Have a computer negotiate for you

In complex negotiations – from the sale of a house to the formation of a coalition – it is often impossible for people to oversee all aspects. CWI’s senior researcher Tim Baarslag is developing algorithms that can do this and can help people achieve better negotiation results. Baarslag recently received a NWO-Vidi grant of 800,000 euro.

Publication date: 26-07-2021

In complex negotiations – from the sale of a house to the formation of a coalition – it is often impossible for people to oversee all aspects. CWI’s  senior researcher Tim Baarslag is developing algorithms that can do this and can help people achieve better negotiation results. Baarslag recently received a NWO-Vidi grant of 800,000 euros to develop new negotiation and coordination algorithms that can strike multiple partial deals with multiple partners. With this grant, Baarslag can develop his own innovative line of research and set up a research group within the next five years.

During the time-consuming formation of the Rutte III cabinet, you suggested using a computer to speed up negotiations. Maybe an idea for upcoming formations?

‘I still believe that. A coalition agreement can be seen as a collection of agreements on all kinds of social issues. The “space” for all possible solutions is enormous. The human way of looking through this is to go through them one by one. “Let’s agree on one topic first, then move on to the next.” But it leads to better results if you discuss several subjects at the same time, so that you can make trade-offs. This “multitasking” is very difficult for people.

The complexity increases enormously as soon as you discuss even a few subjects at the same time. All facets of the different subjects interact, and so you end up with an explosion of possibilities. A computer could calculate these combinations of possible solutions and come up with proposals for outcomes that would be of interest to everyone. Of course, sometimes these are funda- mental social and ethical issues, for which the computer can be no more than an advisor and the human dimension must be the deciding factor.

So, the next formations should be a piece of cake?

‘No, unfortunately, it is not possible to use a computer for this at the moment. One of the problems here is that most of the theories about negotiations on which the algorithms are based are about one-to-one negotiations. For example, it is quite possible to apply an algorithm to bidding on a house. But Dutch politics almost always involves more than two parties forging a coalition. That is much more difficult, the current algorithms cannot handle that yet. They are working on it, though. That is what I like about my field: it is so new, there is so much work to be done.

Apart from the negotiation theory, a negotiation computer must, of course, know what you want. With my Veni grant, I am researching how a computer can find that out. A computer must have data to understand what we find important in negotiations. In the example of the cabinet formation: do you care more about defence or about education? And what are the ratios between those different issues for you? We have developed a method with which you can get an incredibly good idea of what is important to the user to achieve a good result, even with a limited number of questions, sometimes fewer than twenty.

Algorithms are increasingly able to help people with their negotiations and eventually even handle them for you autonomously. Suppose you are talking about a new job. You could put the computer to work with your wishes for salary, days off, and other conditions of employment.The computer could then negotiate with your potential new employer and come back with a possible job proposal for you, without you having to do anything yourself. And there are countless areas where negotiation algorithms can play a role, from the distribution of inheritances to self-driving cars that have to give each other right-of-way.’

In 2020, you joined the Young Academy, a platform for young scientific talents. What are you going to do?

‘I want to change the way we scientists communicate about the social impact of innovations, especially in my field of AI. Take, for example, all the worries about social media bubbles, discrimination by algorithms, loss of jobs due to automation, fake news, impact on politics – that’s quite something. But the same applies to other disciplines, such as biology with designer babies and physics with nuclear energy. These really are innovations with great societal impact.

Scientists often paint a very rosy picture of what a new technology means. This makes sense, because they want to highlight the positive aspects of their work. But when critical questions about negative impact come from the media, citizens or the government, they often have insufficient answers.

That is what I want to think about: how we can make it normal for us scientists to think proactively about potential negative impact and contribute solutions. For example, did you create a new negotiation algorithm? Great, but also think about what it means for people who do not have access to this negotiation algorithm. Or what happens when everyone uses it?’

In 2019, newspaper Het Financieele Dagblad selected you as one of fifty entrepreneurial talents under 35. I read in the accompanying special that you would like to become a professor.

‘Haha, when I was four years old, I said that it was indeed my dream to become a mathematics professor, although, at that time I did not know what that meant, of course. I have been studying mathematics according to plan, but if I ever become a professor, it will be in AI. I would very much like to have my own research group. Fortunately, this subject is fresh, and there is a lot to be done in it. Finding new ideas to explore is not the problem. It is mainly a matter of finding the manpower to work out the ideas I have and the possibilities I see. In addition to my four-day position at CWI, I work at Utrecht University one day a week as an assistant professor, so who knows what the future will bring.’

Someone with your knowledge and skills is also very interesting for the business community. Do they know where to find you?

‘I am currently working with Acumex, a Danish company that wants to use negotiation algorithms when trading in products. If, for example, a construction company wins a large tender to build a residential area, negotiations take place with all sorts of suppliers to purchase stone, cement, steel, right down to the sinks. That is currently done by people behind a desk, over the phone, and with Excel sheets. I am helping this company to see if this can be done online using negotiation algorithms. They called upon my expertise and together we do research to take it to the next level. So, it is not so much that the business community attracts me as a potential employer – I like science far too much for that – but I do enjoy seeing applications in business. As a scientist, you want to lay the foundation on which others can build.’

About NWO-Vidi winner Tim Baarslag

Tim Baarslag is a tenured scientific staff member in the Intelligent and Autonomous Systems research group of CWI, studying AI and negotiation strategies. He is also a part-time Assistant Professor at Utrecht University, a Visiting Assistant Professor at Nagoya University of Technology and Visiting Fellow at the University of Southampton. In 2020 he joined the Young Academy, the prestigious platform of the Royal Netherlands Academy of Arts and Sciences (KNAW) for fifty young scientists who are internationally recognized for their talent, to set up activities in science and society.

Tekst: Wim de Jong

More information