User and Staking Slot Tiers

All users are assigned a tier based on their level and the levels of the other users in the system. There are three tiers, and the levels corresponding to a tier are the lower, middle, and upper thirds of the range from level 1 to the max level user in the system. For example, if the highest level user in the system is a level 9, then users from level 1 to level 3 are LowerTier, users who are level 4 to level 6 are MiddleTier, and users who are level 7 to level 9 are UpperTier users. The level cap is currently set to level 15, so eventually, when a user reaches that level, the LowerTier will consist of users from level 1 to level 5, the MiddleTier will have users from level 6 to level 10, and users who have reached level 11 up to the max of 15 will be the UpperTier users.

The slots that are available for users to reserve are tiered as well, in the sense that only users whose tier is greater than or equal to the tier of the slot can reserve the slot. When data is requested, the slots available always come in triads. A triad is a trio consisting of a LowerTier, MiddleTier, and UpperTier slot. This is why the required lead must be divisible by three when the creator sets it. For example, consider the datasets in a request batch for which the creator configures a required lead to be 6. These datasets will all have 6 slots, which are composed of 2 triads, which means there are 2 LowerTier slots, 2 MiddleTier slots, and 2 UpperTier slots. If the data in a dataset takes too long to finalize, the MiddleTier and UpperTier slots will gradually decrease to become LowerTier slots.

There are several reasons for this design, but the two major considerations were accuracy of data and rewarding long-standing users who provide the backbone of the ODO system. In theory, the longer someone is providing and endorsing data for the system, the more likely they are to be reliable. This is tracked by their reputation score, and the idea is that by having (at least initially) more slots open to more experienced users there is a higher likelihood of getting the correct data as soon as possible. This also works to reward users who stick around because there will be twice as many spots available to MiddleTier users and thrice the number of spots for UpperTier users as there are for the LowerTier users, while not starving out LowerTier users from participating at all. The gradual lowering of the tier requirement for the slots is to try to move datasets along that are taking too long to get higher-tiered users to endorse.