Monetizing Reputation

What is Reputation?

reputation the beliefs or opinions that are generally held about someone or something.

This is a very broad definition of reputation that is intuitive. From this definition we can see that reputation is multidimensional (a set of beliefs) and that it comprises of the “popular” opinion. A reputation has nothing to do with “truth” or “fairness”.

A smear campaign can “ruin” someone’s reputation, meaning that most people will generally believe something they perceive as negative about someone else. Presenting the facts does nothing to fix someone’s reputation unless the majority are open to having their opinion changed.

Scope of Reputation

In the most general sense, reputation is the most popular opinions among everyone who has an opinion about you, but reputation can change depending upon which sub-population you poll for an opinion.

Republicans will hold a positive opinion of their fellow republicans and a negative opinion of democrats. There is an infinite number of ways to group people (gender, race, religion, country, family, etc) and groups can be of any size ranging for the world population down to a single individual.

Depending upon the group you select you can get polar opposite “general beliefs” about an individual.

The Purpose of Reputation Systems

When you meet someone new you have no information about them. This lack of information means that you must assume their reputation based upon your opinion of humanity as a whole. This means you might assume they are out to get you if they can, you may assume they are selfish, or you may assume they are friendly and trustworthy. Each of us has a different default reputation that we assign a random individual.

The default reputation we assign people we have no information about (random handle on the internet) quickly morphs into our stereotype for a group of individuals as we learn some facts about them (race, gender, behavior). The more we get to know someone the better informed our opinions are.

Getting to know someone is time-consuming, expensive, and potentially error prone. When attempting to decide how to behave toward someone we don’t know it is often helpful to ask those we trust what their opinion is of someone. This shortcut works well when there are many people whom we trust who have first hand experience of someone else. It works less well when the people we ask simply repeat what they “heard” from others or their own stereotypes.

Feedback based Reputation

The most basic form of digital reputation is eBay-style feedback on individual transactions. This feedback is robust, nuanced, and from people who had direct experience with the individual. Like all reputation systems, the value of this feedback is highly dependent upon the reputation of the individuals leaving the feedback.

This kind of reputation system takes the approach of “giving you all the data” and allowing you to draw your own conclusions. This is an amount of work that may make sense for a large purchase, but the cost of reviewing the information is prohibitively high for many applications, such as governance.

Algorithm based Reputation

This approach to reputation attempts to combine structured input from a population of individuals to produce a calculated result. Whether this result has any meaning depends entirely upon the robustness and accuracy of the data provided by the people using the system.

There are at least two dimensions required for any such algorithm to render a meaningful opinion:

• do you trust someone to tell the truth about direct experience
• do you trust someone’s judgment about who they can trust

A computer algorithm can compute a personalized estimation of what your opinion would be if you were to manually investigate. All the algorithm needs is a robust well connected network of statements, lines of trust / distrust, and a meta-layer of information about who can be trusted to make good decisions about who can be trusted.

This algorithm gets increasingly complex as you attempt to model additional dimensions of reputation.

Consensus Reputation

Building a self-governing community depends upon defining a “consensus reputation”. The general algorithm outlined in the previous section is able to estimate your opinion, but the goal with consensus reputation is to render community judgements and measure community opinion.

Direct polling is the most common means of arriving at a consensus. This process assumes that all polled individuals have direct knowledge or that they have spent the time asking their social network for an opinion (Feedback based Reputation) and then submitting their conclusion to the poll.

Once again, most people are not willing to pay the price necessary to gather the information. In this case they either abstain (providing no information) or they vote (providing random information at best, or misinformation at worst). Assuming a large enough sample size, the “random” noise cancels out.

In all cases the result of the poll is highly subject to the signal to noise ratio. Too much random noise combined with intentional misinformation can cause the consensus opinion to be “undefined” at best or “wrong” at worst.

The type of opinion is heavily “centralized” in the sense that every individual must participate in every poll. Lack of voter turnout and rational ignorance weakens these systems.

The typical solution for these kind of reputation systems is to have everyone vote on whom they trust to make judgments about others. In other words, representative democracy is nothing more than a reputation system where the “king of the day” has the power to reallocate individual resources through taxation and property forfeiture. Typically a process of appointing judges, and conscripting jurors is implemented.

Assuming such a government restricted itself to decisions regarding the allocation of its own currency, then you would have a fully non-violent self-governing community. Moral problems only occur when an individual has their physical person and property threatened.

Decentralized Reputation Governance

In a decentralized model there would be no need to elect any one individual with the power over the reputation of all others. Under this model individuals have the ability to act in ways that give and take reputation from others. It is a peer-to-peer trading in reputation.

