Social media may let the powerful predict the future, but are we ready to be herded like sheep?
Hint: We already are
The new TV adaptation of Asimov’s classic Foundation has brought renewed interest in predicting the future. Now it may be possible but the consequences of letting this power fall into the wrong hands could be catastrophic.
The problem is that it already has.
Written in the 1950s, Foundation assumed that individual lives were irrelevant to the march of history just as individual atoms are irrelevant to the behavior of gases. Later in life, with the rise of chaos theory, Asimov admitted that he believed psychohistory — his word for predicting the future — was implausible. Chaos theory suggests that, as perhaps many of us suspected, you cannot separate the actions of individuals from the events that shape nations and the world.
If you just look at history, you can see that the actions of individual assassins, for example, Lee Harvey Oswald, the Kennedy assassin, or Gavrilo Princip who murdered the Archduke Ferdinand and started WW1, changed the course of history. To say that the actions of these individuals don’t matter and that the march of history is always one of masses, is naive and ignores the sensitivity of world events on small differences in conditions.
Nevertheless, Asimov may have sold himself short in suggesting that psychohistory cannot be used as it is in the series. We cannot know, as Hari Seldon appears to understand, what will happen. We can only know what might happen if certain conditions are met. Moreover, it is always a gamble to tamper with history because the unforeseen can derail any long term plan. This is part of the reason for the Second Foundation. Any attempt to shape history must be guided so that priors (pre-existing knowledge that feeds into a probability calculation) can be continuously updated.
In 2005, I was an intern at the RAND Corporation, the famous think tank lampooned in the classic cold war film Dr. Strangelove as the “Bland Corporation”. (They were credited with the doctrine of Mutually Assured Destruction that kept the cold war powers in check.)
I was still in graduate school and the RAND internship had been a rare opportunity to spend twelve summer weeks working on policy for one of the world’s top policy research institutes.
Overlooking Ocean Avenue and the Santa Monica beaches of Southern California, RAND was where John Nash of Beautiful Mind fame had spent some of his time, earning extra cash and turning his game theory ideas to solve problems of national security before returning to MIT.
Every day at lunch time I would run alongside the Pacific, thinking about the mathematics of prediction and planning.
There I met James Dewar, who was involved in developing a technique for predicting the future. The technique focused on dealing with uncertainty in your assumptions to shape what will happen. It led to a feature article in Scientific American which you can read for free here.
Yet, can we really shape the future? Or is it a myth? If we can, then why don’t we and what are the consequences when someone else does it for us?
Predicting human events with social media
When it comes to predicting the future, compared to people, climate science turns out to be easy. It’s all basic physics, but we could not begin to predict the future of the climate without a lot of data, and this is what has been missing in predicting the future of human events, until now.
All prediction is data driven because data feeds your priors. The less data you have, the wider your uncertainty and the more likely things are to go sideways. A good predictive model will, of course, take all uncertainty into account, but such models tend to be useless. If you look at many models of Covid infection rates, such as the long-term IHME model, for example, they tended to either have enormous error bars, meaning that they were predicting a wide variety of possible futures, or they were wildly wrong. All from insufficient, “boots on the ground” data.
In the past, predicting human events was largely impossible because of this lack. The news only reports events after they have happened and often without any clear idea of why. Meanwhile, surveys and other data gathering methods are spotty at best because they are only designed to find what the surveyors are looking for. They can’t catch trends and the unexpected.
But now we have huge additional sources of data about human activity that we never had before. This source can help predict events before they happen and may even lead to knowing the outcomes of elections as well as when recessions are going to occur. That source is social media.
The reason social media is so important as a data source is because it contains the contents of people’s day to day lives, and, while people may not realize it, they are part of a much larger web of social interconnections and their actions can have outsized consequences. From starting a fashion trend to influencing local politics to guiding the zeitgeist in any number of ways, social media has an enormous influence. It also reflects what people are doing day to day. This information, never available before, can be studied now to develop a true psychohistory. (Psychohistory is actually a completely different discipline, but we’ll use Asimov’s word here.)
While it is unlikely that mathematics can predict thousands of years of future development, research has already shown that things like box office revenues of movies can be predicted from Twitter chatter. Elections, microeconomics, market trends, and the dissemination of information (and misinformation) are all now predictable from social media data mining. How far is it from that to predicting long-term trends to go from marketing to influencing the fate of empires? If, as in climate science, we are not concerned with specific events but in general trends, it will be a huge step forward towards Asimov’s dream...or perhaps nightmare.
In the hands of those who have our best interests at heart, the Hari Seldons of the world, this ability may seem like a good thing. Unfortunately, the groups that do control such data mining techniques in the West are big tech corporations that use that data to sell us products. They do not have our best interests at heart.
In nations such as China, this technology is used without any pretense of privacy guarantees to control people, limit dissent, and keep the ruling party in power.
Meanwhile, political parties in the USA use this power to control the vote via sophisticated gerrymandering. Media companies use it to increase their circulation. Anyone that has access to this data, uses it to their own ends, not ours. People are being herded like sheep and not for their own good.
In a world where data is increasingly one of the most valuable commodities, we see it consolidated in the hands of a few and those from whom the data was extracted are denied a right to it. This must change.
People have a right to know what data is being collected, but they also have a right to understand how they are being manipulated and what is permissible. Like Hari Seldon’s predictions, the world of social media software along with the wider world of data science algorithms is highly secret. For example, if you suddenly start receiving junk digital mail from a local company, it may be that they used a service to geoframe your home address. You may have unknowingly opted into sharing your location with an app on your phone, but the question is whether you have a right to know that this was done to you and who provided the data to whom? Or is it sufficient to have a vague, lengthy service agreement in legalese that nobody reads?
Opting out of social media, not having a smartphone, using a VPN are all options to avoid tracking and having your data stolen, but in a world that is increasingly connected this is synonymous with living off the grid. It isn’t feasible for many of us. (Particularly those of us who make a portion of our living from social media!) Rather, we need regulation and insight into what these companies are doing, and we need it before the fates of empires fall into the hands of the Googles and the Facebooks of the world.
Asur, Sitaram, and Bernardo A. Huberman. “Predicting the future with social media.” 2010 IEEE/WIC/ACM international conference on web intelligence and intelligent agent technology. Vol. 1. IEEE, 2010.
Rousidis, Dimitrios, Paraskevas Koukaras, and Christos Tjortjis. “Social media prediction: a literature review.” Multimedia Tools and Applications 79.9 (2020): 6279–6311.