Nagorno-Karabakh: Peace ex Machina?

Negotiation support systems (NSS) let computers be brokers in multidimensional and highly complex negotiation problems. In the last 20 years, platforms like Smartsettle and Inspire were used in thousands of serious negotiation cases around the world, many of which were successfully resolved.

Some years ago, the first author of this article was involved in an attempt to use NSS for settling a dispute between the Israeli government, which wanted to shut down one of its departments, and a labor union, which represented the employees who were affected. The main parameters were the number of workers that were to be dismissed and the compensation packages they would receive.

In the beginning of such a process, both parties must convey their preferences to the computer program, which is often operated by a trusted mediator. For instance, the labor union could have stated that in order to give approval, for the first 10 employees that were dismissed it would be necessary to pay 150,000 shekels per head into some social-compensation plan, while for every additional dismissal a payment of 300,000 shekels would be necessary. Also the Israeli government could state complex preferences, for example its indifference between firing one worker or having two workers transferred to other workplaces. The complexity of admissible preferences in NSS is virtually unrestricted, as these are “explained” to the computer in a special programming language.

The negotiation will begin with the computer proposing a solution. If the initial proposal is rejected and changed by one or both of the negotiators, this feedback will be used to generate a new proposal. This leads to a sequence of suggestions in which the computer gradually fathoms the set of feasible outcomes, so that, if things go well, there is convergence to an agreement.

Once a solution is reached which is accepted by both parties, the negotiation will not necessarily stop but rather enter the so-called post-settlement stage. Here, the computer will propose further changes of the outcome which are Pareto improvements (i.e. better for both sides), if there are any. Interestingly, in practice these post-settlement changes are often rejected by the negotiators (cf. Gettinger et al. (2010): “Do they agree once more? An analysis of the factors that influence agreement in the post-settlement phase”, Proceedings of GDN 2010).

NSS has been considered and applied most frequently for labor contracts and in business-to-business negotiations, but sometimes also in international disputes, for example relating to water distribution (Thiessen et al. (1999): “Computer-Assisted Negotiations of Water Resources Conflicts”, Group Decision and Negotiation 7, and several follow-up publications). Recently, an NSS algorithm was devised for sharing the resources of the Caspian Sea among the adjacent countries (cf. Madani et al. (2014): “A negotiation support system for resolving an international trans-boundary natural resource conflict”, Environmental Modeling & Software 51). The distinctive feature of NSS, which makes it so successful, is its piecemeal approach: instead of immediately striking a perfect deal which makes everybody happy, it generates a trajectory of changes, often starting with the status quo, and the parties only have to agree to make small steps one by one.

Could this approach be applied to one of the most deadly and notorious conflicts in the South Caucasus?


The Nagorno-Karabakh conflict between Armenia and Azerbaijan emerged in the last years of the Soviet Union and has been a major obstacle for development and prosperity for both countries ever since. Besides the huge human losses on both sides, the economies of the whole region are severely affected, as the unresolved conflict yields high opportunity costs through unrealized trade and investment as well as through the inefficient use of transport and communication lines (cf. Terterov and Niculescu (2012): “A Pragmatic Review of Nagorno-Karabakh Conflict Resolution: Can Economic Incentives Help Break the Current Stalemate?” European Geopolitical Forum). Everybody knows that except for fierce venture capitalists, investors shy away from turmoil, and having a latent conflict in the neighborhood is a liability for the whole region.

Moreover, excessive defense spending is a burden for both countries. In the 2014 Global Militarization Index of the Bonn International Center for Conversion, which measures the significance of the military sector compared to the whole economy, Armenia and Azerbaijan have Places 3 and 10, respectively.

As the status quo is so unpleasant for both sides, it becomes more likely that – in line with the NSS approach – one can find a different but similar state which is slightly better for both.

One possible starting point is military spending. When both countries would simultaneously decrease their levels of military expenditure, the ratio of military power would remain the same and the probability of winning the war, if it would break out, would not change for either side. This would be a Pareto improvement. Yet, as always, “the devil is in the implementation”: if one country does not reduce military expenses by the same proportion as the other, the balance of power will be changed, which is undesirable for the one who reduced more. Hence, the change must be truly simultaneous and transparent.

While this is difficult, it is much easier in a piecemeal approach than if one aims for a comprehensive solution. One could, for example, begin with a reduction of soldiers on both sides by, say, 2000 troops. Third parties trusted by both Armenia and Azerbaijan could play an important role – both countries have excellent diplomatic relations for example with Israel. The trusted party would ensure that the agreement is kept, i.e. that the number of soldiers is really reduced by 2000 on both sides; while Armenia would not want to reveal its military strength to Azerbaijan and vice versa, both sides might be willing to give the necessary information to the third party. This is similar to NSS, where all negotiators inform the arbitrator about their preferences. The third party could also provide military expertise to ensure that the measures taken do not change the balance of power.

Subsequent steps might go beyond military spending. There are lots of economic, political, and diplomatic parameters that could slowly but steadily be moved in a direction that is good for both sides. Finally, at the end of a long journey, after trust was built and economic ties were established, one might even talk about land; again in a piecemeal fashion, maybe through a sequence of land swaps, village by village, square kilometer by square kilometer. A comprehensive solution to the conflict, however, requires that both parties want their conflict to come to an end. While this seems unrealistic at the moment, during the long process of small changes also the attitudes of the negotiators towards each other may change.


The approach featured by NSS has the advantage that small steps can be made without an immediate change of the overall rhetoric and official stances. Big changes will be avoided by wise rulers, as they might be seen as a loss of face.

Far from losing faces, Armenia and Azerbaijan could gain international influence and prestige if only they manage to gradually sort out their dispute in a civilized way. In the eyes of the world, this would be much more impressive than the kind of futile saber-rattling in which both governments are currently engaged. Armenia would reduce its unhealthy dependence on Russia. Azerbaijan would be able to devote a larger share of its much reduced (or dwindling) oil and gas revenues to develop the economy and create jobs in the periphery.

In the short run, less of the scarce resources of both countries would go into the military sector, and more could be used for schools, universities, pensions, civil infrastructure, and many other things that make a country attractive. At times, the conflict may be useful for the governments in both countries to draw attention away from internal deficiencies, but the improvements in life quality would compensate for the dismantling of this habitual enmity. Hence, peace would not necessarily destabilize either government.

While a comprehensive settlement may be out of reach for the next decade, the piecemeal approach of NSS can bring about small improvements quickly, even if both parties hardly trust each other and do not intend to resolve the conflict altogether. It is high time to get started, be it public or behind closed doors!

Florian Biermann and Aram Grigoryan

29 October 2015 22:32