Kleros: an arbitral tribunal in the blockchain space?

Litigation is blunt. So blunt, that very few (un)willing participants in disputes enjoy being swallowed up by the hassle, time and money it takes to resolve an action through the law courts. It’s little wonder that commercial entities nearly always write into their contracts an option that allows them to resolve a dispute, should any arise, by ‘alternative’ means.

But ‘commercial parties’ can no longer mean just the big corporates. Today, it’s the freelancer, the Etsy retailer, the dog-walker, and the personal trainer. It’s as though we’ve returned to the agora of Ancient Greece, which belonged to solo-traders and entrepreneurs (in our terms, if not theirs). Commercial people like these don’t always have attorneys on retainer. And it’s commercial people like these who feature prominently in the case studies dreamt up by the team at Kleros – who’ve certainly taken inspiration from the ancient Greeks.

Kleros plans to implement a “fast, inexpensive, transparent, reliable and decentralized dispute resolution mechanism that renders ultimate judgments about the enforceability of smart contract”. (See Kleros’s Whitepaper, page 1.)

At a glance, their plan feels inventive in many of the right places. For instance, it does not try to overwrite what already works in practice, and it keeps a familiar system of hierarchical “sub-courts” with an appeal process to allow parties to challenge decisions.

The real innovation lies in the adjudicator(s) at the heart of the model. Kleros works by incentivizing ‘jurors’ to perform the work of an arbitral tribunal: they vote on how to resolve the dispute, and they give reasons for their vote, just as arbitrators do. The requirement that jurors justify their votes is crucial to the trust and integrity such a system must engender in order to be successful in the blockchain space. It also speaks to a uniquely human ability that algorithms so far lack: to be able to articulate logical (but non-numerical) reasons why a certain conclusion has been reached.

This matters: machine learning cannot do so, but the losing party to a dispute can read an arbitrator’s reasoning, fully comprehend it, and agree or disagree (and on that basis decide whether or not to appeal).

This lies at the heart of what ‘justice’ is. But it also leaves the door open for mischievous jurors to adopt the reasoning of another, doing no work for the same reward; or, perhaps worse, for jurors to attempt to influence each other without being held accountable. What is Kleros’s solution? Hiding each juror’s vote to ensure that any one is not influenced by any other. (See Kleros’s Whitepaper, page 5.)

If this can be made to work in practice as smoothly as on paper, it will attract and incentivize jurors who are (or who will become) expert in certain domains, such as “freelancing”, “air transport” and “car insurance” (to name a few of Kleros’s proposed “sub-courts”). And this starts very much to resemble adjudication by a panel of experts.

But that is not to say the plan is flawless. The probability of any given juror being selected “is proportional to the amount of [tokens] deposited” by that juror. Moreover, the increased chance of being selected is likely to increase the “weight” of that juror’s vote. This could mean in practice that voting-power is acquired and monopolised simply by having more money to deposit into the system.

That is troubling. A fairer system would ensure that credibility as a juror is earned, not bought. Kleros finds a way to penalize “incoherent” jurors, so it can surely find a way to reward jurors adjudicate coherently (and consistently). That reward should, if anything, be reflected in the “weight” of that juror’s vote. It all comes back to finding ways to create, embed and demonstrate the trust necessary to make such a system work.

But if it can indeed be made to work, then Kleros, and other systems like it, will demonstrate the potential of a model that will be integral to the development of smart contracts. In time, smart contracts will, hopefully, incorporate clauses stipulating that the parties may resolve disputes by means of this, and similar models of alternative dispute resolution.

---

This is my entry to the @OriginalWorks Kleros competition. I have no affiliation to Kleros, apart from being an interested observer!

kleros2018