1Back in 2013, a quintessentially cyber-geographical project breathtakingly ambitious in size, scope and ingenuity came very close to winning the European Commission’s “Future Emerging Technologies” (FET) 10-billion euro Flagship award (Helbing and Bishop 2013). In one fell swoop, the FuturICT project would have completely transformed social science by combining information and communication technologies (ICT) and global data within a complexity science framework. The latter would have been built on a technological infrastructure consisting of a “Planetary Nervous System”, a “Global Participatory Platform”, and a “Living Earth Simulator”. The anticipated merging of social science and ICT would have led to critical technological and scientific outcomes (in particular, the “ability to understand our global socio-techno-economic system”) and to “a more resilient and sustainable global society”. The project’s ultimate justification, according to its mission statement, would be that “… Revealing the hidden laws and processes underlying societies probably constitutes the most pressing scientific grand challenge of our century.”
2Media hype went on overdrive. The British Broadcasting Corporation’s website proclaimed: “Earth project aims to 'simulate everything‘. The Living Earth Simulator will collect data from billions of sources. Is it possible to build a social-science equivalent to the Large Hadron Collider?” (Morgan 2010). Scientific American magazine ran a cover story titled “The Machine that Would Predict the Future” (Weinberger 2011). In the end, the FuturICT proposal did not win the FET competition, but its champions remain committed to continuing their efforts, and the project’s website is still live. Other, similarly ambitious global syntheses of technologies and data will almost certainly follow.
3Several orders of magnitude below the grandeur of the FuturICT vision, the notions of Big Data (in its acknowledged dimensions of volume, velocity, variety, veracity, and value), ubiquitous sensors, data mining, machine learning and now also deep learning, evidence-based decision making, and so on, are common currency today. Together they constitute the latest expression of the age-old human yearning to seek knowledge of the future or of the unknown, as reflected in the myriad forms of divination still practiced in traditional societies, as well as in the most sophisticated scientific predictions and forecasts of our time. While reliable numerical predictions of natural phenomena have in some cases a history much older than modern science, forecasts of societal phenomena such as those envisioned by FuturICT have only recently become the object of systematic study. The first recorded efforts may go back no further than the 1960s, when the tensions of the Cold War led a handful of military analysts to seriously consider the art of forecasting in the human realm (Kahn 1960; Gordon and Helmer-Hirschberg 1964). Currently the torch of systematic forecasting or foresight (‘la prospective’) based on “more or less rigorous methods and on data” is carried by the Paris-based non-profit organization Les Futuribles (www.futuribles.com).
4In recent decades, the world-wide societal transitions from the industrial age to the post-industrial and the information age have led to massive numbers of expert forecasts in all areas of the social sciences. Many such forecasts have also been formulated in geography and related disciplines, often concerning the fate of cities in an increasingly wired world. The ‘death of distance’ has been proclaimed, and eventually forcefully denied, but not before generating corollaries that make us smile today. Cities will disintegrate since people will be able to work remotely from anywhere. Dense city centers will be things of the past as face-to-face communication will no longer be required. Traffic jams will disappear as commuting will be replaced by telecommuting. Leisure time will increase and family life will improve as more and more people will be working from home. Long-distance business travel will be drastically reduced thanks to advanced video-conferencing. E-commerce will spell the end of brick-and-mortar shops. And so on. The fact that these predictions were made by respected scholars gives us pause. It seems that the experts in question were among the first to detect a number of radical emerging trends, but could not foresee the complex nexus of forces and human responses that often led to outcomes that were the opposite of those predicted. For example: urbanization proceeds unrelenting; denser and taller city centers are proliferating around the world; downtown living is a growing trend even in the traditionally suburban USA; regional shopping centers are thriving at the expense of traditional department stores, and small specialty shops tend to flourish; traffic jams may be getting worse, and telecommuting never really took off; more and more people do their computing on the road rather than from their home or office; and for professionals, there are more conferences that ever to fly to.
5The problem with trends especially in the social world is that they seldom last. They die out; they merge with other trends; they reverse themselves; they morph into something other. That means that their utility for policy and understanding is time-limited. Also, discontinuous changes happen that are often part of trends not considered significant or plausible until they actually hit. The Brexit ‘surprise’ of June 2016 may be a case in point. One of my favorite quotes is attributed to Henry Ford, speaking of his game-changing Model T car: “If you had asked people what they wanted, they would have said ‘faster horses’”. The point is that non-experts have difficulty predicting disruptive innovations, even if related emerging trends had been visible for quite some time, as was the case with the diffusion of the automobile. This is the opposite of the problem with many experts, who tend to over-predict based on emerging trends, as argued above.
