Facebook 1, Government 0. What the Facebook Saga Showed the Data Science World

Having the house to myself a few weeks back, as my family were away visiting the in-laws, it was a great opportunity to kick back, relax, and get up to speed on my TV shows. Unfortunately, it was not to be, due to being hooked on the drama that was the attempted (and failed) Zuckerberg crucifixion on Capitol Hill. To many, it was akin to “The Empire Strikes Back”, although these days we’re not entirely sure who is the Empire and who the Rebel Alliance. Now that the dust has settled and our news channels have resumed normal service (reporting the same old never-ending political scandals or the Royal wedding), it is interesting to take a step back and look at what just happened.

As someone who was also sitting around university residences and engineering faculties around 2005/6 trying to figure out the new MySpace before Facebook hit (clearly with a lot less talent and Harvard resources than the Zuckerberg – Winklevoss combo who first kicked off the idea), these recent events, at least in my opinion, have opened a brand new and fascinating chapter in the debate around the “internet economy”. While the financial and stock price impacts of these recent events on Facebook have been almost zero to date, I think what is certain is that the general population are slowly understanding the implications of having a fully connected information infrastructure, and governments and regulators are beginning to grapple with it as well.

A blip on Facebook's share performance

Ambiguity, contradictions and double standards abound in this whole saga: for example, the American government has been fiddling with net neutrality policies for the last year (which could end up discriminating against smaller companies), and yet quizzed Facebook on their own neutrality and global competitiveness. Not to mention the free rein the FBI and other agencies have on all of our data, in the name of Freedom and Liberty of course.

Given the roughly 28 years of the World Wide Web (remember those carefree, pre-dotcom days?), it has taken us considerable time to build a critical mass of users, information and infrastructure to more fully utilise its capabilities. In the 90s, we had the problem of too few users and too little useful content, compared to what we already had on a CD ROM edition of an encyclopaedia. Today this situation has reversed and the internet is arguably bogged down with a lot of useless, outdated and redundant information – at least for the less internet savvy. However, this is where data science comes into its own, to automatically find what is useful from the pile, or at least what it thinks individuals might find uniquely useful. Even a simple Google search potentially uses a lot of your personal information (e.g. previous searches, location, and browsing history) to provide you with the most relevant answers.

This, I believe, is the crux of what the politicians on Capitol Hill have misunderstood from the start. Targeting relevant content based on user profiles, using advanced methods such as Machine Learning, is actually quite clever, useful and increasingly necessary. What Zuckerberg spelled out quite simply, is that the entire business model of almost every useful website today revolves around using user data to target information and adverts, giving platforms like Facebook and Google their significant revenues. Even a seemingly unrelated website is often running Google ads, which use similar targeting mechanisms. It has been around for ages, and the main use of “cookies” even before we started filling in social media profiles.

The fear of course arises from the misuse of these methods to influence audiences. As a case study, the Cambridge Analytica intrusion is an obvious one, where the methods are accused of influencing votes across the US elections and potentially Brexit. However, I believe it is only gaining the media attention it is at the moment because people think it worked in helping bring about the unexpected results. If the democrats had won and Britain voted remain, I doubt we would have ever known about Cambridge Analytica. In truth, we still have no idea whether it made a difference or not, nor will we ever, especially considering those voters leaning to the right tend to be older and not as hooked on social media as their millennial children are.

I’m sure I wasn’t the only one who chuckled at Ted Cruz’s accusations about Facebook censoring conservative views – if they had done even more censoring then perhaps they would avoided this particular issue altogether! (In the Capitol Hill interrogation, Ted Cruz noted several right wing oriented pages that were taken down for various reasons, probably as they were central to cases of hate speech. The irony here is that if Facebook had censored the right-leaning fake news that was allegedly being propagated through Cambridge Analytica algorithms, then they would have avoided this fallout altogether.)

In the large majority of interviews of the general public, internet users do not seem to care too much about what their data is used for – they understand that they are responsible for what they put in the public domain and have absolute control over it. Many are perceptive enough to also not believe everything they see online, and especially those who have grown up with using the internet as a resource know that fact checking is an important part of using internet resources, where proper expert review is obviously lacking (and the very reason for Wikipedia’s peer review mechanism).

To sum up, I believe what this latest chapter has really taught us is:

• The real power behind connected information systems is the advanced use of data. While this may sound obvious, we’re only really starting to scratch the surface of what we can do with it. There is a lot more exciting stuff coming, which isn’t just about targeted advertising. (I explore a few of the ideas in my previous blog posts.)
• Governments are worried about the potential power of these advanced methods when it comes to online data – possibly because they are not in control of it, nor do they really understand it, yet.
• The public, especially younger generations, generally do not mind having their data used for purposes of targeted information, provided it is fair and unbiased.
• Regulation will be ever evolving in this space – we already have GDPR coming into play in Europe, but much of the rest of the world has quite a bit of catching up to do. Regulation is needed, and social platforms will have to change, as no internet-based business model is currently perfect. I hope it will not become over-regulated and stifle innovation, because if Western economies inhibit innovation in this area, it will open the door for the less regulated countries to overtake in the field of artificial intelligence (and I believe they are already beginning to do so).

Potential disadvantages of GDPR