In cyberspace, parties meet that would often not have met in the offline world, because they physically live in far-away countries or on different continents. Outside cyberspace, ordinary consumers do usually not enter international agreements. In cyberspace, they engage in small or medium transactions, which they would usually not have done in the offline world. In these cases, courts are too expensive mainly because of costs of filing, travel and legal counsel and they are too slow. Alternative dispute resolution methods except arbitration are often not effective enough, because they produce case outcomes that are not legally binding enough. Such methods only engender contracts - if they work. To be enforceable by state authorities, contracts again require a judgement -which is too expensive. As it is expensive, the winning party will have only limited incentives to go to court, and the losing party will also have only limited incentive to obey the contract, because it is unlikely to become enforceable. Arbitration could provide this effectiveness, because the recourse to courts is only minimal and therefore much less expensive. But from this legal effectiveness follow other difficulties. Arbitration requires the parties to give up rights, which does not induce trust and which is the reason why arbitration currently still faces a series of legal obstacles. For consumer claims, where the cost problem is greatest, there is for instance a problem of arbitrability under a number of laws. So how should small and medium-sized disputes by resolved in cyberspace? Non-binding arbitration - a dispute resolution process akin to arbitration, but not binding under the public legal system, its effectiveness relying on in-built enforcement mechanisms - may be an answer to this question. Non-binding arbitration, I contend in this article, may in fact be more binding than traditional arbitration, because it is effective without being subject to so many legal obstacles, and it may be designed so as to avoid many issues related to costs
