Quote from: razdva on July 13, 2016, 12:35:24 PM

Understand it, but... An example. We have 2000 rep holders and 10mln events to reporting period. If we take 100 report per 1 holders, we will have 40 holders per 1 events, with different amount of rep. So it will be very centralization and unreliably. Also if it will be not high liquid market, so reporters be a less then 40 per one events?
What if market itself will decide the result of events? If event for a long time traded on liquid market with the odds close to 100%, this market outcome may be determined without the participation of reporters with same 100% accuracy. Because It will be infinitely costly bet against events, whose outcome is already known
I will try to read blog post, but not native language

I'm not sure I understand your question. Are you asking whether there will usually be a lot of reporters for each event? The answer is, it depends -- the number of reports (rep) assigned to each event depends on the volume in that event's market. Reporting also has two "backstops" to make sure events resolve correctly:

The first is that after reporting, events' results can be appealed by posting a bond, which re-submits the event for reporting, this time to all reporters. If the second result is the same as the original, then the bond is lost. If the second result is different, then the bond poster gets their bond back, plus Rep from the reporters who reported incorrectly the first time.

The second is (if the first backstop still gets the answer incorrect) that any Rep-holder can pay to "fork" Rep, which splits Rep into two classes. This works as follows:

Quote

In one class the reported outcome for whatever event was the cause of dispute is said to be right, and rep is redistributed accordingly.

In the other class/branch, the event is readjudicated and remove the outcome on the pre fork branch as a possible reporting choice, then rep is redistributed as normal, except people who said the original outcome in the original fork lose rep.

Note: Whichever outcome was truly the correct one would determine which branch had rep that actually held value. This would be akin to a Bitcoin hard fork scenario. The winning fork, of course, would be the one with the most voluminous markets, which would be determined by the fork that represented reality, which would have a higher market cap. Markets go to the most popular fork by default after 2 months.

You can see the nuts-and-bolts of the 2nd backstop on the forking.se contract.