All products featured on WIRED are independently selected by our editors. However, we may receive compensation from retailers and/or from purchases of products through these links.
SAN FRANCISCO -- You're alone in your car or office, or you're walking into town to run some errands -- pick up some stamps, make some copies. But in the grand scheme of things, you're not really alone.
In fact, there's a decent chance that someone nearby could probably lighten your workload, without really adding to their own. Who are they, you wonder, and are they willing to help?
If ideas for body-worn personal agents are ever translated into products, wearable computers will some day help you to find them and negotiate to-do list trades.
"You tell your agent when you meet someone to try negotiating with the person's agent," researcher Gerd Kortuem said at a session of the International Symposium on Wearable Computers, which took place Monday and Tuesday in San Francisco.
"Both agents know the complete task-sets of both users. At the end, each agent proposes a deal that is optimal." The optimal deal ensures that at least one party -- hopefully both -- scores a win, and no one loses (i.e. does more work than they would have on their own). One person picks up an extra book of stamps while they're at the post office, for example.
Kortuem and his colleagues at the University of Oregon's Wearable Computing Research Group believe this is how the ideas of game theory -- where competitive situations can be meshed for a mutually beneficial outcome -- can benefit future users of wearable computers.
The users might be co-workers in an office or even bike messengers meeting at an intersection.
"They meet and at the same time their wearable personal agents start communicating," Kortuem said. "They cooperate only if there is the opportunity to enhance the users' goals."
The agents would maintain a user's task list, becoming fully aware of the locations and activities involved, whether it's walking down to the dry cleaner, dropping off mail at the post office, or buying groceries.
The software knows each user's schedule, address book, to-do list, and personal interests. The agents incorporate that data -- which Kortuem noted is already on most people's computers and PalmPilots -- with the intelligence it gathers as the day progresses: whom you meet, where, and when. When an encounter occurs, it produces a negotiation.
If both users approve, a deal is struck. It's a modernization of the time-worn tradition of borrowing butter from your neighbor. They do it because they know you'll do it for them one day. Inconvenience avoided.
"It uses a cost function based on where you are now and where you have to go in order to perform a task," Kortuem added.
A car buyer, for example, visits a dealership, hoping to buy a certain Volkswagen at a certain price. The saleswoman, meanwhile, has a different car in mind, and sales commissions are dangling before her like money on a tree.
While price-finding services on the Web such as Priceline.com can leverage millions of networked users, wearable agent research tries to leverage isolated users in the vicinity of other isolated users.
The ability of wearable computers to detect changing surroundings would provide the smarts necessary to know when two people and their goals intersect -- whether for a few seconds or a few hours.
In the example of the car shopper, the agents could bring the two parties closer to a deal more seamlessly than their anxious users ever could.
Such systems would depend on exchanging tasks that are equal in their simplicity. Anyone can pick up someone else's dry cleaning, for example, just as any other person can buy a book of stamps. But few users would trust anyone else to withdraw cash from an ATM machine.
The value of the tasks must also be the same to both users. "If the other person is in a wheelchair and I am on a bike, the cost is clearly different," Kortuem said. For that reason, it's important that agents' strategies remain private.
The protocols harbor no illusions about human qualities, however.
"It works on the assumption that people are selfish," said Kortuem's University of Oregon colleague Jay Schneider. "And if people aren't selfish, it works even better."
To succeed, both users have to trust that their agents will seek a fair deal. Thus, the protocol incorporates values of fairness. "It's not fair that one user always picks up stamps for the other.... So both of us get even distribution of work."
Although the protocol will always reduce the work of one user, it will reduce the work of both users whenever possible. And it guarantees that neither user will do more work than they would have to without an agent working on their behalf.
Ultimately, Kortuem, Schneider, and their fellow researchers envision large-scale wearable communities, with groups of nearby users interacting and cooperating in many ways and for many reasons. End result: a less redundant, more efficient group.
As Schneider put it, "The community has a tremendous net gain."
