On TCRs and Predictions Markets: Counterweights and Collusion
Egging Calvin Ayre: Satoshi’s True Vision
On April 14th, an Augur market opened with the question of whether or not Calvin Ayre will get egged in the next three months. The market’s resolution source is noted as “General knowledge,” with the market resolving to “yes” should Calvin Ayre get hit by an egg within this timeframe. Along with this is a media requirement showing the egging, and an important note stating that a video of Calvin Ayre egging himself won’t count.
However, bundled with these requirements is the market having to resolve based on a second event: a member of TCR Party creating a Twitter poll with the question of “Should this egging video of Calvin Ayre be invalidated on Augur.” Not only must the event occur for the market to settle on the event occurring, but a secondary event, the Twitter poll, must also be satisfied in order to settle to a “yes.” This forces the hand of the reporter to only settle the market as “yes” if the following conditions are met:
- A video is posted of the egging
- A TCR Party member starts a poll to ask about the validity of the video
- Twitter comes to consensus on the video not being staged
There are plenty of immediate issues associated with this approach. The first is the fact that the bet is leveraging the TCR’s current membership list at the time of the creation of the market rather than putting a hash of the list itself in the description (as noted by Eric Wall). The second issue is that there’s no incentive for anyone on the TCR to post a poll to begin with which would throw off the reporting of the market even if a media artifact was presented. The third issue is that a Twitter poll is presented typically to an open-environment, allowing the voters to be anyone on Twitter rather than a fixed set of arbitrators (making it easy to Sybil attack).
How do we rule out the manipulation of a predictions market with an extremely narrow, individual focus, and the ability for the event to be easily staged?
Could a contained TCR provide a larger, trusted arbitration body for a predictions market with money on the line?
For example, if the event occurs, it would require some kind of media artifact to be produced in order to be reported on by an unbiased oracle. However, what is to stop the subject of that narrow market from creating a version of the event that wasn’t “planned” by a particular predictions market, and reaping the benefits?
Before going any further — it would be helpful to go over what TCRs are and what TCR Party is —
Token Curated Registries and TCRParty
TCR Party is a Token Curated Registry on Ethereum’s Rinkeby testnet. A Token Curated Registry is a list managed by a public group of token holders. There’s a nomination process (to add members to the list), a challenge process (to challenge listings), and a voting process for token holders. For more information on the primitive, check out Mike Goldin’s primer. TCR Party came as an experiment in both abstraction and subjective curation.
TCR Party is a registry of 99 participants that are supposed to be representative of the “best minds on Crypto Twitter.” TCR Party became a natural fit for the Augur market because the event is directly related to Crypto Twitter.
Currently, TCR Party has 529 users since its launch, 53 of which were active in the last week. It has been relatively quiet for the last two months, as its hype cycle has settled and the registry has slightly solidified.
TCR Arbitration for Isolated Incidents
What considerations are there in arbitrating outcomes for predictions markets?
Let’s begin with examining the prediction’s market size and availability of outcome information against its reporting difficulty.
To start with, the widest scope of a market falls along the lines of a question like “will there be oxygen on earth in two months” due to it being accessible to any reporter from any location and the topic. This is an event with a universal scale. However, if the market settled “no,” and we’re all still alive, the reporter is obviously lying.
Going even deeper, we have markets of a general scale that have widely accessible information. For example, this could be a large sporting event or the weather of a particular region with a question such as “Is it going to rain in Brooklyn, New York on August 15th?” The reporter can either be from Brooklyn, New York, or have enough verifiable information to prove that claim.
The next layer is at a localized scale with information that might possibly be unclear. A market like this could be for the results of a particular local junior high school baseball game which very few people attended — but the results could be obtained through someone recording the statistics of said event. The event was scheduled formally, happened in a somewhat populous area, and had humans accounting for the results. There may be participants interested in betting on this kind of event.
The last level is extreme isolation. This is a hyperlocalized isolated incident happening at any given moment within a particular timeframe. This is a great example of a market that might need a second level of arbitration due to the time notice and horizon (near zero). Without any form of media artifact, the event may have never “happened” due to not enough proof being presented to those following such a market. These markets come with extreme risk, as they are difficult to report on and verify.
If that tree falls, how can we prove that it made a sound? When predictions markets narrow in scope, reporting on the outcome of an event becomes a bit harder due to a lack of publicly available information.
It’s important to note that these markets might not even generate enough interest to be worth reporting on, as the cost of reporting and sleuthing may outweigh the benefits associated with reporting on more generalized markets. The crypto community may be interested in an egging market like the one presented and it may generate a few hundred dollars of interest, but the time associated with validating the video may not be worth the extra arbitration steps unless they were easily automated, or there was enough of a financial incentive involved.
