Pre-Proposal Essay: Rewards Pool Redesign

SUMMARY — This blog post is a pre-proposal essay on the future of Hivemapper minted rewards . We do not have a formal proposal at this time. We will reserve the MIP-11 designation for this topic, and open discussion on an ongoing basis to inform an eventual proposal. We welcome input on the best way to fairly reward Buzz, AI Trainers and Map Features through the minted rewards pool.

–

We had fewer decisions to make about the design of the rewards protocol when the Hivemapper Network launched in late 2022.

There were fewer ways to contribute and earn HONEY back then. You could collect images with a dashcam and upload them to the network. You could perform QA tasks on those images. The operator of the network (which is currently Hivemapper Inc.) would receive rewards for bearing the cost of processing, storing and serving images. HONEY would be burned when images were consumed, and reminted to the contributors who submitted those images. That was it.

The original rewards pool was designed for this simple state of affairs, with just four types of transactions. The largest was Map Coverage, which we will call Map Imagery for clarity here. The other three were QA and Map Editing; Operational Reward, and Consumption Reward.

The network outgrew this simple structure over the course of 2023 by adding several new categories of contributions that had been described in project documentation since before launch.

We added Bursts, a new type of Burn & Mint reward that pays a bounty to contributors who map specific areas of interest where customers are willing to burn HONEY as an incentive.

We launched an AI Trainer bounty, funded from our own treasury, to incentivize AI Trainer activity until we could determine the right long-term approach to funding AI Trainers through the weekly mint.

Most recently, we issued an airdrop to incentivize people contributing to Hivemapper directly through the Hivemapper app using Buzz.

Today, this is where we stand. In current state, at least 87% of the weekly minted rewards pool goes to contributors for submitting Map Images, while 10% goes to the administrator of the network (currently Hivemapper Inc.) to cover processing and storage costs and up to 3% goes to contributors for AI Trainer tasks.

Rewards pool, current

Both the AI Trainer bounty and the Buzz airdrop reflect our approach to cultivating network growth. We do not hesitate to fund new modes of contribution until they can be rewarded through the rewards pool. At a certain point, however, this transition needs to occur.

With the launch of a next-generation dashcam later this year, and the launch of smartphone-based contributions through Buzz, the time has come. We need to restructure the allocations of the reward pool to make sure that the rewards pool fairly rewards all of the work being done by contributors to build the map.

For example, the next-generation Hivemapper Bee will be able to extract and position Map Features on the edge — which the original Hivemapper Dashcam lacks the processing power or stereo cameras to do.

Many map data customers tell us they prefer to consume Map Features over Map Imagery, and our rewards pool should reflect the value being created by each mode of contribution.

We do not yet have a proposal for the new rewards pool allocation. We are publishing this post to be transparent with contributors who are deciding which dashcam to buy. We welcome an open discussion about how the rewards pool should be divided among network contributors, and whether additional changes are needed.

Please note that contributors currently receive 90% of the rewards pool in current state, and we do not expect this to decrease. Also note that the transition to a new rewards pool allocation would take place gradually, and would not occur until Hivemapper Bee fulfillment is well underway.

Here are some guiding questions:

1. What factors should influence the rewards allocation between Imagery, Map Features, AI Trainers, Map Features and Buzz? How might we evaluate their relative value to the network?
2. Considering that the next-generation dashcam will generate more valuable map data than earlier iterations of dashcams, but will generate that data in a more passive way, how much of a premium in weekly rewards would you consider fair?
3. Global Map Progress is currently based on Coverage, Activity and Resilience metrics, which are solely based on Map Imagery contributions. How might we incorporate other modes (Buzz, Map Features) into Global Map Progress?