Veselin Popov explains how he and his colleagues from Cambridge University use digital prints to build psychological profiles and discusses drawing the fine line between technology and manipulation.
By Ana Klissarska
The work of this young man sounds like something out of science fiction yet the technology is already outstripping our imagination. What until yesterday was the storyline of a film, today is reality. From our activities online, Veselin Popov and his colleagues from Cambridge can learn a lot about us. Does this means the birth of a digital Big Brother or that online data will improve our lives?
Veselin is Director of Business Development of the Psychometrics Centre of the University of Cambridge – a multidisciplinary research institute specialising in online behavior and psychological evaluation.He graduated in Law at Trinity College Cambridge and applied his skills in a variety of fields ranging from music and the UK sports industry to the Bulgarian Ministry of Finance. His interests include the right of intellectual property, privacy and data protection, digital marketing and entrepreneurship.
Veselin, can you tell us more about your work?
The Psychometrics Centre of the University of Cambridge is a research institution specialising in psychological assessment, online behaviour and Big Data (electronic traces, which each of us leaves in the network). We conduct academic research and training in psychometric methods. This essentially means how to ask questions that measure personal characteristics or intelligence through data provided by our computers and mobile devices.
Using Big Data, for example, the social media information of tens of millions people around the world, we study how psychology is expressed online and how much our digital behaviour can reveal about us. We also try out our study research in practice and collaborate with organisations in the business, education, health sectors and beyond on topics such as personalisation, recruitment, security and artificial intelligence.
Where is your research applicable?
Our research and methods are applicable virtually in any area where humans are involved. The reason for this is that our lives are becoming increasingly digital and our digital traces all contain potentially useful psychological information. We work primarily with social media and search data – which pages you like, what language you use, which websites you visit, etc. But data from IoT devices (Internet of Things i.e. connected with the internet), smartphones, wearables, transactions, health-tech and other platforms can also be relevant. The insights from our work have so far been applied mainly in occupational and educational testing, marketing and human resources, but we are exploring various projects in finance and cybersecurity.
How has the digital world changed people from your point of view?
Being always-perfect is a huge psychological problem, particularly for young or vulnerable people. The pressure to curate an online presence that is attractive to others (including employers, lenders, governments and others with no business in our private life) is growing, and companies like Google, Facebook and Twitter are highly skilled at commoditising these identities and selling them to the highest bidder. Benefitting from the innovations of the information age while mitigating its negative social consequences is a delicate balancing act that we, as a society, must continuously engage in. Some would say that we have already lost our rights to genuine privacy, but I think we can still fight against this.
Ultimately, networking platforms and search engines are not common media, as the tech giants would like us to believe. Algorithms embedded in these services actually frame the way we communicate, amplifying certain emotions and predispositions and creating new ones entirely. They detect which version of the news we might prefer to read at a given moment, and lead us deeper into our own echo chambers. But this is how they are designed, with personalisation and convenience of the user, so we have to take responsibility for our own behaviour as well: we might need to think more carefully before clicking on things, which is a kind of effort that not everyone is willing to put in. Laws and governments are supposed to protect us in these situations.
How else can this Big Data be used?
Our digital traces reveal our psychological profiles, so they could be used in many ways. Besides marketing, they could also be used for better communication for important public issues – for example to encourage volunteering, charitable donations or civic engagement in issues like climate change or social reform.
The education sector is also increasingly interested in using Big Data methods to improve teaching and develop well-rounded forms of assessment. It is conceivable that school examinations that take place once a year and put enormous pressure on students might instead use digital footprints collected throughout the school year (in controlled settings) to better identify students’ potential and personalise their learning process.
Big Data also has huge potential in the health sector, from new drug discovery to personalised therapy. Researchers at Harvard and Vermont University have shown, for example, that the colour filters used by Instagram users can be associated with psychological diseases. Depressed users rarely apply filters, but when they do they mostly use ‘Inkwell’, which makes photos black-and-white, whereas healthy users tended to prefer ‘Valencia’. The number of faces in photos is also predictive.
Chatbots such as Wysa and wearables such as those made by Empatica are further examples how our devices are increasingly trying to offer psychological support, and illustrate the growing trend towards human-machine compatibility.
