, , , , , , , , ,

It all boils down to free will. It always boils down to free will. Or can be boiled down to free will. I was thinking the other day that you can take any statement about human beings, or a specific human being, and connect it to free will in three sentences. But that’s absurd. Every statement about human beings is -already- a statement about free will.

There was a research paper that did the media rounds a little while back. It is some work done at The Psychometrics Centre at the University of Cambridge and Microsoft Research, Cambridge. In it, they study the Facebook Likes of 58,000 volunteers, and also use the demographic information and the results of some psychometric tests of those volunteers to come up with some pretty startling predictive statistics. For instance, they use machine learning to develop models that take as input Facebook Likes and predict with high accuracy personal and sensitive information such as sexual orientation, race, religious beliefs and political leanings.

Do you think that human beings are complex? Do you think they are unpredictable and unknowable? What if someone told you that your Facebook Likes, and a simple linear equation could together determine whether your parents are together, whether you use drugs, and what your satisfaction with life is? Well, that’s exactly what this paper does. Some of the findings seem obvious and therefore non-creepy: a Like for Sun Tzu indicates Competitiveness, and a Like for “Sometimes I Hate Myself” indicates a Neurotic temperament. Other findings are non-obvious. For instance, a Like for “Curly Fries” apparently indicates high intelligence, and one for Mountain Biking indicates a Calm and Relaxed temperament. The point where things start to get creepy is when Likes for Britney Spears and Desperate Housewives turn out to be moderately indicative of being gay, whereas less than 5% of the users labelled gay actually have obvious Likes, such as for groups like “Gay Marriage.”

This is creepy because now, suddenly, the world at large can deduce some pretty personal things about you, and you weren’t even aware that you were giving anything away. Better still, these models are largely just linear models — that is, mathematically, they are about as simple as it can get. In light of the usual perception that every human being is uniquely complicated, this feels like a slap in the face. All our self-regard suddenly seems quite narcissistic.

The thing is, there is no reason to think humans are beyond modelling. As organisms our wiring is largely the same from one human to the next. In fact, if you think of Asimov’s Foundation series, and Hari Seldon’s psychohistory, the argument is that the future of the human race can be predicted quite accurately. There is something to be said for this idea in the sense of the law of large numbers. A tennis ball consists of a large number of atoms, each with some probalistic behaviour, but as a whole, the behaviour is quite deterministic — when you strike a ball, you can predict its future quite well. If you have a large number of humans, each made of a large number of atoms, and consider it all as a single entity, you can perhaps predict the future of humanity as well as that of the tennis ball.

Or at least that is what the analogy is. It’s not clear how well it stands up in reality. It appears that disruptive ideas and inventions (e.g. steamboats, personal computers) can come about through the actions of a very small number of individuals, and a model based on the law of large numbers cannot account for such “anomalies.” In fact, the real brilliance of Hari Seldon, who, let us remind ourselves, is fictional, is that he predicts the disruptive events too. (Are they still called disruptive then?) Even more importantly, he knows how to manipulate them even when they are in the distant future, by means of counterbalancing trends. While I think that human beings are pretty far from doing that today, I am not aware of any solid arguments that rule it out in the future.

Which about brings us back to the question of free will. Are human beings pre-programmed? Maybe pre-programmed with some probabilistic behaviour, but pre-programmed nonetheless? If some innocuous-seeming Facebook Likes are enough to predict some pretty complex things about us, are we really just simple machines, which are not truly free to make choices? Was I just meant to write all this, staying up late to do it, and it wasn’t a choice at all? (And, worse still, does it mean that my long history of sleep-deprivation is destined to continue forever?!)

Free will is a topic that has vexed scholars from across the planet for centuries together, so I don’t think I am going to be able to provide an answer — or at least not tonight! But it is definitely one of the most interesting and fun questions to ponder.