I don't doubt that the market can sort out the problem of ambiguous or badly-worded contracts (to within reasonable tolerances) but that's not the problem I'm talking about. The key innovation in this proposal is to use a p2p system of consensus witnesses to settle the contracts. It looks to me like the p2p system will collapse into de-facto centralization, at which point the incentive system, which is supposed to be designed to incentivize getting to true outcomes, will actively prevent the market from correcting.
Alright, you didn't like my concrete example of why the system can be self-correcting; so could you offer a concrete scenario of how an existing prediction market contract from Intrade or somewhere could collapse into a perverse equlibrium? (Being partially centralized is not a bad thing: Bitcoin mining is pretty centralized these days. I don't mind if everyone is getting their figures from an authority like the BLS.)
Your example doesn't address the bit that I'm worried about. The p2p witness system that has to get to a single consensus decision to settle outcomes is a different thing from the free market system that allows people to choose the best contracts. The latter is indeed self-correcting based on regular free-market principles.
Intrade settlement doesn't have an equilibrium to collapse into - it settles based on a single, central authority (itself), which is incentivized to supply the right data to encourage people to trade and give them a cut.
To illustrate, let me contrast how it's supposed to work to what I think could go wrong. (It might be that I'm just misunderstanding the proposal, in which case let me know):
A. Supposed to work:
1) Everybody makes an independent decision on whether X happened.
2) Everybody has an incentive to match the consensus.
3) The easiest way to match the consensus is to get the correct answer.
4) The answer with the most votes is the correct answer.
B. The way I think it could end up working out:
1) Somebody publishes a convenient data feed saying what happened, pulling from the BLS and a bunch of other sources.
2) 51% of users use the data feed, since it allows them to participate correctly in every vote.
3) Everybody has an incentive to match the data feed.
4) Anyone who wasn't already using the data feed uses it, or gets spanked whenever their judgement diverges from the feed.
5) The data feed produces bad data, because their systems are buggy or they're getting paid off by someone.
6) Anyone who gets the correct data directly from the BLS or wherever gets punished for it by the incentive scheme.
Now, it's true that the lucky monopolist who ended up as the Bill Gates of data feeds probably doesn't have an incentive to screw it up on purpose, and it probably won't be too bad in practice. Often the winners of natural monopolies are pretty good at their jobs - Windows isn't all that bad... But I doubt this is what the designers intend. You have a lot of cost and complexity with all the voting stuff to make the decisions about outcomes decentralized, but then you end up with a de-facto centralized structure anyway, and worse a centralized structure that's hard to fix if it goes bad. If you knew you were going to end up with authorities anyhow I think there would be simpler, more robust ways to do settlement. (For example, skip the voting and let people specify a list of settlement authorities in the contract and a protocol to choose between them if they diverge. That puts your settlement authorities in self-correcting free-market competition as you describe for contracts, rather than lumbering you with the one who won the battle to be the de-facto go-to guy.)
Anyway let's see what the OP says about the centralization issue - maybe I've misunderstood the proposal, or maybe they've got a solution to it.
[Lots of little edits for clarity.]