Next revolution

nThe Next Scientific Revolution
How data mashups can help save the world by Tony Hey
56 Harvard business review November 2010

True PDF release: storemags & fantamag

storemags & fantamag - magazines for all
Hbr.orG

Tony Hey is the vice-president of the External research Division of Microsoft research and the coeditor, along with Stewart Tansley and Kristin Tolle, of The FourthParadigm: DataIntensive Scientific Discovery (Microsoft research, 2009).

For decades computer scientists have tried to A VISITOR WALKING the halls of Microsoft Research’s teach computers to think like human experts by emcampus in Redmond, Washington, today is likely to overhear discussions not only about computer sci- bedding in them complex rules of linguistics and reaence but about a surprisingvariety of other subjects, soning. Up to now, most of those efforts have failed from which way a galaxy rotates, to a new AIDS vac- to come close to generating the creative insights and solutions that come naturally to the best scientists, cine, to strategies for managing the planet’s precious physicians, engineers, and marketers. The most supply of fresh water. What could these issues possibly have incom- talented experts not only have a deep understandmon? And why would Microsoft—ostensibly a soft- ing of data but also are able to see the possibilities ware company—be involved with them? The simple “between the columns”; they can find the nonobvianswer is data—vast amounts of data. So vast that ous connections within or between disciplines that make all the difference. when we run the programsthat analyze some of We have reached a point, however, where even the databases, the temperature of the building that houses 10,000 microprocessors shoots up sev- the experts are drowning in data. Digital information is streaming in from all sorts of sensors, instrueral degrees. Today our computer scientists find ments, and simulations, overwhelming our capacity themselves in partnership withleading scientists in a wide array of disciplines—astronomy, biology, to organize, analyze, and store it. Moore’s Law has for decades accurately predicted that the number of chemistry, hydrology, oceanography, physics, and transistors that could be placed on an integrated cirzoology, just to name a few—working on efforts such cuit would double every two years, and until recently, as drug development,alternative energy, and health this decrease in transistor size was accompanied by care cost containment. And, yes, even commercial increased microprocessor performance. To increase software projects. We believe that a new generation of powerful software tools, which support collabo- performance today, we must program multiple processors on multicore chips and exploit parallelration and dataexploration on an unprecedented ism. The multicore revolution has arrived just as we scale, are about to enable revolutionary discoveries face an exponential increase in data. That increase in these fields.

ILLUSTrATIoN: JAMES JoYCE

November 2010 Harvard business review 57

THE BIG IDEA THE NEXT SCIENTIFIC rEVoLUTIoN

The Four Paradigms of Science
THEORY
beginning in ancient Greece and China,people tried to explain their observations through natural laws instead of supernatural causes.

EXPERIMENTATION

by the 17th century, scientists like Isaac Newton tried to make predictions for new phenomena and would verify hypotheses by conducting experiments.

COMPUTATION AND SIMULATION

The advent of highperformance computers in the latter half of the 20th century allowed scientiststo explore regimes inaccessible to experiment and theory, such as climate modeling or galaxy formation, by numerically solving systems of equations on a large scale and in fine detail.

DATA MINING

Using more-powerful computers, scientists begin with the data and direct programs to mine enormous databases for relationships. In essence, they use computers to discover the rules by studying the...