Sandholm Among "The 73 Biggest Brains in Business"
Tuomas Sandholm probably wouldn't call himself a genius—his cv lists his windsurfing ranking (#1 in Finland, 1986) pages before his NSF CAREER Award—but the editors of Condé Nast Portfolio see it differently. Sandholm recently made their list of "The 73 Biggest Brains in Business" as one of the three most influential academics in the business world.
Sandholm, a professor in Carnegie Mellon's departments of computer science and machine learning, earned the honor for his pioneering work in expressive commerce powered by combinatorial optimization, which helps buyer companies and their suppliers negotiate the best price for goods and services by selectively comparing variables that traditional sourcing systems lack—and in a fraction of the time.
Generally speaking, commerce rests in a complex matrix of prices, preferences, and constraints, where suppliers offer the cheapest price possible and buyers decide which companies to use.
The process, called sourcing, is difficult: decisions can be time consuming and the cheapest price isn't always the best deal. Sometimes companies have allocation preferences that complicate the final award.
To optimally match supply and demand, Sandholm developed the world's fastest market-clearing algorithm, the Advanced Sourcing Application Platform (ASAP). ASAP improves speed of business and global competition by combining math, computer science, and algorithm engineering to quantify increasingly complex problems that result from a greater number of variables.
ASAP makes it possible for companies to negotiate with multiple partners simultaneously and on a much larger scale. The added time advantage translates to savings by decreasing the number of man-hours required to conduct business.
For example, old sourcing models required a shipping company to bid on a contract by guessing what other companies might bid for the same contract; using Sandholm's expressive commerce model, the company could determine its best price possible in any number of given scenarios regardless of its competitors. Likewise, the company awarding contracts could determine which allocations are the most preferable based on factors beyond price, such as reliability, projected growth, and number of suppliers.
As the flagship product of CombineNet, for which Sandholm is the founder, chairman, and chief scientist, ASAP has saved customers over five billion dollars.
"Unlike traditional reverse auctions for sourcing, which attempt to obtain savings by squeezing suppliers," Sandholm explains, "expressive commerce creates a win-win between the buyer and supplier. This is because suppliers can 1) bundle goods and services in self-selective ways that they can bundle efficiently, 2) specify novel alternatives in creative ways, 3) express their economies and diseconomies of scale, 4) express their side constraints, and 5) do not suffer from exposure problems in bidding."
In addition to electronic marketplace analysis, Sandholm, together with graduate student Andrew Gilpin and Sam Ganzfried, developed algorithms for automatically generating game-theoretically strong strategies for poker. That work has resulted in some of the world's best programs in the American Association for Artificial Intelligence's Computer Poker Competition every year since 2006.
In collaboration with graduate student Pranjal Awasthi (and previously with Professor Avrim Blum and graduate student David Abraham), Sandholm also develops algorithms for improving kidney transplants that help match living donors with unrelated patients. The algorithms support not just pair-wise exchanges, but three-way swaps, four-way swaps, and so on, as well as donation chains triggered by altruistic donors. The newest algorithms are also mindful of what might happen in the future, such as new patients and donors coming into the exchange. The result is more successful exchanges among a broader range of patients and lives saved.
Sandholm is also the director of Carnegie Mellon's Agent-Mediated Electronic Marketplaces Lab and winner of the 2003 "Computers and Thought Award" presented every two years by the International Joint Conference on Artificial Intelligence.
Condé Nast Portfolio is a publication of Condé Nast.