Study that claimed majority of Copyright Directive opposition came from the US assumed all English-language tweets came from Washington, DC


Members of the European Parliament have been carpet-bombed with a "report" claiming that the historically unprecedented opposition to the pending Copyright Directive was the result of "US meddling in the EU lawmaking process," with 21 pages of alarming charts and figures to support this conclusion.

However, the report is completely wrong, because it relied on analytics provided by Talkwalker, without understanding the assumptions that Talkwalker uses to fill in missing data.


Specifically, when Talkwalker encounters a tweet whose location field is blank, it guesses which language the tweet is in, then assigns the capital city of the most populous country where that language is spoken as the account-owner's location. Every German-language tweet is reported as originating in Berlin, and every English-language tweet is reported as originating in Washington, DC.

The opposition to the Directive's Article 13 is correlated with technical knowhow, and that is correlated with privacy consciousness (hence opponents of the Directive are likely to have blank location fields) and also with discourse in English. Also, English is the most-widely spoken language in the United Kingdom, which is one of the 28 member states of the European Union (for now). These tweets are in English, and have no location specified, and will all be classified as originating in DC.


So Talkwalker's analytics falsely reported vast numbers of opponents to Article 13 were voicing their opposition from Washington D.C., including Julia Reda, the German MEP who has led the Parliamentary opposition to the Directive from her offices in Brussels and Strasbourg (the analytics also claim that Reda is based in Berlin, because she also tweets extensively in German, of course).


This vast, easily avoidable, obvious and sloppy error was compounded by still more outrageous instances of analytic malpractices. For example, the report's authors measured the location of tweeters who included English-language hashtags, rather than those in other European languages.

The document's authors are lobbyists with the German industry group Interessenverband des Video- und Medienfachhandels in Deutschland e.V., and they have posted their report for your own inspection. One of the authors is co-owner of the filedefense.de service.