A case for predictive systems

Talented people do not need to have a vision of the future ten years out or even ten days out, write Vivek Ranadivé and Kevin Maney in The Two-Second Advantage: How we succeed by anticipating the future – just enough . What is needed is “a highly probable prediction just far enough ahead to see an opening or opportunity an instant before the competition. That's true for athletes, artists, businesspeople, or anyone in any field.”

The prediction, the authors observe, only needs to be two seconds out, though the actual time may be hundredths of a second or several minutes, depending on the situation. They cite Malcolm Gladwell's insight in Blink: The power of thinking without thinking , that judgments made in two seconds are often more accurate than those made after months of analysis.

The book opens with the story of Wayne Gretzky, ‘The Great One' in ice hockey, who would not skate to where the puck is but to where it is going to be, as if he seemed to be two seconds ahead of everyone else. Exploring the question whether there are lessons from Gretzky that have implications for, say, running a department store, the authors find that a number of companies are implementing some of the first predictive systems, aimed at getting the right information to the right place just a little ahead of time.

Examples mentioned in the book include Southwest Airlines (which is developing a system that will let it watch its inventory of planes, the weather, ticket prices, and other factors and constantly adjust, perhaps sensing that a storm is coming and refiguring the airline's entire schedule and moving passengers to different flights before routes snarl); and Xcel Energy (which is testing a system in Boulder, Colorado, that uses two-way meters and outage monitoring equipment to build memories of how electricity moves through the distribution grid and what seems to trigger problems).

Tracing the history of computer evolution, the authors note that despite being good at gathering, storing, recalling and analysing a lot of data, computes are really autistic. How so? “Computers have been made to excel at exactly those things humans are bad at. They can store and recall every detail or do calculations in an instant. Most humans can't do those things because we spend so much energy thinking about higher-level concepts.”

While, therefore, there is the useful, symbiotic relationship between computers and humans, the authors underline that if we want computers to be a little more human, to be a little bit talented, what comes in the way is that the business of collecting and delving into mountains of data eats up too much processing and storage resources and ultimately takes too much time. “To take machines to the next level, we're going to have to teach them how to forget.”

A section devoted to ‘Talented systems' takes us through Rajesh Rao's lab at the University of Washington to see robots that could be ‘the cousins of the robot from Lost in Space, Rosie from The Jetsons, and the C-3Po.' They have square, electronics-laden heads, human-shaped bodies, lights, buttons, and herky-jerky movements, the authors describe. “They look like toys but are nothing of the sort. The robots were created to help Rao learn about human brains…”

Riveting read, though you can glean a lot through two-minute page-flips, too.

COMMENT NOW