Map Improvement Proposal 6 (MIP-6)

9/1 UPDATE: After an open and productive discussion with many members of the community, we have decided to finalize the MIP-6 proposal. It will take effect with the reward cycle that is processed on Sept. 6, and will apply retroactively to the week running from Monday, Aug. 28 to Sunday, Sept. 3.

We will be carefully monitoring the upcoming reward cycles to ensure these changes achieve the purpose of making Region Progress simpler, fairer, and better aligned with our mission of scaling map coverage.

Our thanks to everyone who helped to shape this decision.

—

Hivemapper decided to use crypto rewards because they create incredibly powerful incentives that align the economic interests of contributors and the network. With the right incentives, everyone is on the same team building a product that will reach scale and generate value.

In March, we implemented MIP-2 to fix problems with region boundaries and region weights that were undermining this incentive structure. With MIP-6, we are now proposing changes to the Region Progress formula. We believe these changes would make the incentive structure simpler, fairer, and better aligned with our mission of scaling map coverage.

Under this package of changes, some regions would see their Region Progress increase while others would see it decrease. Our analysis indicates that Global Map Progress would increase by about 5% under this proposal, resulting in a corresponding 5% increase in the global rewards pool, all else equal. As always, we are focused on doing what is best for contributors in aggregate, rather than in any specific region.

Here’s a summary of what would change under the proposal.

• Coverage would be stricter, to better distinguish the regions making the most progress. To maintain 100% Coverage in a region, contributors would need to refresh 15% of unique km every 28 days, compared to the current rate of 10% every 90 days.
• Activity would be stricter, with a new formula that fixes a bias against small regions. To maintain 100% Activity in a region, contributors would need to submit at least 30% as many total km every week as there are road km in the region.
• Resilience would be easier, scaling up as contributors join a region. To maintain 100% Resilience, no contributor can be responsible for more than one-fifth (20%) of total km submitted over the prior 28 days.
• Region Progress would be calculated based on a weighting of 60% Coverage, 30% Activity, and 10% Resilience, rather than an equal weighting of all three metrics. As an example, if Region X had a Coverage of 70%, Activity of 80% and Resilience of 50% after this change, Region Progress would be (0.7*0.6) + (0.8*0.3) + (0.5*0.1) = 0.71 or 71%. This reflects that increasing the number of unique km regularly refreshed by the network (Coverage) is more important than total km (Activity) or the number of contributors (Resilience) at this stage.

The problem we’re solving

There are still big differences in HONEY rewards from region to region that do not reflect the amount of progress made toward building a valuable map. In other words, the current Region Progress formula could be a better reflection of actual region progress.

New regions make progress very quickly. One or two contributors can easily drive a region to 20% or 30% Region Progress without deep or sustainable coverage.

In contrast, more established regions make progress at a slower pace. Eventually, their progress might plateau, primarily due to the challenge of achieving Resilience. A significant number of regions with 100% Coverage and Activity are struggling to reach 20% Resilience — as a result, there is a cluster of regions with Region Progress in the 60–70% range that are seeing diminishing rewards as more contributors join the region.

On closer examination, we discovered problems with all three components of Region Progress, which are Coverage, Activity, and Resilience.

Coverage

With the current formula, Coverage does not give a sufficient reward to regions that make the most progress.

More than 15% of regions maxed out Coverage in recent weeks, by crossing the threshold of mapping 10% of unique roads within the last 90 days. This threshold is too low, and does not fairly reward the regions making the most progress relative to other regions that are making a bit less progress.

At some point in the future, 15% will likely also be too low. We expect the Coverage threshold to be steadily increased over time to continually reward the regions that continue to make progress.

With this MIP, we propose two changes to the current formula:

• Increasing the threshold from 10% to 15%.
• Reducing the time horizon from 90 days to 28 days to better align with the freshness that map customers expect.

The result:

• The number of regions achieving a 100% score would be cut in half.
• On a global basis, the weighted average of Coverage Progress would drop by 14 percentage points.

Activity

With the current formula, Activity is biased toward larger regions.

• As currently defined, a region reaches 100% Active Progress when a sufficient number of contributors drive at least 100 km in the region. That is easy in a larger region, but difficult in a smaller region.
• For example, consider the small, dense regions of central London, such as the City of Westminster, home of Buckingham Palace. Most roads have been mapped at least once. Every week, more kilometers of map data are submitted than there are kilometers of roads. This seems like a perfect example of a highly active region. And yet, the Activity Progress in Westminster has ranged from 20% to 40%.

We propose a new formula based on the number of total km submitted relative to the number of road km in the region, not per-user activity. If contributors submit at least 30% as many total km every week as there are road km in the region, the region gets a 100% score. If contributors submit less than 30% of that amount, Activity scales down to 0.

The result:

• Some smaller regions would see their Activity score increase.
• Some larger regions would see their Activity score decrease.
• On a global basis, Activity would drop by 13 percentage points.

Resilience

In the current Region Progress formula, Resilience is far too difficult to drive up because it requires a very large share of hexes in a region to be mapped by multiple contributors within the same time period.

• The weighted average Resilience across all regions is just 5%, which is a significant drag on overall Region Progress.
• For example: Seoul has 100+ weekly active contributors. If one of these contributors were to move to another region, it would hardly impact the level of map coverage. Despite this, Seoul has been unable to surpass 25% Resilience.

We propose a new formula that is based on the vulnerability of a region to losing any individual contributor.

• If no contributor was responsible for more than one-fifth (20%) of the region’s submissions over the previous 28 days, the region will receive a 100% Resilience score.
• This means that Resilience is highly correlated with the number of contributors in a region. It is mathematically impossible to reach 100% Resilience with fewer than 5 active contributors, and it gets easier and easier to maintain 100% Resilience as more contributors enter a region.

The result:

• Nearly all regions would see an increase in Resilience — especially larger regions. With the current formula, only one region has greater than 50%. With these changes, about a half-dozen regions would have 100% and nearly 100 regions would be above 50%.
• On a global basis, Resilience would increase by 40 percentage points.

Community Feedback

With these proposed changes in aggregate, Global Map Progress would be expected to increase by 2 to 3 percentage points, or about 5%.

The Hivemapper Foundation welcomes community feedback on the MIP-6 proposal. To provide feedback, please share any questions, suggestions, and comments in the #mip-6 channel of Hivemapper’s Discord Server.

The comment period will be 7 days, running from Friday, Aug. 25, 2023 until Friday, Sept. 1, 2023.