Crypto NYC hosts Relevant
On Tuesday, June 19th, Crypto NYC hosted Relevant for our weekly lunch presentation. Relevant’s presentation was led by one of its co-founders, Slava Balasanov. Relevant was founded by Balasanov and Analisa Teachworth, previous co-founders of 4real.io, a digital production agency. Relevant’s mission is to create a digital platform that incentivizes quality content curation and production. Roughly thirty members of the Crypto NYC community were in attendance. Balasanov gave a thirty-minute presentation followed by a Q&A session with the attendees.
Relevant’s mission is to solve the problem of increasingly poor content that is flooding Web 2.0 properties. Relevant’s team views the problem of poor quality content as stemming from the poorly designed incentive structures built into the Web 2.0 infrastructure. Advertisers track clicks, which incentivizes clickbait and spam production, for example. The destabilizing political impact of fake news adds gravitas to these problems and adds urgency to finding a solution. Relevant hopes their application can lead a pivot away from this model by creating a token backed qualitative metric which will encourage quality content production.
Relevant is producing a curation application that uses a reputation-based system to rank content. Participants in the Relevant network are ranked according to their expertise and previous contributions to the network. A prediction market incentivizes participants to rank content, with the prediction market helping to determine where curated items rank in the Relevant application.
The Relevant ecosystem is organized around communities that cover specific areas of content, analogous to subreddits. Any Relevant user will have the ability to bootstrap a unique community. Each community will have an admin or team of admins that manage the community. The Relevant team envisions each community will be led by a real person, or at least a recognized avatar, to add legitimacy and staying power to the community. These admins will also play an important role in creating the baseline trust that will be useful in protecting the network against attacks.
Participants in each community will have multiple ways of interacting with Relevant. Users can build their reputation by commenting on work that has been curated by the community. The more the community values a person’s participation, measured by interaction and likes, the more that participants reputation grows.
Reputation and trust can be built in the system directly and indirectly. If one participant positively engages with another user’s work, that will build the reputation of the creator of the content. Reputation can also be transferred transitively along a chain of trust. If Pam approves of Jim’s content, and Jim approves of Dwight’s content, then Pam will indirectly approve of Dwight, as an example.
Relevant’s prediction market is similar to Steemit’s system. Participants can predict how helpful a piece of content will be to their community. There is a brief window of time after content is posted where participants can make their predictions. Participants can choose how many of their tokens to stake when making each prediction. When content is ranked for display, participants are rewarded for how accurate their predictions were, commensurate with the size of their staked position and factoring how quickly they made their prediction.
Crucially, the ultimate content ranking is based on the reputation of the participants in the participation market, not the size of their stakes. This discourages whales with large token holdings from creating self-fulfilling predictions by massively staking their predictions and skewing the ultimate rankings.
Relevant will be launching an ERC20 token that will power the internal economy of the Relevant application. Relevant plans to build their token with a built-in annualized inflation rate of 10%. Tokens generated via inflation will be deposited in a reward pool and will then be distributed to participants in the network based on the quality of their predictions. Tokens are awarded to top posts for each time period, and then distributed to participants who staked on top posts.
In any system built on reputation, maintaining the integrity of the reputational system is of primary importance. Since Relevant is a decentralized system, it needs a way to punish and exclude bad actors that functions independently of manual human supervision. Networks such as Facebook have struggled to combat fake news, highlighting the difficulty even centralized systems have in policing content on their networks.
Relevant has designed their system with this concern in mind. The most common method of attacking a reputation-based system is called a Sybil attack, in which a system is flooded with fake profiles that distort rankings. Relevant uses Eigentrust++ to protect the network from Sybil attacks. Eigentrust++ was originally built to validate nodes in P2P file sharing networks.
Eigentrust++ uses a set of initial trusted users as a foundation for users building their reputation. Participants in a community must have multiple connections to the initial trusted set in order to build their reputation. This prevents two false participants from repeatedly upvoting each other and building undeserved reputation scores.
As each community grows and changes, the trusted set of users will change as well. Relevant will use a modified version of a Token Curated Registry (TCR) to manage the set of admins and trusted users.
A TCR is a decentralized registry which token holders are incentivized to manage. Relevant’s TCR will incorporate some of the unique aspects of Relevant’s governance structure into its management. Management of Relevant’s TCR will incorporate the reputational ranking of voting members, not simply their token holdings. This will give a greater voice to higher quality participants who may lack a substantial holding of tokens, while reducing the wealth of large token holders whose contributions to the network are less meaningful.
Relevant currently has a desktop and mobile application in internal beta, with a test community running already. Balasanov and his team are busy building a centralized version of the reputation protocol. Relevant plans to launch a new blockchain-focused community soon. The team aims to have an alpha version of the platform in six months.
After the presentation was completed, the floor was opened to questions from the Crypto NYC community. Many of the community’s questions focused on the structure of Relevant’s internal token economy, and on how effective Relevant’s token-based system would be at incentivizing participation in the system. Several attendees requested clarification on how the multiple layers of the Relevant infrastructure would interact.
Crypto NYC hosts a weekly round table lunch at our headquarters in TriBeCa. To sign up to participate in our lunch program, please sign up via our website for additional information! On Tuesday June 26th we will be hosting Tarrence van As of Common Interest. You can register to attend here.
