A deep dive: Cambridge Analytica

Cambridge Analytica have been successfully controlling the minds of thousands of people – if not hundreds of thousands – since 2013. Working for political parties for over 200 elections in the United States and the United Kingdom, as well as Argentina, the Czech Republic, Kenya, India, and Nigeria, they may have singlehandedly swung the outcome of two controversial knife-edge votes. In 2016 they acted in the U.S. Presidential Elections and likely the U.K. European Referendum – the full extent of their actions may never be known.

It all began on the 21st April, 2010, when Facebook announced Open Graph, a tool for 3rd party application developers allowing them to ask the user for permission to access their personal data; at the time this also included access to the personal data of all the friends of a user. As of 2017, the average number of ‘friends’ for every profile was 338 people (median at 200); given the number of people who directly accessed the app is anywhere from 185,000 to 300,000 we have an enormous amount of potentially affected people. Cambridge Analytica claim the number of people whose data was harvested was 30 million, whilst Facebook claim this number is 87 million. Whichever way you look at it, the sheer amount of data that could have been sourced surpasses the level of current known mass-surveillance techniques used by governments.

In 2013 Dr. Aleksandr Kogan, a research associate at the Department of Psychology of Cambridge University built a 3rd-party profiling application to use for his research. By 2014, Facebook had updated their terms and conditions which no longer permitted new 3rd party applications to harvest friends’ data or be used for psychometric profiling. However this did not apply to the application that Kogan had built in the previous year. In early 2014 Kogan had been introduced to SCL Elections and in May of the same year formed the company Global Science Research (GSR) to create a data-licensing contract between itself and SCL Elections, an affiliate of Cambridge Analytica. The contract stipulated that the funding transferred to GSR was to be used “to further develop, add to, refine, and supplement GS psychometric scoring algorithms, databases and scores” from which Kogan and his partner Joseph Chancellor could not profit directly from.

“It will be extremely difficult to take back control of our data”

The Guardian published an exposé in December 2015 on the involvement of Cambridge Analytica in aiding Ted Cruz’s presidential campaign with the use of Facebook user data. After this publication Facebook ceased their collaboration with Dr. Kogan and asked him to delete all the data acquired by GSR. Facebook and Dr. Kogan had a close relationship, with Facebook sharing aggregate user data to feed into GSR models. These models were used by GSR to match personalities with voter record data provided by SCL Elections to be able to target voters.

The contract between GSR and SCL Elections stated that the predictions made by this model were comparable to, if not more accurate than, those made by close acquaintances of test groups. This claim, however, is unsubstantiated and there is no reason to believe that targeting on an individual basis was possible. This does not mean that group targeting was not possible with a sufficient prediction rate for it to be viable: even with a conservative 50% correct prediction outcome on a group size of a couple of hundred can be an important factor when some elections were won by a few hundred votes. Once Cambridge Analytica established their target groups, media content would be created, curated, and published based on the groups’ personal interests and political beliefs; they would then would use marketing platforms with specific targeting tools, such as Facebook itself, to reach voters with the right material.

Something that might hit closer to home is with a company called AggregateIQ, who were affiliated with the SCL group, the parent company of Cambridge Analytica. AggregateIQ was hired by the ‘Vote Leave’ campaign team called BeLeave, £625,000 was donated from the official Vote Leave campaign for the referendum, all of which was used to pay AIQ for their services. In total £3.5 million was paid to AIQ for their services by the different Vote Leave campaigns.

It would be impossible to ever estimate how successful Cambridge Analytica, SCL Elections, and Global Science Research were at converting masses of data into votes; however, these events show us some of the more dangerous effects of powerful computing technology we have developed. Social-media services like Facebook are wielding powerful currency in this new age, this currency being our personal data. With millions of people sharing data publicly with minimal privacy restrictions, it becomes increasingly profitable to record and store as much data as possible on as many people as possible. In a dystopian society it is easy to imagine governments going through all our data and controlling what we see on a massive scale. Legislation can at best control where our data is stored, who it is shared with, and what the data is used for, however it is slow to be written and implemented. It will be extremely difficult to take back control of our data.