Another study finds that when the talking voice of an in-car navigation system is adapted to match the personality of the driver, the driver pays more attention to the road, has fewer accidents and talks more to the car. So Big Data could also make us safer on the roads. Let’s just hope we will talk more to each other in the future and not just to our devices!
One of your programmes – Apply Magic Sauce – constructs psychological profiles of Facebook users according to their ‘likes’ with five dimensions: openness, conscientiousness, extrovertness, acceptability and neuroticism. Ain’t the human race more complex than this?
BIG5 Personality is the most commonly used index for describing individual differences in personality. It does not put people into frames, but rather recognises the enormous psychological diversity between people and provides a convenient method for finding stable traits. Your score on a given trait does not determine your score on any other trait and the scales are independent. This five-factor model replicates well across all geographies and cultures, with the exception of some Asian countries where a sixth ‘tradition’ factor is added to the model.
BIG5 can strongly predict your job performance, health outcomes and relationships – this is another reason why it is popular. The self-learning models in Apply Magic Sauce that predict personality perceive humans in hundreds of thousands of dimensions, but people can only really hold 5-7 of these in their head at one time. That is why psychologists use simplifications such as the five-factor model, to make the results more easily interpretable and actionable.
How much can we trust someone’s digital profile?
Academic research has repeatedly shown that people’s online profiles tend to be accurate reflections of their offline behaviours. Of course, the extent to which this is a true, partial or distorted reflection may still vary drastically person to person, particularly if the individual is motivated to misrepresent aspects of himself. My view is that no single online profile can ever give you a complete picture, but with access to multiple sources of digital information (Facebook profile + Tweets + emails + credit card for example), computer algorithms can arrive at very accurate predictions.
The fact is that our digital behaviour is being used, one way or another, by a panoply of private and public organisations to determine our access to services. In future, we will face a choice between: a) keeping our profiles accurate and up-to-date, thus eroding control of our private sphere, or b) misrepresenting who we really are online, and get services from institutions that need our data to function. The dystopian vision of this ‘socially ranked’ society is well portrayed in an episode of Black Mirror and we can see a real example of the same in China’s surveillance state. Our consultancy and applied research projects encourage better practices among the companies that operate in the Big Data ecosystem.
Is it possible to abuse the information we give about ourselves in social networks?
Yes, it is certainly possible and we unfortunately continue to see examples of this every day. The Edward Snowden revelations were only the beginning of public awareness in the United States and Europe. Recent investigations into Russian interference in the US elections and microtargeted advertising in Brexit remind us that the full truth has not yet come to light. It can also be argued that ‘misuse’ and ‘legitimate business purpose’ is a fluid notion and determine by company practices, occasional regulation and customer opinion. Tracking these trends can be difficult as it depends on the perspective you adopt, but there does seem to be progress being made, for example with investigations into data analytics in politics by the UK Information Commissioner’s Office. Legal initiatives such as the General Data Protection Regulation and e-Privacy directive are also helping to maintain ethical considerations.
Last year it was said that Donald Trump won the Presidential election because of the methodology developed by the Cambridge Centre for Analysis of Digital User Behaviour even without your knowledge and acceptance. The same analysis is used for Brexit. Do you think that this is fair play? Where is the boundary of private space?
The company Cambridge Analytica has nothing in common with Cambridge University. It is owned by the British company SCL Elections and funded by American billionaire Robert Mercer. This is about a digital agency that, seeing our publications in the Proceedings of the National Academy of Sciences in March 2013, decided to develop a similar database and similar abilities to build psychological projections from Facebook. It happened, according to their CEO, through buying data from data brokers. In my opinion, it’s not fair play. But let’s not forget that Cambridge Analytica work not only for Trump, and also for the most common financial and media companies and employers’ agencies (under a project called CA-Commercial). It is clear that they profit from public attention. I think right now the limit of personal space has no influence over their work.
Do you have a Facebook profile?
Yes, of course I do. It’s even under my real name. I don’t think it is right to do research on platforms that you are not personally familiar with, because each platform has its idiosyncrasies.
I try to use Facebook as little as possible because I know what personal data I reveal by using the site.