I am happy to say that the initial goal of creating and automating a quality-based structure on top of the Steem blockchain has been achieved. The q-filter voting bot has been been up and running successfully for two weeks now and we are now ready to showcase our rewards distribution system that incentives users to distribute their votes and make decisions on quality rather than simply maximize rewards.
Image Source
Q-Filter Voting Bot
Most voting bots require users to either pay for votes or bid against other users for a portion of a vote. Other bots simply reward users for using their service with little regards to the validity or quality of the content provided by the users that use the platform. The problem is that the current generation of bots are not robust enough to address these concerns and are focused on generating revenue for their owners.
Q-Filter is different in that it uses information from voters in order to estimate the general feeling towards a post and then by using stake-based weighted averages the q-filter bot gives a partial upvote based on user consensus. This also allows q-filter to upvote good and quality content and distinguish somewhat between the two and flag bad content granted that there is enough bad sentiment against that content.
This is achieved by utilizing the full range of voting from a full flag (-100%) to a full upvote(100%). However, one of the issues by utilizing this type of range voting is that many users have little incentive to do so and often most users leave a full upvote or a very small upvote independent of their own quality perception. Thus, in order to incentivize voting in a more nuanced way, the planned distribution mechanism was shifted away from content creation and curation to solely voting and curation.
Incentivized Voting
To incentivize voting, a voting lottery has been designed to award SBD to one lucky voter in the voting pool. This incentivizes a more horizontal voting distribution as there will be competition for each piece of content. By having horizontal distribution, the creators are rewarded more effectively and thus do not need to be additionally compensated.
However, a completely random voting lottery would not encourage nuanced voting. Thus, the goal of the voter will be to estimate the final stake-based weighted average with their vote. However, this begs the question: why not simply wait and then vote according to average? We avoid this potential issue by adding a component of randomness to one's vote. Earlier voters will have less randomness added but less knowledge of the final average while later voters will have more randomness but more knowledge of the final average.
So we have a fair system that encourages users to hunt for that average but disincentivize voters using the system to vote at full strength on content that already has a lot of votes. But this system is still experimental and in development and does not address many possible concerns. There are still ways that this could be abused but with continuous development and brainstorming, we can discover mechanisms that achieve desirable results.
How to Participate
Creators
Content creators simply need to use the 'q-filter' tag. After 48 hours, the q-filter bot will vote on that content based on its weighted average.
Curators
Voters simply need to vote on content under the 'q-filter' tag. Shortly after the post delivers its rewards, a vote lottery will occur and SBD will be delivered to one voter outside of that post. The q-filter bot is ineligible for this award. Partial voting and flagging are encouraged.
The SBD delivered is the total SBD of this account divided by the number of posts in the pipeline. In order to maintain funding for this project, more regular posts will be written to help to fund the lottery fund. At a minimum, as long as this project stays small, there should be little competition for those rewards.
Future posts will go into some of the details of development and talk about some of the algorithms and ideas in detail. At the moment the voting bot and voting are open and currently up and running. Have fun.