26 ms·

> Settlement - once the reporting has been complete, and it’s verified by the community, the bettors who bought shares on the correct outcome are paid - as is t

> Settlement - once the reporting has been complete, and it’s verified by the community, the bettors who bought shares on the correct outcome are paid - as is the market creator. Can anyone comment on how settlement actually works? A persistent problem with new tokens is handwavy explanations of absolutely core features. With a human legal system, humans can argue over intent. When “code is law”, pure intentions alone will not do. I would love to use a functioning prediction market, and I believe it may be one of the legitimately useful non-money applications of blockchain. But I really wish they would focus on explaining how it solves the key problems rather than marketing the concept.

otoburb 8y ago
>>Can anyone comment on how settlement actually works? The Augur whitepaper[1] explains the gory details, but at a higher level when the market creator defines the prediction market and event states through a smart contract, they must also specify the "designated reporter" and the "resolution source". The designated reporter should be somebody you trust (perhaps even yourself) to report on the event using the specified data source. If the final event state is invalid, then you (the market creator) forfeit your validity bond, which is a form of punishment for creating a market that ends up in a state of ambiguity. If the event state is valid (i.e. one of the set of outcomes that can be unambiguously determined) then the settlement proceeds. After settlement finalizes, reporters get their REP token payout, market creators take their ETH cut, and traders who still hold shares corresponding to the settled reported outcome get their ETH. I believe each reporting cycle is still 7 days. There are some edge cases where an event outcome can be disputed by other reporters in case the designated reporter is incorrect, but the system economic incentives are such that this type of behaviour is heavily discouraged. [1] https://www.augur.net/whitepaper.pdf https://www.augur.net/whitepaper.pdf
prepend 8y ago
Likewise, I was trying to figure this out. And haven’t yet. In most prediction markets the reason for centralization is as the market maker or arbiter. For example predictit.com will definitively say if some testified or not during the market term [0]. So arguments over what testifying means get handled fairly. In augur the community seems to determine this based on how much they spend to reconcile the transaction. So lots of people could tip the scale. On big tickets this would likely attract attention, but on smaller items it could be a problem unless members of the community have reputation and identities. It seems like, from reading the faq [1], that it really just boils down to a proof of stake consensus method with multiple rounds of dispute before and after a market is finalized. With a documented fork procedure. So if consensus gets manipulated, it is very clear and the market itself loses reputation through that they tried to design in their fork process. But would also get likely checked by a “hard fork” away from whomever is running that particular blockchain and manipulating accounts. The “ethics arbitrage” opportunity would be to look for markets where the outcome is realistically fuzzy “did someone lie” or has an invalid outcome and tip the scales without detection. [0] https://www.predictit.org/Contract/9262/Will-Trump-testify-to-special-counsel-in-2018#data https://www.predictit.org/Contract/9262/Will-Trump-testify-t... [1] https://docs.augur.net/#tentative-outcome https://docs.augur.net/#tentative-outcome
- JackFr 8y ago
  Yeah, so I read the white paper -- fairly closely at first, but as I got bogged down in the details of 'first public reporter' and the formula for the size of the 'dispute bond' and disputes etc., it became very clear that despite all of the thought and hard work that went into this, that the link between the real world and the reported outcome is reliant on trust and community norms. And the more one thinks about it the more obvious it becomes. Two ways to think about it: 1) The initial market is betting on an outcome -- how is that different from the dispute resolution phase, which can also be reduced to betting on an outcome. 2) Imagine a purely imaginary event -- and read the whitepaper with that in mind. All of the the processes set up still function perfectly -- as designed. In fact there is no need of a real event. So if there is no need for a real event, why is there an impetus to report a real outcome even if there is a real event. That isn't to say it can't work, but if it does it will be through trust and community norms, not Ethereum.

otoburb 8y ago

>>Can anyone comment on how settlement actually works? The Augur whitepaper[1] explains the gory details, but at a higher level when the market creator defines the prediction market and event states through a smart contract, they must also specify the "designated reporter" and the "resolution source". The designated reporter should be somebody you trust (perhaps even yourself) to report on the event using the specified data source. If the final event state is invalid, then you (the market creator) forfeit your validity bond, which is a form of punishment for creating a market that ends up in a state of ambiguity. If the event state is valid (i.e. one of the set of outcomes that can be unambiguously determined) then the settlement proceeds. After settlement finalizes, reporters get their REP token payout, market creators take their ETH cut, and traders who still hold shares corresponding to the settled reported outcome get their ETH. I believe each reporting cycle is still 7 days. There are some edge cases where an event outcome can be disputed by other reporters in case the designated reporter is incorrect, but the system economic incentives are such that this type of behaviour is heavily discouraged. [1] https://www.augur.net/whitepaper.pdf https://www.augur.net/whitepaper.pdf

prepend 8y ago

Likewise, I was trying to figure this out. And haven’t yet. In most prediction markets the reason for centralization is as the market maker or arbiter. For example predictit.com will definitively say if some testified or not during the market term [0]. So arguments over what testifying means get handled fairly. In augur the community seems to determine this based on how much they spend to reconcile the transaction. So lots of people could tip the scale. On big tickets this would likely attract attention, but on smaller items it could be a problem unless members of the community have reputation and identities. It seems like, from reading the faq [1], that it really just boils down to a proof of stake consensus method with multiple rounds of dispute before and after a market is finalized. With a documented fork procedure. So if consensus gets manipulated, it is very clear and the market itself loses reputation through that they tried to design in their fork process. But would also get likely checked by a “hard fork” away from whomever is running that particular blockchain and manipulating accounts. The “ethics arbitrage” opportunity would be to look for markets where the outcome is realistically fuzzy “did someone lie” or has an invalid outcome and tip the scales without detection. [0] https://www.predictit.org/Contract/9262/Will-Trump-testify-to-special-counsel-in-2018#data https://www.predictit.org/Contract/9262/Will-Trump-testify-t... [1] https://docs.augur.net/#tentative-outcome https://docs.augur.net/#tentative-outcome

JackFr 8y ago
Yeah, so I read the white paper -- fairly closely at first, but as I got bogged down in the details of 'first public reporter' and the formula for the size of the 'dispute bond' and disputes etc., it became very clear that despite all of the thought and hard work that went into this, that the link between the real world and the reported outcome is reliant on trust and community norms. And the more one thinks about it the more obvious it becomes. Two ways to think about it: 1) The initial market is betting on an outcome -- how is that different from the dispute resolution phase, which can also be reduced to betting on an outcome. 2) Imagine a purely imaginary event -- and read the whitepaper with that in mind. All of the the processes set up still function perfectly -- as designed. In fact there is no need of a real event. So if there is no need for a real event, why is there an impetus to report a real outcome even if there is a real event. That isn't to say it can't work, but if it does it will be through trust and community norms, not Ethereum.