I think youtube needs to consider backing off regulating political content.<p>The fact is politics and morality are inherently intermingled. One can use words like extremist, but sometimes the extremists are the "correct" ones (like our founding fathers who orchestrated a revolution). How could any system consistently categorize "appropriate" videos without making moral judgements?
I'm much a proponent of automation as anyone else. But I think right now Google is trying to do something way too hard. By looking for "extremist" material, they are basically trying to determine the intention of a video. How can you expect an AI to do that?
Let's look at the bigger picture. First, in March some newspapers find an extremist video. It has ~14 views and YT advertising all over it. They make a a big deal out of it. As a result YouTube looses ad clients and tons of money.<p>Then, as a response, they make an alg. They don't want people to call them a "terrorist platform" ever again. Hence they take down the videos.<p>Now, this algorithm is hurting the bystanders. IMO the real problem is a public and business reaction to the initial event.<p>And this peace of news is an inevitable consequence.
Something people need to keep in mind when parsing this story is that many of the effected channels were not about militancy, they were local media outlets. Local outlets that only gained historical note due to what they documented as it was unfolding.<p>In Syria outlets like Sham News Network have posted thousands upon thousands of clips. Everything from stories on civilian infrastructure under war, spots on mental health, live broadcasts of demonstrations.<p>Everything.<p>Including documenting attacks as they happen and after they have happened. Some of the effected accounts were ones that documented the regime's early chemical weapons attacks. These videos are literally cited in investigations.<p>All that is needed to get thousands upon thousands of hours of documentation going back half a decade deleted is three strikes.<p>Liveleak is not a good host for such outlets because it is not what these media outlets are about. Liveleak themselves delete content as well so even if the outlets fit the community it would not be a 'fix.'
i really don't know how to describe my feeling as a syrian when i know the most important evidence that witnessed the regime crimes were deleted because of wrong call. And it's really confusing how artificial algorithm get confused between what is is obvious as isis propaganda and a family buried under the rubble and this statement makes things even worse. mistakenly? because there is so many videos? just imagine that may happen to any celebs channel. Will youtube issue the same statement? dont think so.
What I don't like about those web giant services is, to get a human support, it requires to start social pressure like this.<p>If they fucked up something by automation, contacting to human support is hopeless unless you have very influential SNS status or something.
What about all the speech that's censored that doesn't have enough interest or political clout to make people aware of the injustice of its censoring.
Google (parent company of YouTube) already sees itself as the protector of the public's eyes and ears. They might be contrite now but they behave as a censorshipping organization.
I think YouTube really needs to hire more humans to review flagging of videos rather than leave it to a loose set of algorithms and swarming behavior of viewers. They assume wrongly that anyone who flags a video is honest. They should always assume the opposite and err on the side of caution. And this should also apply to any Content ID flagging. It should be the obligation of accusers to present evidence before taking content down.
Automation is the only real solution. These types of conversations seem to always overlook how normal people don't want to watch such videos. Do you want to spend your day watching this stuff to grade them?