Come to think of it - The art(ifice) of negotiation

Delta and Delft Integraal/Outlook often write about innovative ideas that offer great promises for the future. But what has happened to them a couple of years on? What for instance has happened to the idea of the pocket negotiator, a computer-based advisor that helps you to reach the optimal result in negotiations.

(Illustration: Floris Wiegerinck)

Delta, 29-03-2007
The computer can give you advice for an offer that pleases both parties. That seems relevant for daily life. 

The research conducted by professor Catholijn Jonker’s man-machine interaction group, at the faculty of Electrical Engineering, Mathematics and Computer Science, got a boost from a Vici grant she received in January 2008. Now she has three PhD students and one post-doc working on various aspects of the automated negotiator. As yet, it does not fit into your pocket - it’s merely a piece of software that runs on all types of PCs. Eventually though the program could run on a laptop or a PDA, or even as an iPhone App, to assist people at the negotiating table. The program’s subject matter could be anything, but knowledge so far has been collected on two specific subjects: real estate and terms of employment.

PhD student Dmytro Tykhonov shows a demo of yet another field of expertise: party organising. There are several issues involved here: the kind of catering; what type of drinks; where to party; what the invitations should look like; music/band, dj or cds, and finally: who will clean up afterwards. “We chose as an example something that would appeal to students,” Dmytro says. Jonker explains that, no matter what the subject, there must be at least five issues to negotiate about, which is part of the theory of negotiation. Jonker uses the book, 'Negotiation Analysis', by Harvard business school professor emeritus Howard Raiffa, for the business scenarios.

The user makes choices from a menu for each of the items (chips & nuts; non-alcoholic drinks; party centre; live music and cleaning service) and then value each of the choices to correspond with their subjective importance. Tykhonov for example is crazy about nuts (ten points), but he doesn’t care about the looks of the invitation (1 point only). He also needs to prioritise the various issues by giving them a certain weight (varying between 0-1), telling the computer to what extent the issue is negotiable. Once the user profile is finished, the fun begins. After two iterations, Tykhonov is ninety percent content. But then he makes an offer that actually suits him less, prompting the computer to tell him that he is damaging his own interests.

Jonker points out that people often do this in negotiations, because they cannot keep score of all their priorities simultaneously. Research revealed that people reach about eighty percent satisfaction levels, whereas computers reach up to ninety percent, depending on their negotiating strategies. Jonker thinks she and her colleagues will need another three to four years before their intelligent agents are user-friendly enough to hit the streets.