I disagree, Chris. Much experience on FX has shown that interesting questions (those that aren’t routine repetitions of previous questions) often result in realities that diverge from the obvious expectations of nearly everyone involved in describing the possibilities. In those situations, we’ve found that trying to interpret intent leads to more confusion than sticking to the letter of the question as asked.
If a [prediction market] sticks to its written description of what the claims mean, then careful readers are rewarded, and they learn that they have a good chance to predict how the judge will interpret the question and events in the world. If questions are determined based on “intent”, then everyone has to spend time deciding which aspect of the question the judge will decide was more important, when reality decides not to conform to the question’s expectations.
Sometimes (as you argue was the case with the North Korea question) the result is surprising and disappointing, but choosing the other approach leads to much less participation as people who see that something surprising is preparing to happen or has happened back out of their bets rather than waiting to find out what the judge decides is important. I’m much happier when the participants spend their time figuring out what will happen in the world, rather than when they have to spend their time predicting how the judge will react. Strict construction gives us a predictable world.