While some studies can go against the wisdom of the crowd idea, we strongly believe an incentivization model within the Finnoq protocol is a solidifying force that holds crowd wisdom together. As no verifiable outcome (open-ended question) is asked on the Finnoq protocol, the need to reward truth and honesty is imperative. In the first part of this two-part series, we explore the staking function as well as game theory subsets. Then, we look at short-term parameters within the Finnoq protocol used to incentivize participants. With these concepts and parameters in mind, participants will naturally generate trustworthy collective opinion statements.
Basics: Staking, Nash Equilibrium, and Schelling Point
Before jumping into how the Finnoq protocol incentivizes knowledge and honesty to guarantee crowd wisdom, staking as well as two additional game theoretical notions have to be discussed: the Nash Equilibrium and Schelling Point. All three are interrelated to each other and serve as a useful foundation toward understanding the Finnoq protocol.
The Finnoq protocol requires that FNQ Tokens are locked for a free, definable period of time, known as “staking”. In doing so, community participants are able to vote and receive rewards by placing their FNQ in Staking Contracts. The longer FNQ Tokens are staked, the more potential rewards one can gain. A staking function dissuades users from attempting to “game the system” through untruthful responses, because they will lose credibility, opportunities to stake further tokens, and actual FNQ by doing so. Crowd wisdom research has shown that when there is something to gain and lose (penalties and rewards), participants respond more truthfully and with greater attention. This is an important game theory subset, but must go in conjunction with the two discussed below.
The Nash Equilibrium was discovered and pioneered by John Nash in 1950. He demonstrated games when a set of actions of each player is finite, possessing at least one mixed-strategy equilibrium. The Nash Equilibrium is utilized when players are faced with incomplete information. In a Nash equilibrium, every member of a group makes the best decision possible, and nobody does better by changing strategy. Distributed computer systems such as the blockchain benefit greatly by being totally “cheat-free” when nothing is gained by shifting strategy.
In the same article, the Schelling Point is discussed as a solution “that people will tend to use in the absence of communication because it feels special, relevant or natural to them.” For example, Thomas Schelling posited that if asking New Yorkers to determine a place to meet up, they’ll choose Grand Central Station, because of how common and natural it feels. Schelling argued that the ability to coordinate without communicating was key to understanding how strategic games are solved. The literature has gone further, noting that when there is an implicit incentive to coordinate responses among group participants, it can be done without communication. In other words, participants seeking only to be part of the majority can coordinate with or without communication, putting a damper on traditional scoring methods (i.e. simple averaging). This means that crowd wisdom can be tainted.
Under circumstances when participants have no verifiable outcome, the tendency to deviate from a chosen strategy, deconstruct the Nash Equilibrium, and funnel responses into a Schelling Point can be high. Understandably, as verifiable outcomes (such as what the weather will be or who will win a match) possess finality at the end of the wager. Finnoq’s protocol takes these principles and constraints into account when generating true crowd wisdom.
The Finnoq Protocol's Short-Term Incentivization Structure
How does the Finnoq protocol incentivize its users to give candid and knowledgeable answers in the short-term? Through the following ways the Finnoq protocol can ensure this:
- Bayesian Truth Serum (BTS) scoring method
- Surprisingly Popular Algorithm (SPA)
- Sub-stake per vote
Bayesian Truth Serum (BTS) Scoring Method
The BTS scoring method aggregates subjective and truthful responses when objectivity cannot be obtained by the study itself. Each participant gives a personal opinion and the percentage distribution of others’ answers. The result is broken up into two considerations: information and prediction score. The information score is the accuracy of each personal opinion, while the prediction score is the accuracy of the percentage distribution among other participants. By doing so, the BTS can be leveraged into an algorithm that obtains better results for those designing studies.
Surprisingly Popular Algorithm (SPA)
The SPA is a recent mechanism - inspired by the BTS - used for getting better results when not enough people know the best answer. They will create a collectively wrong result under simple averaging methods. The SPA’s impact can reduce errors by up to 24%. Under such a design, respondents give their opinions and simultaneously predict the percentage distribution of other respondent answers, as in the BTS method. The algorithm chooses the most-popular response than originally assumed (also known as the so-called “surprisingly popular answer”). Due to the fact that this is a new research topic, the algorithm is evolving. When employing the SPA compared to other useful algorithms, results were more-stable and bolstered with the SPA method.
The Schelling Point could be seen as a challenge at best and exacerbation at worst when employing BTS and SPA. By asking respondents to focus part of their attention on other participants, the natural tendency to move toward a Schelling Point is obvious. Thus, the Finnoq protocol relies on many other mechanisms to ensure incentivization. In the short-term, a nifty idea has been baked into the cake: a sub-stake on each vote.
Sub-Stake per Vote
Participants giving their candid knowledge must sub-stake on each vote. The purpose of sub-staking FNQ Tokens is to reward the confidence of knowledge. When a participant is more confident and gives an opinion in-sync with the crowd, the participant has more to gain. Alternatively, when participants are more confident and are “wrong”, they have more to lose.
For those losing their sub-stake on a given vote, these lost earnings go to those who were part of the crowd (where distribution is commensurate to confidence level). This sub-stake refocuses the participant on individual success, indirectly creating part of the criteria necessary to ensure a Nash Equilibrium. One’s confidence level on each vote is scored over time, and its implication for reward possibilities will be discussed in the second part (stay tuned).
Holding off the Schelling Point and generating a pure Nash Equilibrium is our challenge and interest in terms of incentivizing participants who give straightforward and knowledgeable opinions. The more factors placed into the system, the harder it is to find obvious, uncommunicated Schelling Points. Already with the three mentioned above, it becomes more difficult to just vote with the majority in hopes of obtaining rewards. We can determine if knowledge is based on an individual or perceptions of what the others will do; further, participants must sub-stake some FNQ to demonstrate confidence in each vote.
On top of this, three additional factors must be considered. Part two will step even further into incentivization, from a long-term perspective. Stay hungry for knowledge until then.