ICELAND VOLCANO: Air travel – computer says No
19 April 2010
Frankfurter Allgemeine Zeitung
The aviation standstill was not caused by facts, but by a computer simulation. Increasingly powerful computers now take decisions off our hands. Not only in the air, but in everyday life, too. So we need an overseer to control computer decisions, urges the FAZ.
Klaus Walther, Lufthansa’s corporate spokesman, believes in technology. So when he looks up at the empty sky over Europe and bemoans the want of intuition and common sense that grounded the planes, ears perk up. Walther just wants to fly. And yet his protest against the flight ban is a milestone in the nascent criticism of digital-age technology, a chapter in the history of modern society’s systematic handing-over of power to computer models. Needless to say, the airlines have their own interests at heart. But Walther has never been known to put profit before safety. And those unwilling to board a plane at the moment ought to get this straight: the invisible cloud completely paralysing air traffic is not composed of ash and dust, but of numbers crunched in a computer. The chain reaction unleashed by a volcanic eruption today might be triggered by plenty of different eruptions tomorrow: geological, economic, social, you name it.
Today computer simulation is stopping civil aviation – entailing costs that run into the hundreds of millions per day. What will it do tomorrow? What is it already doing? And at what cost? The cascade of decisions that eventually paralysed aviation on the European continent took place without any empiricism, measurements or collating and matching up data. The simulation outputs we all rely on come from the Met Office, the UK’s National Weather Service. The authorities’ cautious stance is understandable. After all, who wants to be responsible for a crash? So it’s not about contesting the soundness of simulations on principle. The problem is they’re treated so much as facts as to force decision-making processes that leave no room for experience, intuition – in common parlance: common sense.
The danger of a new authoritarianism
Are there any regions without clouds? Can we run test flights to determine actual ash distribution so as to generate hard data? Nothing doing: a single simulation sufficed to interfere with the fates of millions of people and grind Europe to a halt. That has to do with the fact that the simulation generates its own social algorithms, leaving nothing to the discretion of the authorities. The agents in these models are indeed people, but basically they have to act like the algorithms that triggered a cascade of market reactions in the financial crisis because the parameters so determined.
And suddenly we’re all spectators: passengers, pilots, other weather services, the authorities. A “human response” to the machine is no longer possible. If the degree of predicted complexity is high enough, “fate” ceases to exist. And where fate does not exist, everything becomes a matter of legal responsibility, as the simulation is always right if and when disaster strikes. Computers, says US mathematician Steve Strogatz, now calculate things even the most brilliant mathematicians can’t verify. Which leads to a new authoritarianism: the computer findings are becoming a “spectator sport”, we can only cheer or boo them, but we can’t get our minds around them because we don’t understand anymore how the computer arrived at its results.
Today your plane is grounded, tomorrow your career
Simulation is becoming prediction. In this age of social digital networking, similar predictions are made about people all the time. The Florida State Department of Juvenile Justice, for instance, has just announced it is going to use IBM software to predict the social behaviour of juvenile offenders. The British judicial authorities use the same “predictive analytics”. Studies on the predictive power of Twitter or Google show that such prognoses are astoundingly spot on.
That’s what makes them so tempting and dangerous. If not only flight safety, but also social mobility, intellectual competence, health can be predicted with an air of scientific certainty, just a few parameters will suffice for the regulators of life to step in and take over. They’re already at it in companies and government agencies. Your plane is grounded today, your career might be too tomorrow. Only those living in the past can believe that scepticism about the nascent power of computer models is just nostalgia for a pre-industrial era. The point is we need to establish overseers to take on computer-made decisions and raise objections based on empirical data and intuition: that is a task modern society has to face. If we don’t, we might all end up grounded soon.