to me (and many other people). I uploaded a video of mine that included a segment of old, definitely public domain material. Shortly thereafter, my entire vid was flagged by YouTube's Content ID. Why? It took some digging to figure out, but it turns out a Content ID partner had uploaded a video of their own that happened to include a section of the same public domain material I had used. This apparently made it look like my video was infringing, since Content ID assumed the section of my vid that matched their vid was in violation. Wrong! But Content ID partners get the assumption of being correct, and there's no way for an average user to assert that something is public domain a priori. I was able to get this reversed by careful explanation on the appropriate forms, but I wonder how many people would just throw up their arms and say, "To hell with it!" and not bother? This is not an easy situation to solve, but the explicit assumption that Content ID partners are correct and that takedowns or other actions are immediate -- with a protest required to get blocks, etc. removed after the fact, strikes me as increasingly problematic. - Lauren
8 plus ones
Shared publicly•View activity
View 4 previous comments
- It does need to be mostly algorithmic; human intervention doesn't scale. As of four months ago, they were up to 72 hours of video uploaded per minute. For certain kinds of content, you do want it to be killed as instantly as possible.
OTOH, Google's general reliance and trust of algorithms is also a problem. Someone on here noted this morning that he got a Google+ warning dialogue about spamming, which he absolutely doesn't do. The opaqueness of the system, and the apparent inability to report algorithmic failures to real live humans, is a serious problem.
https://plus.google.com/u/0/108450240479866108610/posts/aJAMsydUDSHSep 5, 2012
- Hrm... Any possibility that YouTube/GOOG could keep track of what clips are identified as public-domain (or otherwise non-infringing), and stop flagging them?Sep 5, 2012
- You can't directly infer which segments of a video are public domain when they're included with other video material. If you had a comprehensive database of public domain material ... but that's a tall order. And of course this doesn't address fair use at all.Sep 5, 2012
- I can imagine a system where everyone who registers a video with YouTube, asserting ownership rights, must also fill out a detailed schedule of exemptions (0:32..1:27, 3:42..4:03, etc, are public domain clips). Potentially quite a lot of additional effort. And also, the system should inherently allow for fair use -- some percentage or absolute number of seconds of a copyrighted video is always allowed. But that's all hand-waving; I have no idea how it would all work. Something obviously needs to be done; Leo Laporte constantly comments about how including the tiniest fragment of a YouTube video in one of his shows, for legitimate journalistic purposes, invariably triggers a takedown when the show gets posted to YouTube.Sep 5, 2012
- As I said: public domain, or otherwise non-infringing. Which might include fair use.
Though in truth, courts have a hard time sorting that last bit.Sep 5, 2012
- Discussion of copyright enforcement bots starts at about 32:00: http://twit.tv/show/this-week-in-google/162Sep 6, 2012