Every curation begins with a KurateDAO database. There are 3 players:
The founder of the database (the “curator”) defines the purpose and rules of the database and writes them into bylaws. Like the rules of bitcoin, once the bylaws are set by a database they can’t be changed. This curator’s goal is to create the best database around to attract users.
Once the database is created, “scouts” gather rows of information and submit them to the database. Each row submission requires the scout to submit one crypto token as an ante. This prevents spam and sets up the crypto-economic game we describe next.
If another user of the database thinks a piece of content submitted by a scout should be excluded (because of quality, accuracy, relevance, etc. as it relates to the bylaws of the database), that user can stake two tokens saying the content should be removed.
The battle of the faction now begins! If another user wants the row to be kept in the database, they can stake twice the previous stake of two tokens, which would be four tokens, to keep the row in the database.
This staking process continues for a specified time, and the last side standing gets their way. Tokens are then distributed to the winning side. However, if the betting crosses a certain threshold, the voting will pause and the database curator makes the final call. The threshold prevents someone with a lot of money from getting “bad” content in the database. The threshold is set by the curator and the dynamics of the database. If the threshold is low the curator will have to do more work, if the threshold is high they’ll have to do less work.
The double or nothing staking process is a prediction market. Prediction markets allow for betting on the future outcome of a question. In this case, the question we ask the market: will the curator accept this row as part of the database?
Here’s an example of the curation process in action:
Curator Sandy creates a database dedicated to reviewing restaurants in LA. She wants to ensure the highest quality possible because she gets paid when people use the database. Users either pay explicitly to access the data or ads are run alongside the data and deposited in the treasury.
Once she launches it, reviews and restaurant locations start flooding in. Some come from customers, some from restaurant owners, and others from AI bots searching the web. How will Sandy figure out which information to keep? By incentivizing users to help her.
The countdown has started ticking, and users can vote to exclude it for a set period of time (for this example let’s say 30 minutes). During this time, if somebody votes to exclude the restaurant, then others can vote to re-include it.
When Squidward thinks an article on the database grossly exaggerates the quality of the paddies he just ate (perhaps the author was the restaurant owner), he will likewise vote to have the article removed. If Sandy actually just doesn’t know what quality sushi tastes like, more experienced connoisseurs will counter Squidward’s vote and be rewarded for defending the restaurant’s reputation. Scouts can look at the curator’s previous curatorial decisions to inform what to submit.
But, what happens when things are spiced up to the point that a lot of people have different opinions about the restaurant?
At this point, Sandy steps in to mediate the dispute. She will decide if the restaurant is or isn’t worthy to be on the list. Sandys attention is then only given to issues where his users are thoroughly on the fence. When her expertise isn’t needed for those issues, she can kick back and enjoy her database as it evolves into something more and more valuable. The strength of the system is that it widens the pool of people who can contribute to database curation.