This is a great start to a revolutionary, innovative idea. There are things that will need to be addressed in order for it to really revolutionize legal systems. For example, a lot of criminal law can be very emotional. However, if Oracles know that they will earn based on what the majority votes, many of them may actually vote against what is just/right and instead choose what they think the rest of the majority is likely to choose because of emotional or other factors. Oracles need to be given an incentive to pick what is actually "just" regardless of what they feel the majority or the "mob rule" is most likely say. It can be hard enough for many people to stand up for what is "right" despite what everyone else thinks, and it becomes harder if you know you will be penalised for it (not rewarded/paid for your work).
Additionally, sampling factors will need to be addressed to reduce bias. If you allow any Oracle to vote on an issue, you are going to have very skewed results because a certain issue is likely to chosen by many of the same people who have an interest in that issue, rather than an objective view. Also, different cultural and other factors can make a big difference, and there need to be constraints somewhat similar to jury selection (though hopefully better) in order to ensure more objective results.
This idea is great, and I hope it becomes further refined. I would recommend they review and apply some of the similarities to things tested out in other countries, etc such as that discussed in wired: https://www.wired.com/2012/05/st-essay-voting/ and implement those ideas into this concept to make it work as well as possible with as little bias as possible using correct statistical randomized sampling as much as possible.