Algorithms can be Manipulated

All reputation algorithms whose computed result can inflict a positive or negative economic outcome to an individual will incentivize market participants to exploit the rules for maximum profit. The success or failure of the system depends upon the ability of honest individuals to defend their reputation against attacks from dishonest individuals attempting to acquire reputation dishonestly.

One of the most common types of manipulation is known as a Sybil attack. This is an attack where people create fake accounts. There are only two solutions to Sybil attacks: use a trusted individual to certify uniqueness of individuals or base the system upon a scarce resource. This isn’t a problem for personalized reputation systems, but for consensus reputation systems it becomes critical.

A resource based system is “for sale” to the highest bidder. This means that its security, reliability, and trustworthiness depends upon that distribution of the resource being in the hands of trustworthy individuals combined with the total value of the resource in question. The harder it is to buy up a majority of the reputation resource, the more difficult it is to cheat a resource based system.

Reputation as a Resource

The concept of Reputation as a Resource is the basis of proof of stake systems. It is also the basis of Steem Power. Those who hold this resource are trusted with the power to distribute this resource to new users. Unfortunately, Reputation as a Resource cannot operate like other resources.

You do not own your reputation. It is popular mythology that we should own our reputation, but your reputation is nothing but the beliefs others have about you. You do not own their thoughts and beliefs. We have the power to impact what others believe about us through our actions, but the absolute power to change an opinion resides in the opinion holder.

If we define reputation with regard to “trust”, it is said that it is “hard to earn, easy to lose”. To base reputation on a resource it must be a resource that is “hard to earn and easy to lose”. If someone can earn a reputation and then use said positive reputation to defraud others while being immune to the loss of their reputation, then the value of the reputation resource as a whole will decline until it is meaningless.

Negative Reputation

If someone who has no reputation does something “bad”, then the result is they should earn a negative reputation. Alternatively, lack of any reputation at all must be viewed as the “most negative”. All unique individuals who verify and prove their uniqueness should “earn” a non-zero reputation.

Property Rights in Cryptographic Reputation

Crypto currency was born out of a desire for honest money that is free from political control. It is highly valued due to its ability to enforce property rights against all aggressors so long as you can secure your private keys.

To base reputation on a cryptocurrency-like resource means that the allocation of the currency must be entirely subject to evolving / changing community opinion. This is the very nature of both reputation and politics.

In a hypothetical resource based reputation token, the community must have the power to vote with their Reputation to remove the Reputation from someone else. This means that when you buy reputation tokens on the market you are betting that other owners of the reputation token will continue to believe you deserve that reputation.

In a free market, wealth is the measure by which you have served your fellow man can be measured by the profits he has earned. The degree to which wealth is a reliable measure of reputation is to some extent determined by the overall degree of thievery in a society.

Stake your Reputation

When someone acts they risk harming their reputation. It is the risk to your reputation that motivates individuals to behave in a society. With a token based reputation system, losing your reputation carries a direct and immediate economic loss. This is compared to a non-token based reputation where your loss of reputation only impacts future income from lost sales.

Reputation needs to be Monetized

A self-governing free-society based upon non-violence needs to have a solid reputation system that is trusted by all. When people trust the reputation system, they know who they can trust in business and who they should shun. They know who to befriend and who to outcast.

Fraud is how bad individuals monetize good reputation today. They utilize the trust others have placed in them to embezzle money. The largest frauds are currently perpetrated by those who have the greatest trust and reputation.

If someone can gain more by selling their reputation directly than they can earn by fraud, then there is no longer any financial incentive for fraud. Fraud will only occur if it can go undetected long enough for the fraudster to sell their illicit gains before the community can act to revoke their reputation.

Historically loss of reputation has resulted in a loss of theoretical future income. If reputation were to be monetized, then individuals face the potential of the very real present value of their reputation.

When the richer in reputation you get the more you have to lose by falling out of favor. To maintain your position of reputation wealth will require increasingly competent public relations. Meanwhile, investigative journalists would have financial incentive to grow their reputation by alerting the public to anything that would indicate abusive behavior.

Conclusion

Reputation need to be monetized by converting it into a token that is entirely revokable by decentralized consensus algorithm. This would necessarily be a new type of token based upon a different kind of “property right”, namely the communities right to assign and revoke reputation. Steemit has already provided the most scalable solution to date for allocating tokens to those who earn reputation with the community, perhaps a similar system could be employed for removing tokens from those whom the community feels does not deserve them.