Both the EU referendum in the UK and the presidential election in the USA have generated a lot of debate over what influenced the results. They were close campaigns. There are many things that could have led to a different outcome. I’ve been particularly interested in the debate over the role played by technology, the web and data.
I think the debate is missing how politics risks becoming driven by data rather than informed by it.
Technology-driven progress, globalisation, fake news, social media, malicious and mischievous actors
Technology is a major strand in the debate about globalisation, nation states, jobs and inequality. The web and data are at their best when they are world-wide, open and know no boundaries but it is essential that we use technology-driven progress to build a better society for everyone.
Technology and the web play a big part in the increased consumption of news online and on social media rather than through more traditional media channels and in particular the changing economics of media and the rise of fake news.
Technology, the web and data are also present in the investigations into the potential role played by organisations and people that may be malicious, for example foreign governments, or simply mischievous. In the UK a report claimed that 1/3 of the tweets on the EU referendum in a one week period were created by bots.
Whilst debate about hacking and bots continues in other countries, such as Germany, this story seems to be at risk of slipping off the radar in the UK and USA. A more informed debate about their effects and purpose would seem useful.
But there’s a fourth element that I’m barely seeing debated at all. Data-driven politics.
Politics has always gathered and used data to help it make decisions. This data comes from door knocking, censuses, opinion polls, focus groups and election results. In our current age of data abundance there are ever more and cheaper ways for anyone to gather and use data. Some of the uses by political parties seem to be at risk of copying the worst excesses of online marketing.
In the UK Labour leadership contest in 2016 organisations such as Momentum and Saving Labour talked of capturing email addresses and the reach of their social media channels. Neither group has been open about who is in control of this data, whether it is secure from hacking or how it is used.
“Their dream was of a system that could put information from Twitter, canvassing, polls, websites, apps, into one giant IT programme that would then churn out extremely sophisticated models that would reveal the areas most likely to vote Leave, down to the street.”
Other campaigns and political parties debunked their claims on twitter and proudly said their data tools collected more information. No one questioned whether either was appropriate or healthy for democracy.
The New Statesman reported on the plans of a UKIP funder to start a new political party saying he claimed that Leave.EU’s email database was “a goldmine to anyone doing digital campaigning”. No one asked if it was either legal or right to transfer this “goldmine” to a new political party.
In America the Trump campaign was talking about its heavy use of data before the campaign finished. One insider on the data team said:
“There’s really not that much of a difference between politics and regular marketing.”
I hope I’m not alone in thinking there should be a difference between politics and marketing.
The Trump team used Facebook to target particular adverts to discourage black Americans from voting. Following Trump’s victory there are reports of one of the major data analytics firms being employed on an ongoing basis by both the government and an ongoing Trump campaign organisation.
This increase in the use of data to both listen to and influence people in political debates raises a number of issues.
There are biases in data and in how we use data
Data has biases. This might occur because there are gaps in how we collect data: for example ~10–20% of the UK and US population are not online because of issues such as cost, disability, location or motivation. Data also includes the biases in society such as those affecting gender and race. Bias can also occur through the people who decide how to analyse data and code the algorithms. People write code and people are biased.
If our political parties increasingly use the web and data to get them over the electoral winning line then they are likely to focus their efforts on winning over groups that are well represented in the data and predictable by the algorithms. Other people may be ignored.
National slogans, targeted adverts
The recent campaigns hint at a trend towards very broad brush national slogans (Make America Great Again!, Take Back Control) coupled with targeted campaigns aimed at particular interest groups.
Someone working in the car industry living in the Northeast of England might see an advert telling them that a political party is supporting car factories in Sunderland but see nothing else about that party’s policies or beliefs. The political party can see how that person responds to the advert — whether they comment, share, like, or retweet it — and use that data to tailor their next advert.
Some of these campaigns will come through official channels but targeted campaigns will come from through social media adverts, local (sub-brand in marketing speak) or unoffical channels. Coupled with the ongoing loss of funding for and trust in national journalism this will make it ever more difficult for a coherent national debate where a society makes an informed choice about its future. Instead political parties will tell different groups of people what they think they want to hear based on data.
We risk becoming more fragmented and the importance of values and principles in politics could become ever weaker as politics becomes more data driven.
Now some of this type of political advertising occurs already but technology, the web and data allow it to happen at a larger scale and at a cheaper cost. I can only imagine the voices in political campaigns saying that this is a race and that the process must become faster and more automated through “smart” algorithms. As we have already seen in other sectors these algorithms risk embodying and multiplying the biases in the data.
Use of data in political campaigns will influence how politicians govern when in office
Finally, there is an ongoing debate about the use of data by governments and the private sector. This debate concerns the rights and responsibilities that people and organisations have when data is collected and used. There are calls for greater control by people and more scrutiny by regulators.
This debate needs to include political parties.
If our political parties believe that the only way to get elected is through the use of data and algorithms then they will use them. If that use is not questioned and people are not held to account then that use could be normalised. Politicians might carry those normalised beliefs into office and it risks affecting how they govern and how they legislate.
Politics can be improved by new technology, the web and data.
The web offers ways for more people to be engaged in politics and it gives them more tools to influence politics. The web can help with a transfer of power from the centre to communities and people. Data can provide better evidence for policies and make it possible for us to trial new policies before they are implemented on a large scale and at a big cost. Better use of data can help improve public services and the economy.
These things can be dazzling. But we need to recognise the risks. Not just that some technology innovation is pointless but also that some uses of technology are actively harmful. That they can harm individuals and communities and that copied wholesale into politics they can damage democracy.
Rather than being driven by data we need to encourage politics to be informed by data, to be open about how it uses data and for political parties to use data and technology to help people engage with politics and make better decisions based on both evidence and their values and principles. It’s up to all of us, particularly those of us with knowledge of technology and data, to help make sure that this happens.