That’s a quite interesting debate.
I think the main goal of Kleros is to allow parties to contract with predictability:
- You do the job, you get paid.
- You submit a valid item, your deposit get reimbursed.
- A storm occurs, you get your insurance policy paid.
In all those cases, the fair outcome is the predictable one, so ruling on those is easy.
Should we just enforce the contracts as they are written? Probably not, otherwise someone would end up using Kleros for an automated assassination market and we probably don’t want that to happen. This would definitely not be fair.
If you do nothing, your vote is not taken into account, if no one votes, the default result is refuse to arbitrate. It should always be better to vote than not to so let’s remove the last option to simplify things.
Let’s take a look at the general court policy:
“Refuse to arbitrate” should be used for disputes where both sides of the dispute have engaged in activities which are immoral (ex: refuse to rule on an assassination market dispute).
Immoral activities include: Murder, slavery, rape, violence, theft and perjury.
In this case, ruling 1 and 2 would lead to murder, which is immoral, so according to the court policy, you should vote 0: refuse to arbitrate.
This is the predictable outcome. Note that the policy does not prevent people to vote for options leading to immoral outcomes, it mandates to vote 0.
Is it also the fair one?
It may not be obvious at first sign, as we could think that the fair outcome would be the one that leads to lowest suffering. And 0 immediately leads to 2 deaths instead of 1. But we should look at the long term effects.
If the arbitrator cede to this blackmail and effectively votes 1 or 2, there would be only 1 death in this ruling. But then the blackmailer would have a satisfactory ruling and could continue to use Kleros to decide who to kill. Even worse, other malicious people could see Kleros as nice tool for it and use Kleros to kill more people. Moreover, all the protection about immoral uses could be rendered useless, as malicious people could then just make 0: refuse to arbitrate worse than other outcomes to force jurors to vote.
So voting 0 would also be the fair outcome as it would minimize the amount of suffering on the long term.
In this ethical dilemma example fair and predictable are aligned. But it may not always be the case. I think that’s the role of governance to ensure fair and predictable outcomes are aligned.