Image credit: How To Win a Vaccination Debate
The media are promoting a new scientific paper on the vaccination debate. Their interest is undoubtedly driven by the study’s conclusion that “Russian trolls” (and by implication the Russian state) are amplifying this debate to promote discord in the US. The title describes this as “Weaponization of Health Communication.”
I am very cynical. After all, the media loves to dramatise these matters – and scientists are not immune to the temptation of taking advantage of this and the current political environment. The data the authors present is weak and has a far more reasonable explanation than the one they assume.
Yes, I may well be called a “Russian troll” or one of “Putin’s Useful Idiots” (and it wouldn’t be the first time) for expressing these doubts. But I have read the paper and this was helpful as it provides sources enabling me to do my own checking.
The paper is:
Broniatowski, D. A., Jamison, A. M., Qi, S., AlKulaib, L., Chen, T., Benton, A., … Dredze, M. (2018). Weaponized Health Communication: Twitter Bots and Russian Trolls Amplify the Vaccine Debate. American Journal of Public Health.
That’s just the abstract but here is a link to the full text.
The paper summarises its main claim about “Russian trolls” as:
“Russian trolls and sophisticated Twitter bots post content about vaccination at significantly higher rates than does the average user. Content from these sources gives equal attention to pro- and antivaccination arguments. This is consistent with a strategy of promoting discord across a range of controversial topics—a known tactic employed by Russian troll accounts. Such strategies may undermine the public health: normalizing these debates may lead the public to question long-standing scientific consensus regarding vaccine efficacy.”
The analysis relies on subjective judgment for defining a twitter account as a bot, but it does use two publicly available lists of twitter accounts (and tweets from these accounts) defined as inauthentic or false “Russian trolls.”
These sources are:
- “Russian troll accounts identified by NBC news” which allegedly documented “Russian interference in the US political system” (see Twitter deleted 200,000 Russian troll tweets. Read them here), and
- “Accounts the US Congress identifies as Russian trolls” (see Twitter’s list of 2,752 Russian trolls).
The evidence supporting their main claim is given in their Figure 1: Bots’ Likelihood of Tweeting About Vaccines Compared With Average Twitter Users: July 14, 2014–September 26, 2017. See below:
Tweets from the “NBC Russian Trolls” contain a higher incidence of vaccination keywords than tweets from the average twitter user. To be clear – this is not evidence of promotion of an anti-vaccine message (“Content from these sources gives equal attention to pro- and antivaccination arguments”). It simply shows these collection of tweets contained a higher than average reference to this polarizing subject.
I suspect a similar analysis of this collection of tweets would also show a higher than average incidence for other polarizing subjects in this collection. It is the nature of the tweet selection not evidence of a specific motive.
In fact this claim of “promoting discord” is so commonly used nowadays that it seems to have lost any meaning. Politicians now attribute this motive to much of the Russian social media – and to Russian mainstream media (eg., RT and Sputnik) news reports.
We should note that the authors did not attempt to justify the highly political allegation. They simply aligned themselves with the political message, but the senior author Broniatoski admits “we cannot say that with 100% certainty, because we’re not inside their head.”
Unfortunately, they did not consider for one moment other possible explanations for their results (that is highly unscientific and reveals a bias). I think this illustrates the power of the controlling or prominent political narrative. Anti-Russian hysteria is widespread in the US at the moment.
But there are more innocent motives for such tweets which a more objective analysis would have considered (see below).
The “guilty” tweets
I have looked through the database listing the tweets identified as from“Russian troll accounts identified by NBC news.” The incidence of reference to vaccination in the tweets from“Accounts the US Congress identifies as Russian trolls” was not much different to that for the “average user” so I did not consider them.
There were 203,451 tweets in this collection and I found about 100 (about 0.05%) included a vaccine keyword (vacc*). The paper gives examples of both pro and anti-vaccine tweets from this collection and mine were similar. These were hardly remarkable – indeed most of them were retweets. For example:
- RT @HealthRanger: Don’t miss this: #autism-vaccine link explained by doctors! Â https://t.co/L9ziemow6o Â #antivax #vaccines #adhd
- RT @ActivistPost: States are rushing to pass vaccine mandates before everyone realizes that they’re completely unnecessary at best, harmfulâ€¦
- RT @HealthRanger: Danish #documentary exposes widespread damage caused by HPV vaccine https://t.co/nuQqQ1u0XZ Â #health #vaccines #antivax #â€¦
- RT @HealthRanger: Never inject them. #antivax #vaccines #natural #health https://t.co/oY0XLqRkdH
- RT @pakalert: The Scary TRUTH About Vaccines (Satanic illuminati Vaccines Agenda Exposed Full Documentary) https://t.co/fxs8zOwVnV
- RT @WorldTruthTV: Robert De Niro To Produce Film Proving Vaccines Cause Autism | World https://t.co/telXZBWPRi https://t.co/VrApvqn62s
- RT @CobraCommans: Canadian scientists to test promising HIV vaccine on 600 volunteers @ANCParliament @My_AfricanUnion @AfricaHealthFor
- RT @GStein269: Perry talking about Drugs and Vaccines? https://t.co/lsxJN2Udcy
- RT @SanJosePost: #politics California’s vaccine bill passes Assembly, next hurdle: Gov. Jerry Brown
- RT @varadmehta: Having a vaccine truther chair a commission on vaccine safety is something that merits actual outrage. But media only has oâ€¦
- RT @blicqer: Major HIV Vaccine Trial Set to Begin in South AfricaÂ https://t.co/fPkW3XYV32 @TheRoot https://t.co/I5iRgU42Yn
The #VaccinateUS hashtag
The paper describes the #VaccinateUS hashtag as:
“designed to promote discord using vaccination as a political wedge issue. #VaccinateUS tweets were uniquely identified with Russian troll accounts linked to the Internet Research Agency—a company backed by the Russian government specializing in online influence operations.”
