Political parties are making increasing use of data analytics to track voters’ behaviour on sites such as Facebook and Twitter in order to send them political messages they will find persuasive. Parties have been campaigning on Facebook as part of the current UK election and both Labour and the Conservatives are said to have set aside around £1m for the purpose.
Yet there is a lot of uncertainty about exactly what is going on – and growing concerns that the harvesting of social media data was secretly used to influence swing voters in the EU referendum. Elizabeth Denham, the UK Information Commissioner, has launched a formal inquiry into the implications of deploying personal information in this way, focusing on both the referendum and “potentially also … other campaigns”
It is a welcome move, but it is limited to looking at the law around individual privacy and control of personal information. This is governed by data protection legislation and is mainly about ensuring that voters are not targeted as individuals – as opposed to in larger groups, such as a particular region or pensioners or whatever.
The inquiry will last several months – the political parties deny any wrongdoing – but it will not look at the electoral laws that also exist to protect voters. And this is an area that needs to be closely examined. Data analysis is making it look completely unfit for purpose.
The first problem is the election spending limits. There is a mix of old rules that set the limit for spending by each candidate in a constituency, and newer rules that create a cap on national party spending in campaigns.
Targeted social media advertising is being undertaken by central party organisations, since they have the resources to spend on data analytics tools. This may be counted against national spending limits.
But if social media advertising is concentrated on voters in particular marginal constituencies and involve messages connected to local issues likely to determine the campaign in question, it should count as local spending. Candidate limits for constituencies are low (around £15,000; varying according to the number of electors) so parties have an incentive to count social media as national spending.
The law makes no provision for monitoring the content of social media communications to determine if they are “local” or “national”. But even if you created such a regime, it wouldn’t solve the problem. The Electoral Commission has the power to fine parties for such infringements, but the parties would probably just see this as another cost of campaigning.
Also, the laws apply only to spending during the official election campaign period. Regulation does not extend to the monies spent in the inter-election years, when much of the expensive work of data analysis is being commissioned.
Apart from the spending controls, the use of social media in election campaigns is entirely unregulated. This is startling, given the tight legal controls applied to other campaigning methods: statute prohibits paid political adverts on TV or radio. (Party election broadcasts use free airtime; parties only need to meet production costs.) This longstanding ban is aimed at protecting our elections from American-style auctioning of airtime to the wealthiest campaigners.
This is about fairness -– the notion of a level playing field. Three years ago the UK government vigorously and successfully defended the ban before the European Court of Human Rights when a group of animal rights campaigners argued it ran counter to principles of freedom of expression.
Why then should unrestricted social media campaigning be regarded as a good or tolerable electoral practice? If the tool confers advantage on parties with the most resources to fund expensive data analytics, it is surely suspect on the same grounds as banned TV adverts.
There are other issues. Contrary to a recent suggestion by the chief executive of the Electoral Commission, targeted communications to voters on social media are not merely an evolution of traditional letterbox leafleting. They aim to influence voters without making it clear how the data used to tailor the message was gathered or processed.
This disempowers the electorate, making it difficult for voters to assess the communication. Put simply, election messages that don’t openly acknowledge the use of data analytics are an unfair form of campaigning.
Parliament could legislate to prohibit the technique. Or more modestly, it could require that such election communications make clear their use of data analytics. The law could insist that such an acknowledgement be displayed prominently, specifying a minimum font size.
We should not expect speedy reform, however. The government has not even adopted a 14-year-old recommendation from the Electoral Commission that online campaign material be legally required to contain an “imprint” identifying clearly the campaigner who generated the communication. This is a requirement for all printed campaign matter in elections.
Social media communication needs to be recognised formally by election law as a campaign method. Leaving it unregulated, subject only to the flawed rules on spending limits, undermines free and fair elections by devaluing the transparency on which election processes depend.
But since this is a tool that assists political parties to win or retain the power to govern, they have a massive conflict of interest in maintaining the status quo. It is incumbent on the rest of us to shout this from the rooftops until they are shamed into doing the right thing.