6So here comes the suggestion of a computational infrastructure that is immensely powerful, that is completely free of human biases, that has access to continuous, global streams of Big Data and to the best algorithms for detecting emerging trends in the data, that is guided by the precepts of complexity theory, and that is monitored and used by experts in globally distributed ‘observatories’. That is the dream that the FuturICT project embodied. Not surprisingly, there have been questions and critiques of the concept. What about the totally unexpected events that were never part of any discernible trends? How can complexity theory, a basic tenet of which is the loss of predictability in complex systems, serve as the framework for the anticipation of developments in (exceedingly) complex social systems? Is there really such a thing as “the hidden laws and processes underlying societies”, as the project’s mission statement suggests? These are fair questions, but the core ideas of super-efficient data mining and deep learning are robust. Important actionable trends and outliers should be detectable that are too subtle for educated intuition to pick up, even when supported by traditional kinds of models. The possibility to continuously track how these trends develop and change over time and space can to a large extent address the problem of their relatively short shelf life and their concomitant fast-declining utility for policy and understanding, at least for the near and middle future.
7There are however at least two reasons why many critical societal events and futures may always lie beyond the reach of data analysis, no matter how advanced. The first reason is that uncertainty is only partially related to data (or even model) issues, contrary to what is widely assumed in geography and other quantitative sciences. Much uncertainty is qualitative, whereby it is not possible to assign error margins, probabilities, or even probability rankings to possible outcomes. ‘Deep’ uncertainty goes further, addressing the problem of the truly unexpected: the ‘unknown unknowns’, in military parlance, in addition to the ‘known unknowns’, that are often also not quantifiable (Walker et al. 2003).
8The second reason why uncertainty may defy data analysis is what is known as the ‘human factor’. The power of human expectations and beliefs to reinforce or deter predicted outcomes is expressed in the well-known phenomena of self-fulfilling and self-defeating prophecies, studied in psychology, sociology, and economics. In the first case unfounded predictions are made true, and in the second, well-founded ones are invalidated, because of dynamic feedbacks between people’s beliefs and their behavior. More generally, intentionality, which encompasses a person’s beliefs, expectations, desires, intentions, and motivations, often leads to actions and outcomes that are not subject to any laws or statistical regularities. The human factor is thus a significant component of deep uncertainty. A tragic (and very geographical) example of the many kinds of uncertainties and their implications is the disappearance of Malaysia Airlines flight MH 370 in March of 2014. There are the known knowns: the tracked section of the plane’s trajectory, including an abnormal sharp turn and the sudden disappearance of the aircraft from primary radar, while it continued to be tracked in flight by the military. The known unknowns, which include the reasons for the above two abnormalities, the remaining untracked trajectory, the cause of the crash, the possible role of the pilots or others in the crash, the location and extent of the optimal search area, the actual location of the wreck, and so on. (Note that until summer of 2015, even the fate of MH 370 – whether it had actually crashed – was among the known unknowns.) Then there are the unknown unknowns, most of them related to human factors: if someone deliberately caused the crash - why? Were the contradictions, delays, and inaccuracies in the information provided by three neighboring countries at the time of the incident deliberate, and if so, in which cases, and why? Was some critical information suppressed? Does someone(s) still alive know exactly what happened but is not talking? Ironically, later that same year, Europe’s Rosetta Spacecraft put a lander on comet Churyumov-Gerasimenko, very close to the originally planned site – whereas here on earth, where the position of aircraft is supposed to be known exactly at all times, all the technology and expertise of several advanced countries have been unable to locate the lost Boeing 777 after more than two years of efforts.
9In a paper on their ‘Delphi’ long-range forecasting method, Gordon and Helmer-Hirschfeld (1964, p.vi) write: “… potential objections that may be leveled at this approach: its inherently insufficient reliability; its tendency to produce self-fulfilling or self-defeating prophecies which would make it both undesirable and unreliable; the sensitivity of results to [the] ambiguity of questions; the difficulty of assessing and utilizing the degree of expertise; and the impossibility of taking into account the unexpected.” This assessment written by two military analysts of the USA’s RAND Corporation over fifty years ago still sounds true today, despite all the fantastic new possibilities of the Big Data age. This is because the big problem of Big (and small) Data is that it is, and will always be, about the past; even ‘this instant’ is part of the past even before you utter these three syllables. Some day before too long the earth may indeed support a ‘planetary nervous system’ made up of trillions of connected sensors feeding billions of computational devices. It is difficult to even imagine the brand new, exciting kinds of geographies that will be possible. Yet even then, because unknown unknowns will always be with us, much of our earth’s future will remain an enigma.