> What must have seemed like a benign social media strategy managed to fall foul of Facebook’s algorithmic censors, which labeled sections of the declaration hate speech and removed the posts.
Do we know that this was the action of an "algorithmic censor"? I mean, one of the automated censors? My impression was that Facebook still has many human censors [0] in its employ, many of them outsourced in non-U.S. countries. It's possible that a human user flagged this Declaration of Independence post, and a human censor approved the flagging/censoring of it.
Yes, I know that "algorithm", broadly defined, could include a process in which a human user flags something for a human censor, but that's not how "algorithmic censors" is typically used in this context.
Does it matter? There was a problem, they apologized and corrected it.
FB has been under intense scrutiny for providing a platform for hate speech, so why should we be surprised that they use automated means to decrease the lead time between posting and removal?
That phrase is hate speech, it just happens to be contained within one of the founding documents of the US. While it is a worthwhile goal for the algorithm to understand the context of it - I don't think we're quite there yet, and I would rather have a few false positives than the avalanche of shit that would exist without it.
As to whether a human removed it, I suspect that from the perspective of the person doing it, it's better to err on the side of caution than lose your job; especially a non-US citizen.
Yes, it matters. People need to have a better understanding of how these processes work, and not just to think everything related to big-tech is solved by machine learning.
I didn't say it was. I said that we're not there yet, and implied that there will be mistakes along the way.
Can you tell me what crime was committed here? As noted in the Vindicator's own article:
> While unhappy with Facebook’s action, the editor reminds readers that Facebook is a business corporation, not the government, and as such it is allowed to restrict use of its services as long as those restrictions do not violate any laws. Plus, The Vindicator is using Facebook for free, so the newspaper has little grounds for complaint other than the silliness of it.
> The problem The Vindicator faces is that it has become dependent, perhaps too dependent, on Facebook to communicate with local residents and to promote the newspaper. Some Vindicator stories posted on thevindicator.com attract thousands of page views, but usually only after links to them are shared on Facebook. Plus, many bits of information are shared by the newspaper through Facebook alone.
Sorry, but what are we debating here? I never said that Facebook commuted a crime, or otherwise alleged that they acted maliciously. I asked a question about what the article meant by “algorithmic censors”, because I suspect the Guardian is conflating FB’s automated content flagging with its human censor system. Because it would be more of a concern (especially from a technical viewpoint) if FB’s automated system was attempting to discern in this kind of situation.
A debate could be made over whether FB invests enough in its human censor system, perhaps, but I’m not getting into that here. BTW I didn’t and can’t downvote you — HN disallows authors from downvoting direct replies.
Basically a case study in why automated/preemptive content filtering is a bad idea. Let users report bad content, investigate it once it reaches a certain number of reports.
I certainly am not in favor of censoring historical documents, but as the article mentions, the declaration blames the British for stirring up the "merciless Indian Savages whose known rule of warfare, is an undistinguished destruction of all ages, sexes and conditions". That's a pretty hateful thing to say especially when you realize that colonists killing Native women and children was if anything more common than the reverse.
Satan almight. I'm glad you've looked back on a 240 year old document and determined it to be hateful according to modern societal standards. This is clearly a reasonable position to take because, duh: Fuck context, amirite?. You're doing god's work.
Do we know that this was the action of an "algorithmic censor"? I mean, one of the automated censors? My impression was that Facebook still has many human censors [0] in its employ, many of them outsourced in non-U.S. countries. It's possible that a human user flagged this Declaration of Independence post, and a human censor approved the flagging/censoring of it.
Yes, I know that "algorithm", broadly defined, could include a process in which a human user flags something for a human censor, but that's not how "algorithmic censors" is typically used in this context.
[0] http://www.latimes.com/business/technology/la-fi-tn-facebook...