Again, it did not provide any evidence to support this allegation.
The authors claim these tweets “contain a combination of grammatical errors, unnatural word choices, and irregular phrasing.” I did not see this myself – the grammar in these tweets appeared to me to be far better than the average tweets I see. The authors did acknowledge that these “messages contain fewer spelling and punctuation errors than do comparable tweets from the general vaccine stream.”
Tweets with this hashtag are about evenly divided between pro- and anti-vaccination potions (“43% were provaccine, 38% were anti vaccine, and the remaining 19% were neutral”). It occurred rarely in the quarter million tweets.
While they appear to have been specifically written by the account holders or staff at the organisation behind them, rather than simple retweets, they hardly provide evidence for a motive of “sowing discord.”
Here are some examples:
- what will you fill when you get a disease that you could’ve been protected from? #VaccinateUS
- if we don’t have regular chek ups and get #vaccines-what’s the point of doctors’ work? #VaccinateUS
- open your eyes, people! It’s all government conspiracy plan #VaccinateUS
- our government cares only about money so itâ€™s profitable for them to say that #vaccination is necessary #VaccinateUS
- the production of a #vaccine is disgusting #VaccinateUS
- #VaccinateUS FDA state that #vaccines are safe
- #VaccinateUS For sure #vaccines work!
- God bless big pharma. You fools #VaccinateUS
Amplification of the anti-Russian hysteria
Ironically the charge laid at the supposed “Russian trolls” (that they seek to sow discord by amplifying existing electoral or polarizing debates) is actually typical of much of the reaction in our media to stories like this. In fact these media reports are aimed at sowing discord and promoting Russophobia. And, unfortunately, such anti-Russian amplification, or weaponization to use the language of the paper, comes from people I would have thought should know better.
This example from March for Science – a social media group formed after Trump’s election and aimed at mobilising scientists against anti-science policies of the new administration.
” Study finds that 93% of tweets about vaccines between 2014 and 2017 were planted by bots and Russian trolls with the aim of sowing division.”
The 93% is the invention of March for Science as neither the paper or The Guardian provided this figure. And the study did not “find” that Russian trolls were sowing divisions – that was the prevailing assumption they started with. March for Science is simply crudely (very crudely considering their invention of 93%) amplifying the anti-Russian narrative and contributing to weaponization of social media against the Russian Federation.
Bringing this home, the NZ Facebook page Science Community New Zealand reposted the March for Science claim. Here we have social media accounts claiming to be pro-science amplifying an outright lie on social media.
Update: Science Community New Zealand has now removed the offending post – a good sign perhaps.
I am disappointed at such a naively political falsification from organisations which is meant to be promoting science. It does show how persuasive the current anti-Russian hysteria is – but it is especially disappointing to see people who should know better succumbing to it. Or, perhaps, I have been fooled and the real motives of March for Science and Science Community New Zealand have been far more questionable right from the start.
A more realistic motive for these tweets
The motive given by the study’s authors, and usually promoted in the current mainstream media narrative (sowing discord to weaken US society), really does not hold water. That strategy could more legitimately be attributed to ordinary US twitter users who indulge in tweeting on controversial subjects in far larger numbers. Anything added by these Russian trolls is minuscule. If the Kremlin genuinely has such a strategy it should be judged a pitiful failure.
But what about this shady company Internet Research Agency based in St Petersburg? I have no doubt it exists and that it is planting material in social media like Facebook and Twitter. Presumably it is also setting up non-authentic or fake accounts for this purpose.
However, the paper’s claim that it is “a company backed by the Russian government” is not supported by any evidence at all and is typical of the way our media continually falsely claims that Russian individuals and entities are connected to the Kremlin or “close to Putin” – simply because of their ethnicity.
While the company (and many similar companies indulged in similar activity) have no credible results in “sowing discord” (compared with the ordinary, authenticated users of Twitter and Facebook in the US) they do seem to be doing this for commercial purposes. These appear to be similar to the activity of the Cambridge Analytica company which acquired personal data from social media users which they then marketed to political users.
Using fake or inauthentic accounts to retweet messages, or plant original messages, in a polarizing political or health debates is one way of mining personal data. Authentic users who retweet, “like” or repost such messages reveal a preference or bias which is of interest to companies involved in marketing products and ideas. Even seeding social media with pictures and videos of cats and dogs which attract likes, retweets and reposts can help obtain information of use to commercial and political entities.
Hell, Google, Facebook and Twitter themselves are involved in mining account holder’s personal information and selling it to advertisers.
How else do we end up getting social media messages related to topics we have searched for information on, or have commented on in social media. On the surface this appears harmless, even useful (although the continual messages I still get offering travel insurance just because I researched the topic several months ago are rather tiring – and counter-productive as they turn me off the advertiser).
My main objection to this paper is its uncritical and unthinking acceptance of the prevailing political narrative. I think it shocking that a scientific study makes no attempt to question or validate the narrative it relies on.
The data is extremely weak – only someone intoxicated by the political narrative will seriously see the extremely small number of tweets and retweets they found as evidence of a “strategy of sowing discord.”
Finally, the authors make no effort to consider other more reasonable explanations for their data. That is a pity as mining personal data by Google, Facebook, Twitter, Cambridge Analytica, the Internet Research Agency and other commercial companies should concern us all. Targeted advertising is very intrusive and annoying. Targeted political influence is also no doubt occurring and should concern us.
But the old trick of blaming the Russians for these problems is diverting our attention away from the real culprits.
I guess this shows how a bad political climate and destructive prevailing narrative can influence even the most scientific researcher.