Research NewsComputer Science

Hedging Bets on Hard Problems

See allHide authors and affiliations

Science  03 Jan 1997:
Vol. 275, Issue 5296, pp. 33
DOI: 10.1126/science.275.5296.33

You are currently viewing the summary.

View Full Text


Solving certain problems in computer science is a gamble: There's no way to know in advance how long an algorithm will take to yield an answer. In this issue (see page 51), a group of researchers presents a technique for bettering the odds by imitating Wall Street: splitting the computer's investment among a “portfolio” of different programs in the hope that one of them will get lucky.