1101

Maximizing Happiness

by
Scope Correspondent

Elan Pavlov works in the space between theory and reality. An independent computational economist based in Massachusetts, he explores how to use the formulas of classical economics in the real world. His goal: to maximize the total good—defined as whatever it is that makes an individual happy. “We’re trying to allow individuals to find out what their own goals are and maximize them,” he explains.

One way to maximize happiness, according to some economists, is to design a mechanism—a theoretical arrangement to determine who gets what and how much they should pay. Pavlov’s’s interested in the details of implementing these mechanisms, methods for computing prices and allocations. His specialty is auctions.

In a traditional auction, the person who bids highest wins and pays whatever she bid. Often, however, in order to win the item, the winner pays more than she initially intended. Economists view this phenomenon, called the winner’s curse, as sub-optimal.

“A better mechanism,” reveals Pavlov, “is the VCG auction.” Vickrey-Clarke-Grove auctions assign prices in a way that maximizes the total value for all bidders. Winners pay according to how much their participation affects the total value of the system.

For example, say that a bidder, Alice, wins an auction for a painting. In doing so, she causes another bidder, Ben, to lose the auction. To make up for Ben’s perceived loss, Alice should have to pay what the painting was worth to Ben. The price she pays is determined by her effect on the system. (Let’s ignore the seller, assuming that if he wanted the painting, he wouldn’t put it up for sale.)

“In other words,” Pavlov continues, “what is the welfare of everybody else in the world if I didn’t exist? What is the welfare of everyone else in the world if I do exist? I look at the difference between that.” He smiles. “That’s how much I should pay, to make up for my existence.”

AuctioneerPavlov explains that VCG auctions have been mathematically proven to maximize welfare. While this might sound nice in theory, he continues, prices in VCG auctions get very hard to calculate as the number of bidders or the number of items increase. In the field of computer science, general VCG auctions are known as an “NP-hard problem.” This means that they can’t be calculated on human time-spans.

Despite this limit, Pavlov remains optimistic: “Maybe we can approximate, we can look at special cases, maybe we give up—there are a lot of things we can do.” In 2007 he and colleagues solved one special case and found an algorithm that would increase revenue from Google’s Adwords auctions by a factor of up to 1,000.

The plan was rejected.

“From a mathematical point of view, I’m certainly correct,” Pavlov explains, when asked for a post-mortem. “From a practical, company point of view, I’m not sure I’m correct. It could make sense for them not to listen to me because the advertisers won’t buy it.”

The philosopher Yogi Berra is said to have observed that “in theory, there is no difference between theory and practice. In practice, there is.” Pavlov’s aware that putting his methods into practice requires subtle human considerations. Getting from economic model to real world happiness, it seems, requires more than a clever algorithm.

Comments

0 Comments