ADL – Anti-Semitic Posts on Twitter

A recent report released by the Anti-Defamation League (ADL) claims that the organization tracked some 4.2 million anti-Semitic tweets between January 2017 and January 2018. The Media, both traditional and social, repeated the claims widely, giving an impression that Twitter was a hotbed of anti-Semitism.

As with all matters concerning “hate,” in the Media, a closer look is warranted.

First off, anti-Semitism is a very real and a very dangerous social phenomenon. While even one hateful tweet is one too many, the definition of “hate” is nebulous at best and often tailored to the goals of the people defining it. To fully understand the reality on the ground it is imperative to review the definitions and methodologies used in creating  such reports. Let’s check the facts.

The ADL’s report, Quantifying Hate: A Year of Anti-Semitism on Twitter (no authors are named) claims that a review of English language tweets over 2017, using both computer algorithms and human review, yielded 4.2 million anti-Semitic tweets and re-tweets by 3 million users.

“The current findings are based on a complex Boolean query designed to identify language frequently used by anti-Semites.

The query was broadly written to encompass obvious expressions of anti-Semitism, including classic anti-Semitic stereotypes; code words and symbols sometimes used in an anti-Semitic fashion; and also subtle references to anti-Semitic conspiracy theories.”

The query was indeed “broadly written” and was “designed to detect anti-Semitism in the following categories:

  1. Classic anti-Semitic stereotypes (e.g. references to Jews as greedy; controllers of banks, media, governments and academia; under-miners of culture and racial purity; cursed for killing Jesus; etc.)

  2. Positive references to or promotion of known anti-Semitic personalities, authors, books, articles, videos and podcasts

  3. References to anti-Semitic conspiracy theories (e.g. Jewish control of the Federal Reserve; the existence of a “Zionist Occupation Government,” etc.)

  4. Holocaust denial

  5. Epithets used for Jews (e.g. “kike”) and against Jews (e.g. “goddamn Jews”)

  6. Code words and anti-Semitic symbols such as the “echo symbol” (“((( )))”)

The current report includes criticism of Israel or Zionism when such criticism makes use of classic anti-Semitic language or conspiracy theories, or when it ascribes evil motivations to significant numbers of Jews. General criticism of Israel or its policies is not counted as anti-Semitism.

That covers a lot of territory, and, as the report’s Table of Contents indicates, includes several topics that may be, at best, tangentially associated with actual anti-Semitism.

ADL TOC

Including references to Harvey Weinstein, George Soros, Zionism and “globalists” will undoubtedly uncover many genuine anti-Semitic references but will also include many legitimate criticisms of cultural and political movements and players.

Let’s dig a little deeper:

  • As of 2017, Twitter had 330 million active users, of which 100 million were active daily
  • Over 500,000,000 tweets are sent every day, or 182 BILLION a year. A 2012 survey estimated that 38% of all tweets were in English, though this percentage has most certainly changed since then. The ADL study limited its sample to English language tweets only.
  • The United States makes up 21% of all Twitter users worldwide. Lacking demographic data later than the 2012 report referenced above, and excluding all tweets from other English speaking countries, (The UK, Canada, Australia, New Zealand, etc.) and tweets in English from other countries, a simple one-for-one extrapolation (21% of users accounting for 21% of tweets) would suggest that Americans sent more than 38 billion tweets in 2017.

    Obviously, not all American tweets were in English, but even 75% of this incredibly low-balled estimate would yield 28 billion tweets, of which the ADL found 4 million to be questionable, or .014%.

  • An ADL press release notes that, whatever the actual number of English tweets studied for 2017, only 55,000 “were manually reviewed for the presence of anti-Semitism.”
  • A 2017 peer-reviewed study estimated that between nine and 15% of all Twitter users were computer generated “bots.”

In short, the methodology used in the ADL’s Twitter report was, at best, “broad,” and at worst, unreliable. The report includes a disclaimer referencing “the rise of ‘QAnon’ conspiracy theories” that illustrates just how broadly the term “anti-Semitism” was interpreted:

“The vast majority of QAnon-inspired conspiracy theories have nothing to do with anti-Semitism. However, a small percentage of tweets referencing QAnon also referred to Israel, Jews, Zionists, Rothschilds … , or George Soros. 

This study’s methodology does not allow us to determine how many of the QAnon tweets containing those terms actually expressed anti-Semitic sentiment, but an impressionistic review revealed some troubling examples.”

“Impressionistic reviews” revealing “some troubling examples” are not the stuff of hard research. It should be noted that the Anti-Defamation League, like the Southern Poverty Law Center (SPLC), is a private “advocacy group,” which by its very definition means that the organization is “advocating” a particular point of view.

Online tax records indicate that the ADL averaged between $50- and $60 million dollars in donations a year between 2011 and 2016, with 2% to 2.5% of that money going to “program services,” such as this anonymous survey, and 40% to 45% going to “Executive compensation, other salaries and wages.”

With tens of millions of dollars in compensation at stake, a strong financial incentive to interpret anti-Semitism as broadly as possible cannot be ignored.

As we noted at the beginning of this post, the Media, in all its forms, gladly repeats the claims of the ADL and SPLC without performing any review. Lurid claims of “hate groups” everywhere and anti-Semitism on the rise make for profitable click bait. The Media also have an undeniable financial incentive for promoting such “reports” without ever asking to see the evidence.

Again, anti-Semitism is a very real thing and must not be tolerated in any way, but lumping criticism of George Soros’ political activities and re-tweets of anonymous QAnon conspiracy theories is not a legitimate method of documenting it.

Research for the report was allegedly performed by two ADL in-house organizations, the ADL’s Center on Extremism and Center on Technology and Society, with no external peer review or oversight.

If the data is good the results ought to be easily replicable independently. Anonymous reports generated by in-house organs simply do not meet basic research standards.

Big claims demand big proof and the ADL needs to show its work. A poorly estimated 4 million suspicious tweets out of tens of billions posted in 2017 is statistically insignificant.

Prove it or remove it.

Advertisements

Tags: , , , , , ,

One Response to “ADL – Anti-Semitic Posts on Twitter”

  1. Chris Wiley Says:

    The ADL, like the SPLC, is a left wing hack organization, not a non partisan observer of “hate”. Their report “Murder and Extremism in the U.S. In 2017” was the usual conflation of killings by prison gangs, ex boyfriends or family members with actual extremist attacks. Like the SPLC’s map of phantom hate groups, anyone who spends 5 minutes reading this report can see it is fake news (an Islamic convert who killed his roommates for offending his religion is classified as a “white supremacist” attack). Of course, the complacent media, fully enveloped in the national panic over “white supremacy”, uncritically cited the report….and blamed Trump. Their bottom line demands it.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s


%d bloggers like this: