MIC WRIGHT: Efforts to introduce transparency in online political advertising are falling short
PUBLISHED: 10:00 10 May 2019 | UPDATED: 13:55 10 May 2019
MIC WRIGHT says that political campaigners are undermining the democratic process with manipulative, duplicitous online advertising and more needs to be done to introduce transparency.
There was great fanfare in October 2018, when Facebook introduced new rules for political advertising running on its platform in the UK. Requiring advertisers to prove their identity and location before their messages were shown and displaying the group behind each ad to users, along with the introduction of a searchable Ad Library, was repeatedly hailed as "transparent" in media reports.
But taking a look at any one of the current Facebook ads focused on Brexit reveals how the social media giant didn't so much install a window into the world of online political advertising as opt for some frosted glass.
Messages promoted by groups with names like Better Brexit, Now Brexit and the Brexit Defence Force are linked to pages with no indication of the individuals managing them nor of their ultimate funding sources. Facebook also remains reticent to show the targeting criteria used by the advertisers. So, while it's possible to view them in the Ad Library there is no way of knowing who is seeing these messages in their Facebook newsfeeds.
The Guardian reported in April that the Mainstream Network - an operation designed to look like a grassroots campaign made up of pro-Brexit Facebook pages backing a no-deal exit - is in fact overseen by Lynton Crosby's political lobbying firm CTF Partners, with a budget of more than £1 million.
The UK information commissioner, Elizabeth Denham, told the Digital, Culture, Media and Sport Sub-Committee on Disinformation that the activity is being probed as part of her office's wider investigation into the use of political data online.
Denham told MPs that she was shocked to see that the government's recent White Paper on online harms did not significantly address the issue of political advertising: "I was surprised and disappointed that there wasn't more focus on what I think is a huge societal harm, which is around electoral interference and the need for more transparency in political advertising."
The information commissioner was asked by MPs whether Facebook's changes to its rules around political advertising go far enough. She told them: "You can't leave it to an individual company. There needs to be more robust transparency tools and there needs to be regulation that requires companies to have systems in place to give real transparency."
Her view is echoed by the Institute of Practitioners in Advertising (IPA) and the independent fact-checking charity FullFact, which have both called for an industry-owned register of political advertising in the UK, with details not only of content, but of who is being targeted and how much is being spent to reach them.
While Facebook's Ad Library and other measures are clearly inadequate, Twitter and Google have yet to provide public archives of UK political advertising run on their platforms.
Current rules on electoral advertising were drafted to cover posters, billboards and leaflets, not micro-targeted social media posts and customised email campaigns. The Electoral Reform Society's director of research and policy, Jess Garland, put it succinctly: "The UK's analogue-age election rules are a meddlers' charter that leave our elections vulnerable."
While the claims made in commercials for washing powder or cosmetics can be rigorously challenged by the Advertising Standard Agency (ASA), since 1997, it hasn't regulated any form of political messaging. Lies that would be unacceptable in the process of trying to sell pet food go unchallenged when it is voters being served up rhetoric to swallow. It was easier to get away with falsely claiming that Turkey was joining the EU than it is to lie about the nutritional content of a turkey drumstick.
The official Vote Leave campaign spent more than £2.7 million on targeted Facebook ads during the EU referendum. Data subsequently released by the social network to the Digital, Culture, Media and Sport Sub-Committee as part of its inquiry into fake news, revealed that Vote Leave had run ads with 1,433 different messages.
You may also want to watch:
They ranged from bizarre claims that the EU was out to kill the British cuppa to emotional appeals featuring polar bears and bullfighting. The adverts were seen more than 169 million times in total.
As well as relatively straightforward adverts, Vote Leave ran a competition offering a £50 million pound prize to anyone who could predict the outcome of every match played in the 2016 European championships. It was a data harvesting exercise.
But while regulators are trying to get to grips with the online political advertising equivalent of muskets, campaigners already have machine guns and tanks at their disposal.
In 2016, the Trump campaign tested an average of 50,000 ad variations a day to micro-target voters, peaking at 150,000 ads on some days. It did that, in part, by using machine learning to quickly get rid of under-performing messages and formats and bring the most effective lines to the top. Voter profiles are ever more detailed and machine learning is increasingly making connections between data sets that human campaign strategists would never be able to make on their own.
We are not far from a time when it will be possible to quickly and effectively target single voters with messaging geared to their particular hopes and fears.
A recent Channel 4 investigation accused the Arron Banks-funded Leave.EU campaign of faking footage of migrants entering the UK and attacking women, but future political advertising could make even that look mild. Deepfake videos - clips using machine learning to turn real people into convincing puppets for words or actions they might otherwise never say - are becoming increasingly advanced. It is only a matter of time before deepfakes are effectively used for political disinformation and the waters are further muddied.
Last September, three members of Congress - Democrats Adam Schiff and Stephanie Murphy, along with Republican Carlos Curbelo - wrote to the US director of national security, Dan Coats, to warn that "as deepfake technology becomes more advanced and more accessible, it could pose a threat to public discourse and national security".
The issue is not just that campaigns could use deepfake techniques to spread rumour and gossip, but that an increased climate of distrust could make it easier to discredit real videos with plausible deniability.
Donald Trump initially apologised for his comments on the infamous Access Hollywood tapes, but he later suggested that the audio was fake. In a climate where deepfake attacks on politicians become common that kind of obfuscation will become easier for figures who are so willing to brazenly lie about their words and actions.
Currently, the UK is in a situation where the Electoral Commission complains of campaigns submitting invoices for Facebook advertising that "[make] it difficult or impossible to know what the money was spent on and where". While the regulations don't even require that minimal meaningful level of disclosure, and technology companies are not forced to deliver true transparency, the integrity of UK elections will be in doubt.
Ensuring that all online political advertising in the UK has to abide by the same imprint laws as offline advertising is key - they must state who paid for them and published them. But it has to go further than that. We need a publicly viewable algorithmic ingredient list tied to every targeted online political ad - who is seeing this message and what criteria were used to decide that they would? Revealing the recipe shouldn't be left to the whim of individual companies, but a legal requirement.
Political campaigns used to have to shout their messages but Vote Leave's targeted Facebook ads during the EU referendum were whispers. The manipulation would have been far less effective if it had been done in the full light of public scrutiny, where the conflicting tones and disparate imagery would have been apparent. It's easy to sell a lie if only people who don't know any better can hear you telling it.
This is a Wizard of Oz problem: It's currently too hard to know who is behind the curtain, pulling the strings. If voters are able to see how they are defined by campaigns that target them, they may be a lot less willing to take the messages presented to them at face value. Offline political adverts are restricted by their context, jumbled on the doormat with bills and junk mail, or pasted on a billboard.
Online, the political messaging can hide itself as a plea to defend the polar bears or ban bullfighting, while its creators have another agenda entirely. We need regulations and regulators with the power to pull back the curtain and reveal the levers, who's pulling them and who is ultimately paying for it all.