PRESS RELEASE: Shurat Hadin Facebook experiment on Kill Israelis and Kill Palestinians


FOR IMMEDIATE RELEASE: A number of people sent us a video of an experiment by an Israeli NGO, Sherut Hadin. In the video they created two hate pages, once against Israelis and one against Palestinians. Very similar content was posted to each page. They then reported both pages. The anti-Palestinian page was promptly removed while the anti-Israeli one was not. We’ve contacted Facebook about this a few hours after the videos surfaced, Facebook have investigated, and they have now also closed the hate page targeting Israelis.

Facebook’s public statement provided to OHPI is as follows: “Facebook does not tolerate hate speech, including against people on the basis of their nationality. We review all reports and take down such content. Both these pages have now been removed from Facebook.”

We’re pleased to see Facebook has responded promptly to this. We have queried whether the different initial responses were simply random luck, or whether there was something in the nature of the content, the context, or the response by the page administrators which led to the different initial responses.

The sample of 2 used in this experiment is far too small to draw any general conclusions. Research by the Online Hate Prevention Institute based on 16 items promoting violence against Jews on Facebook and 91 items promoting violence against Muslims on Facebook show that content promoting violence against Jews is removed far more often.

OHPI’s CEO, Dr Andre Oboler, stated that, “our own research indicates that, at least in English language content, the promotion of violence against Jews is taken down by Facebook more readily than content promoting violence against Muslims. As we have said to Facebook while discussing this issue, in both cases there is still much room for improvement.” Oboler recently told the BBC that online hate become far worse in nature in 2015, he explained that, “where previously people hid behind pages and fake accounts, by the end of 2015 many people felt their hate was acceptable and were comfortable posting it under their real name or their regular social media account”.

Of all the antisemitic hate in a sample of over 2,000 items, only 5% of it related to the promotion of violence against Jews expressed through Facebook. In a sample of over 1,000 items of anti-Muslim hate, 9% related to the promotion of violence against Muslims through Facebook. The full research into antisemitism will be released on Holocaust Memorial Day, January 27th 2016. The full research into anti-Muslim hate will be released in March 2016, but an interim report is available. We recently presented a preview on both sets of data at the United Nations in New York.

Share this press release:


NB: Scroll down below the video for further commentary

The Sherut Hadin video:

Update (6 January):

Someone wrote on our Facebook page that our response “misses the point”. They explained that as they see it, the point “was not that FB did not notice both of them” but that in the case of the kill Israelis page, “FB reviewed the page, kept it up, and sent a message saying it did not violate community standards – until later shamed publicly” by the video. This was, they said the opposite of the response to the anti-Palestinian page where in response to the initial report, Facebook “closed down the page and sent a message saying it violated community standards”.

The fact that the anti-Arab one came down the first time is the only part of this which we find surprising, and that seems to be down to luck. Some low level person in Facebook was likely having a good day and made the exceptional decision to respond (at first instance) by removing the incitement. It can happen, but it hardly ever does for any category of hate, be it antisemitism, anti-Arab, anti-Muslim, anti-women, homophobia etc. We deal with all of it, and in all cases the initial response tends to be to reject the content, and only later is some of it removed. What is removed is not nearly enough regardless of the type of hate we happen to be looking at. Campaigns in the media certainly do help get some of the remaining items removed. So in summary, the process documented in the video for the anti-Israel page, including what we then reported about it coming down in review, is all quite normal not just for antisemitism but for all types of hate on Facebook. The anti-Palestinian one coming down the first time is the fluke.

What this means is that claims Facebook is institutionally biased against Jews or Israelis because they handle reports about antisemitic content less sympathetically than reports about other forms of hate speech, are simply untrue. People are making these claims because the video, based on a sample of two cases, raises it as a question. While the video is careful to avoid making it a statement (presumably to avoid being defamatory) people commenting are not being so careful… and that may have been the intent.

Facebook does need to become better at removing antisemitism, and they need to become better at removing all other hate speech. We have no issue at all with that. Facebook specifically needs to become better at removing content which is inciting violence. Whether it is Israelis, Palestinians, or refugees in Germany. More needs to be done about incitement and there is no doubt about that. The issue is that a claim implying they are institutionally antisemitic because they are less likely to remove antisemitism than other forms of hate… would be over reaching and untrue. The evidence when collected scientifically and in a more reasonable sample size than “one of each”, as provided in this video, simply doesn’t support that claim. That’s not to say Facebook is immune to liability for not doing enough, across a wide range of hate types. What it does mean is that basing such a claim on competitive victimhood is not only a bad approach that supports a “them and us” attitude rather than dealing with the larger problem, but in this case it is also not supported by reasonable evidence.