For example, a market on baseball with millions of viewers and plenty of sources reporting the same outcome becomes a lot easier to report on than deciding whether or not a video was staged, which is also rife with contention. Typically the highest value markets will fall somewhere between general and localized, as there’s a mass of participants that both have access to the event and can even self-verify if something is amiss with the reporting.
However, there could be a small part of the value zone that requires additional arbitration; a localized market that has a high enough amount of liquidity to merit additional forms of arbitration to be certain about the outcome. This is where designated TCRs may come in handy as particular lists of “legitimacy.”
TCRs as Arbitrators
To begin the process of using a TCR as an arbitration body, we need our primary base TCR of individuals or entities. The token curated registry would represent a base that can understand the market’s constraints and be able to provide a just means of arbitrating. In the case of the Calvin Ayre market, TCR Party was used due to its nature as a natural fit. The Augur market most likely came from the drama currently occurring between Ayre and the community through Twitter — and who better to arbitrate than a list that is supposed to be 99 crypto-savvy accounts.
A TCR as an arbitration body may require forms of quasi centralization to achieve the necessary registry that would be a fit for the market. The reason why TCR Party worked was due to the interface being through Twitter, transactions were abstracted, tokens weren’t sold, and the list was seeded by our centralized team. This may have to also apply to other cases of localized markets, as they may not be “crypto-friendly” topics which TCR Party registry members can handle.
The way I see TCRs coming in as arbitrators is through a voting mechanism that is minimized to the TCR itself. As presented in the existing Augur market, a TCRP member would have to create a poll on Twitter which opens up plenty of different ways to manipulate that vote. To avoid this, we need a verifiable way in which members can signal their choice on how to verify the item in question.
The initial requirement is that it will probably need to be in a familiar medium for the registry, as all participants may not be up-to-speed on how to signal in a “blockchain-based” way. For example, if we were to use TCR Party for this purpose, registry members eligible to participate in these votes would already have a multi-sig set up with our bot and would be sent a message allowing them to respond with their vote. A transaction would then be created on their behalf for verification purposes.
In order to incentivize registry participants to signal and participate (while avoiding complete apathy), there has to be a way in which registry members can be awarded for their participation. In this case, being an arbitrator leads to an additional incentive outside of simply being on the registry for the registry’s sake. This becomes the trickiest part because of the way in which Augur markets are currently set up. It would most likely look like an additional fee that’s payable to an address set up by the TCR creator meant for this kind of distribution.
Addressing Issues
This is by no means perfect and comes with plenty of issues:
Apathy — The first issue that needs to be addressed in such a system is obviously voter apathy and the value of arbitrating as a registry member. If there isn’t a high enough fee going to registry members, they won’t be arbitrating out of the kindness of their own hearts. Also considering that the necessity for additional arbitration comes from markets that are localized or hyperlocalized, the markets that need help might not provide enough money to be worth the time. The only scenario in which this would work is where Augur explodes in terms of interest and there is enough open interest to support these isolated events. However, even then it would mean an excessive amount of activity just to earn a minimal amount of fees.
Inactivity — As mentioned before in the TCR Party example, a mechanism would have to be built in order to facilitate the voting, as well as administer payouts. However, some TCRs may simply have uninvolved parties as registry members, due to the fact that any participant can nominate anyone to a registry. For example, in TCR Party, one didn’t need to be signed up in order to be added to the registry. Even the pope was nominated at one point.
Bribery — Bribery is made easy due to the fact that these registry members will most likely be known and they directly impact the reporting on a market. For example, if the Augur egging market were to gain enough open interest, TCR Party had a private voting mechanism, and Ayre was to stage the incident, Ayre could simply bribe registry members due to the fact that most of them are publicly known and accessible on Twitter. This could obviously occur with any other arbitration body because they will most likely be known actors entrusted with validating a certain event (as they have to have either a particular skillset or inherent knowledge of the market).
Manipulation — Considering the TCR has a final say in the reporting of the event, registry members could simply place bets on what is occurring in their private vote. Because this form of arbitration will boil down to subjectivity, it may be easy for arbitrators to bet on the side of the desired outcome from TCR members before the vote concludes in order to play the market. Not only will they earn from the market, but they will still earn from fees as well.
Token Curated Reliability?
Although TCRs may seem like an interesting form of arbitration layer for reporting on predictions markets, the downsides weigh heavily on the concept’s potential. Considering the fact that the value in having this arbitration comes only from particular markets with low chances of having high value, the incentives aren’t aligned for registry members. Registry members may not even be involved with the curation of their respective TCRs, which also ties into apathy and inactivity. The chances of bribery and internal manipulation are also high if enough value is on the line which defeats the moonshot scenario of these tinier markets taking off.
Stick to one reporter, and don’t make subjective reporting requirements.
