I see Cambridge Analytica is trending again because of this Channel 4 documentary exposing the Trump campaign’s attempted deterrence of Black voters in 2016.
Unpopular opinion: This documentary has some big structural flaws.
Why? Hear me out… https://twitter.com/Channel4News/status/1310625842888880129
Unpopular opinion: This documentary has some big structural flaws.
Why? Hear me out… https://twitter.com/Channel4News/status/1310625842888880129
A general point: How is it possible that not a single expert was consulted on this?
Lots of ppl have studied this & would have been able to contextualise some of the claims…
E.g. @kreissdaniel, @shannimcg, @jesse_b_p, @ajungherr, @sivavaid, @davekarpf, @eithanhersh
Lots of ppl have studied this & would have been able to contextualise some of the claims…
E.g. @kreissdaniel, @shannimcg, @jesse_b_p, @ajungherr, @sivavaid, @davekarpf, @eithanhersh
Had they bothered to ask, @davekarpf would have told them that this is “just” negative advertising (which is still despicable), but not structural voter suppression (eg closing polling stations, etc.). More in this short thread here: https://twitter.com/davekarpf/status/1310796399655694336?s=20
Had they bothered to ask, @sivavaid would have told them that a) Facebook is the main problem here, b) That Black voters were disproportionately targeted because they were more likely to be Clinton supporters ( https://www.pewresearch.org/politics/2018/08/09/an-examination-of-the-2016-electorate-based-on-validated-voters/#demographic-and-political-profiles-of-clinton-and-trump-voters)
@sivavaid has summarised his main points here: https://twitter.com/sivavaid/status/1310702419832119300?s=20
They could have read @jesse_b_p who dissects how journalists/pundits “construct practices of data campaigning as all-powerful despite the limited empirical findings to that effect” & questions the assumed effectiveness of digital campaigning. https://www.tandfonline.com/doi/full/10.1080/10584609.2020.1723751#.XkwLRYpyS90.twitter
They could have read this paper on targeted political ads which finds “small effects of political advertising […] regardless of context, message, sender, or receiver” before making grand suggestions about the persuasive power of ads. https://advances.sciencemag.org/content/6/36/eabc4046
They could have read this illuminating @davekarpf essay on MediaWell which shoots massive holes in the “Cambridge Analytica/Micro-targeting is super effective“ narrative: https://mediawell.ssrc.org/expert-reflections/on-digital-disinformation-and-democratic-myths/#.XhDQjYRCzGQ.twitter
Not once does the documentary properly address the question if this kind of shadow advertising on Facebook actually works. Instead, it simply assumes that this is the case because it fits the narrative.
“Turnout in ward XYZ went down & many people in this ward were marked as deterrence and targeted with ads => the latter caused it“.
But correlation is not causation.
But correlation is not causation.
Evidence that points to other, structural factors at play & helps explain low turnout is only addressed in passing.
We get a brief mention that Democrats (to this day) simply seem to assume that Black voters will automatically vote for them...
We get a brief mention that Democrats (to this day) simply seem to assume that Black voters will automatically vote for them...
...to the extent that Hilary Clinton didn’t really campaign in one of the key examples presented in the documentary, Wisconsin, but that’s it.
Factors they could have mentioned but did not: the obstacles to voting disproportionately affect people of colour. Black voters are more likely than white voters to cite obstacles to voting, eg. “inconvenient polling place” or “transportation problems”… https://www.theguardian.com/us-news/2017/jan/18/american-non-voters-election-donald-trump
…not to mention a wealth of other laws more or less aimed at voter suppression, which strongly affected Black voters in—surprise—for instance, Wisconsin in 2016. https://www.americanprogress.org/issues/democracy/reports/2016/11/11/292322/voter-suppression-laws-cost-americans-their-voices-at-the-polls/
This article also provides a good summary of the many structural problems and injustices against Black voters in the US: https://www.theguardian.com/commentisfree/2019/dec/19/myth-black-people-dont-vote-harmful-stereotype
But sure, let’s talk about the magic power of online ads.
But sure, let’s talk about the magic power of online ads.
Don’t worry, there’s still lots in this documentary to be rightly concerned about.
Yes, we should be concerned about Facebook’s central role in elections, its handling of user data, its cooperation with political campaigns (of all colours) and some of the tools it has built.
Yes, we should be concerned about Facebook’s central role in elections, its handling of user data, its cooperation with political campaigns (of all colours) and some of the tools it has built.
Yes, political adverts on social media should be accessible in a public database, with information on who paid for them, who was targeted, who saw them, for how long etc. so that we can study effects. Not an expert on this but election rules probably also need some updating here
Yes, the way personal data seems to just lie on the street in the US (and elsewhere)—up for grabs for anyone who wants to use it and has the money to do so—is a problem and should be regulated/curbed.
We should also have a wider discussion ab political advertising & limits.
We should also have a wider discussion ab political advertising & limits.
Finally, does the documentary spell a problem for Brad Parscale because it shows that he lied to Congress about the campaign not targeting Black voters (which might lead to perjury charges)? Again, yes.
But the central claim--that these ads are the most pressing worry for US democracy--are on a very shaky foundation & a massive distraction from more fundamental problems.
Finally, writing threads like this gives me no joy. Channel 4 is a top broadcaster with excellent journalists. Exposing Cambridge Analytica as the sham it was, was a milestone of investigative work. But this is also why such obvious oversights set me on

Ways to avoid this:
Talk to the experts (not me, this is just one of my pet interests). They are here, on Twitter, and elsewhere & usually happy to talk even though they gain little from it.
Read academic literature, not just media reports
Tone down the causal language



Addendum: Seeing that this has gone "viral", here are some additional comments that came up which are relevant and interesting. First up, this good critique by @TomDobber who (I think) recently finished his doctorate on micro-targeting. https://twitter.com/TomDobber/status/1310915664857763840?s=20
Second, this critique by @RMAjayi who makes a valid point and points to other scholars/activists on this that I had missed: https://twitter.com/RMAjayi/status/1310956995508396032?s=20
Third, a list of voter suppression efforts with very real, measurable, and awful effects. https://twitter.com/coribush/status/1311328706296770561