Twitter accused of dodging Brexit botnet questions again

Twitter accused of dodging Brexit botnet questions again
From TechCrunch - January 26, 2018

Once again Twitter stands accused of dodging questions from a parliamentary committee thats investigating Russian bot activity during the UKs 2016 Brexit referendum.

In a lettersent yesterday to Twitter CEO Jack Dorsey, DCMS committee chair Damian Collins writes: Im afraid there are outstanding questions that Twitter have not yet answered, and some further ones that come from your most recent letter.

In Twittersletter sent last Fridaythe company says it has now conducted an analysis of a dataset underpinning a City University study from last October (which had identified a ~13,500-strong botnet of fake Twitter accounts that had tweeted extensively about the Brexit referendum and vanished shortly after the vote).

And it says that 1% of these accounts were registered in Russia.

But Twitters letter doesnt say very much else.

While many of the accounts identified by City University were in violation of the Twitter Rules regarding spam, at this time, we do not have sufficiently strong evidence to enable us to conclusively link them with Russia or indeed the Internet Research Agency [a previously identified Russian trollfarm], it writes.

Twitter goes on to state that 6,508 of the total accounts had already been suspended prior to the studys publication (which we knew already, per the study itself)and says that more than 99% of these suspensions specifically related to the violation of our spam policies.

So its saying that a major chunk of these accounts were engaged in spamming other Twitter users. And thatas a consequencetweets from those accounts would not have been very visible because of its anti-spam measures.

Of the remaining accounts, approximately 44.2% were deactivated permanently, it continues, without exactly explaining why they were shuttered. Of these, 1,093 accounts had been labelled as spam or low quality by Twitter prior to deletion, which would have resulted in their Tweets being hidden in Search for all users and not contributing to trending topics in any way.

As we said in our previous letter, these defensive actions are not visible to researchers using our public APIs; however they are an important part of our proactive, technological approach to addressing these issues.

Twitters letter writer, UK head of public policy Nick Pickles, adds that a very small number of accounts identified by City University are still active on Twitter and are not currently in breach of our rules.

He does not say how small.

tl;dr a small portion of this Brexit botnet is actually still live on

While Twitters letter runs to two pages, the second of which points to a December 2017 Brexit bot study by researchers at the Oxford Internet Institute, also relying on data from Twitters public streaming API, which Twitter says found little evidence of links to Russian sourcesliterally right after shitting on research conducted by researchers using our public APIsCollins is clearly not wooed by either the quantity or the quality of the intelligence being so tardily provided.

Cutting to the chase, he asks Twitter to specify how many of the accounts were being controlled from agencies in Russia, even if they were not registered there.

He also wants to know: How many of the accounts share the characteristics of the accounts that have already been identified as being linked to Russia, even if you are yet to establish conclusively that that link exists.

And he points out that Twitter still hasnt told the committee whether the 13,493 suspected bot accounts were legitimate users or bots; who controlled these accounts, what the audience was for their activity during the referendum, and who deleted the tweets from these accounts.

So many questions, still lacking robust answers.


Continue reading at TechCrunch »