Thread by @rossatkin

Really sorry of you aren’t interested in Brexit but, well me, my friends and family have to live it and things are getting pretty fluid right now, so I wanted to pull some facts together about Facebook and the referendum that are worth bearing in mind. Thread! 👇

People like @Anna_Soubry and @ChukaUmunna have been making a good case for @peoplesvote_uk based on the idea that in the original ref people didn’t know what kind of Brexit they were voting for. This makes sense because some people voted leave expecting to stay in SM etc 1/

The problem is the counter argument is also good. The mandate granted to the ref was to decide whether we stayed or left and everyone involved was clear about that at the time. To those with a sense of proportion this isn’t convincing but Brexiteers are people of principle 2/

There is actually a much stronger argument for @peoplesvote_uk but it is obscured by a fog of confusion, created by both a lack of understanding of the technology that was used to win the ref, and the larger than life personalities on both sides of the argument 3/

The misunderstanding is most clearly articulated by @OwenJones84 who went out the weekend it was announced that @Arron_banks was being investigated by the National Crime Agency dismissing the idea that misuse of social media platforms could have affected the result of the ref. 4/

This was seized upon by one of the very people who ran paid ad campaigns on Facebook in a manner deemed illegal by the Electoral Commission, as well as the rest of the #Brexit messaging grid. 5/

https://twitter.com/darrengrimes_/status/1058440119395475462

Whether he meant to do this or not @OwenJones84 ended up making the hard Brexiteers arguments for them. He was put in position because of a consistent reflex in the media to group three distinct issues, that should be treated separately, together as ‘social media and politics’ /6

These issues are:
A) The toxicity of political discourse on Twitter caused by the design of the platform and the presence of professional trolls. Journalists like to write about this because they experience trolling themselves but isn’t that impactful /7

B) Radicalisation of Facebook users caused by the shifting of social norms within closed FB groups, and the platforms tendency to boost extreme content and isolate users from information that would challenge their attitudes or beliefs. More impactful than A) but less visible 8/

C) The ability of political actors to circumvent the principles of UK electoral law by using Facebook’s ad targeting platform. This is by far the most important story and we now know that multiple Leave campaigns broke electoral rules with their Facebook advertising 9/

You shouldn’t have to believe in a conspiracy by the Russians, Steve Bannon or anyone else to view this as very important and to see that the cumulative effect of the illegal activity would have been enough to swing the referendum result. 10/

First let’s look at what been deemed illegal -Grimes spending £675k in coordination with the Vote Leave on Facebook ads by the agency, Aggregate IQ Let’s assume they were charging a 20% fee to administer the spend it leaves £540k to buy actual ads. 11/
electoralcommission.org.uk/i-am-a/journal…

I’ve been buying some Facebook ads lately and the better ones have been getting around 3,500 impressions per £15 spent. That’s 233 impressions/£. I think FB prices are higher now than 2016 but based on these numbers that £540k would have bought 126 million impressions 12/

We know from the blog post by Vote Leave's Dominic Cummings in Jan 2017 that the audience those ads were targeting, and the messages they contained had been carefully optimised over months of focus groups with undecided voters and A/B testing online 13/
dominiccummings.com/2017/01/09/on-…

So the first question is whether, just that £540k, was enough to persuade 1.3 million people who would have stayed at home to turn out and put a cross for leave, or 635,000 people, who were going to turn out anyway but would have voted remain, to vote leave. 14/

It’s worth noting at this point that the people that these ads were targeted at were not hardcore Leave or Remain supporters. They were, people, like most people in the UK, who are not particularly interested in politics and don’t have strong views either way - 'undecideds' 15/

There are lots of these people. Question time usually gets around 2.5 to 3 million viewers, Strictly gets 10 million. The Telegraphs’s circulation is just under half a million, National Trust magazine gets 2.7 million. Politics is a niche interest. 16/

For just that illegally spent £650k to swing the ref they would have needed a conversion ratio of about 1% in the persuading non-voters to turn out for Leave scenario, and 0.5% in the converting Remain voters to Leave ones version. 17/

I’m getting 11% clickthrough to the shop and converting that at about 3%. That leads to a total conversion rate of 0.3% BUT I’ve not had time to optimise my messaging or targeting, my e-shop is a bit rubbish and I’m taking people’s actual cash. 18/

If I can turn up with some paper robots, janky webshop, no focus group or targeting data, at the start of the A/B testing journey, and *ask for hard earned cash* and still get 0.3%, the finely honed Vote Leave machine would have done better for a vote that’s free 19/

This makes even more sense when you look at the actual ads they were running, targeting for example people who are interested in animal rights with content about bull fighting. Many of the claims made in these ads are either totally untrue or at least very misleading. 20/

I actually have to send customers a product that meets their expectations so can’t mislead. As at the time these ads were never expected to be public, there was no need to comply with electoral law that applies to billboards or newspaper ads. 21/
bbc.co.uk/news/uk-politi…

So far we’ve only considering the activity that’s already been deemed illegal and not what’s ‘officially suspicious’. The National Crime Agency are investigating the £2m loan made by Banks and his companies to Leave.EU 22/
bbc.co.uk/news/uk-politi…

We know that most of the people staffing the Leave.EU campaign were paid by Banks’s companies, so we can assume most of this money went on actual ads not administration by let’s be charitable and assume the same 20% admin fee as before 23/
opendemocracy.net/uk/brexitinc/p…

So that £1.6 million, on top of the £540k of ads already been deemed illegal, giving us a total of £2.14 million and meaning conversion ratios of 0.24% in the persuading non-voters to turn out for Leave scenario, and 0.12% in the converting Remain voters to Leave ones version 24/

These ratios are below what most people doing e-commerce would consider a success when they are asking for people’s money, and far below what they would expect when asking people to do something free, like installing an app 23/

Obviously, the kind of people that get very excited about principles will say that you could never compare the noble exercise of democracy with grubby e-commerce, but the truth is that people who are not very invested in an issue can be easily convinced one way or another 24/

When they look at the facts all together, people with a sense of proportion will see that there is a good enough chance that the cumulative effect of lawbreaking could have swung the result that it makes sense to ask the question again. [ends] 😓🤓 25/

Apologies to @hugorifkind and @helenlewis for trolling you before, this is what I was on about. @davidallengreen and @jamiesusskind the legal view would be very interesting if you have time.