?:reviewBody
|
-
A small newspaper in East Texas ran afoul of Facebook's algorithms in the summer of 2018, after their posting portions of the Declaration of Independence for the July 4 holiday was deemed to have violated the social network's hate speech standards. In the run-up to Independence Day, the Vindicator newspaper, which serves Liberty County, about 40 miles north-east of Houston, published the text of the Declaration of Independence on their Facebook page, in twelve separate posts between 24 June and 4 July. According to The Vindicator (information that was later corroborated by a Facebook spokesperson) the tenth excerpt did not appear on the newspaper's Facebook page, despite being scheduled to publish on 2 July. The text of that post read as follows: In their article (published under the headline Facebook's program thinks Declaration of Independence is hate speech) The Vindicator reported that: The following day, Facebook restored the post in question and apologized to The Vindicator in an email the newspaper quoted as follows: A spokesperson for the social network told us that The post was removed by mistake and restored as soon as we looked into it. We process millions of reports each week, and sometimes we get things wrong. The Vindicator's managing editor Casey Stinnett speculated that Facebook's algorithms were alerted by the use of the phrase Indian Savages in the excerpt from the Declaration, something that a spokesperson for the social network confirmed to us in an email, noting that context is crucial in how Facebook evaluates potential hate speech and bigoted slurs. That the phrase Indian savages could be hate speech, depending upon the context in which it appeared, was not disputed even by observers who were critical of Facebook in this episode, such as the libertarian web site Reason: Neither Facebook's algorithms nor their employees labelled the Declaration of Independence, as a whole, as hate speech. Headlines in some news reports (including The Vindicator's own) have been misleading in this respect, suggesting that the social network had determined the entire text of the document was in violation of Facebook standards. However, the other eleven out of twelve excerpts from the document did not alert even Facebook's automated standards filters.
(en)
|