Hivemapper MIPalooza. 4 new MIPs...good or bad?

Hivemapper released 4 MIPs today:

MIP-21: bye-bye side cameras

They are turning off side-mounted cameras. Currently Bee is only allowed to be mounted in the front and your old HDC or HDC-S can go on the side. Once Hivemapper made that annoucement most people could see the writing on the wall that side cameras would be fully disabled.

This will likely have some detrimental effect on region progress scores, which also means rewards could drop for the front camera as well, just due to less activity in each region.

https://medium.com/hivemapper-foundation/map-improvement-proposal-21-mip-21-26be0ceb7458

MIP-22: no more AI trainer reward pool

Previously all of the annotation tasks were done manually by humans and around the start of this year Hivemapper started moving to fully automated AI reviews. Some more difficult tasks are still done by humans, however you generally can no longer get approved for an AI task review account as a new contributor.

To be clear, this MIP isn’t getting rid of AI trainers, but instead it is taking the 10% of the weekly rewards that get set aside for trainers and now setting it aside as an “airdrop” category, which is grouped together as the existing Foundation Reward category. The article alludes to this reward pool being used for testing new features and accompanying rewards.

All remaining human training tasks will get rewarded out of the existing operations fund.

https://medium.com/hivemapper-foundation/map-improvement-proposal-22-mip-22-15d74f318882

MIP-23: token burn for Bee’s object model data purchases

Currently the only token burn Hivemapper supports is for Scout map credits used to spot-buy street view imagery. For the existing Scout map credit purchases Hivemapper burns 75% and 25% gets reminted as rewards for drivers who contributed to the imagery purchased. However the burns have been pretty random with June 2025 seeing 1.225M tokens burned (around $24k USD).

This MIP is introducing the same burn and reward mechanism, but for the object model and classifier information the Bee generates. So this would be a customer buying data that details object type (stop signs, traffic lights, speed limit signs, construction zones, etc) and GPS locations. Pricing detailed as

• 20 Map Credits per 1,000 unique km

• At $0.075 per Map Credit → $1,500 per million unique km

I believe the concept is that this type of data is data that companies who have mapping products would rebuy frequently, such as once/week. Those companies would then integrate all of the latest object data into their mapping systems so that the most efficient travel routes could be updated.

The MIP also specifies that the developer that purchases this data is then able to resell it. So for example Mapbox would buy this data to update their maps and then sell their map API data to their own customers, etc.

https://medium.com/hivemapper-foundation/map-improvement-proposal-23-mip-23-5cd8509d832d

MIP-24: the long-awaited rewards change for drivers

This is the big one that has been debated in their Discord for a while. This MIP does three key things:

1. Specifies rewards are hex-travel based regardless of any images/object data being uploaded. This only affects the Bee.

2. Enable night time rewards for the Bee.

3. New “upload speed score” introduced for all rewards.

Item 1: currently rewards for Hivemapper are hex based, however the app for the old cameras reported image counts. Even if your camera captured 10 image in a single hex you ultimately got rewarded for the first image and all the other images essentially earned 0 rewards.

However for the Bee one of the main technical tasks the team implemented is that the Bee will check with their server first to determine if any data was desired for that hex, and if not it wouldn’t upload anything. When this happened the Bee didn’t receive any rewards for that hex. As a result the Bee earnings have been dismal and this has been the main complaint.

With this change the HDC and HDC-S will continue to earn as they do today, which is essentially hex based other than when these cameras lose GPS connection and don’t capture any images. However the Bee will now essentially get heartbeat rewards for any hex where no data is needed, this way the earnings of the Bee will be on-par with the older cameras, or better just due to the improved GPS performance on the Bee.

Item 2: the Bee has been mapping at night, as can be seen from data that’s being uploaded, but hadn’t not been rewarded for night mapping. Now with this MIP the Bee will continue to earn rewards at night for distance driven. The HDC and HDC-S will still be restricted to earning only during the published mapping hours.

This should help counteract MIP-21, as well as the recent decrease in region progress scores due to people not using their cameras. Night time mapping means the Bee will continue to contribute to the Coverage, Activity, and Resilience categories of the region progress score.

Item 3: the upload speed score. This had been a hot topic in their discord where Hivemapper originally indicated the LTE Bee model, using a SIM card, would earn more than the Wifi Bee model, because the LTE model with a SIM would upload data instantly and Hivemapper wants immediately fresh data. Many people complained that this didn’t make sense because either the LTE Bee could be operated in wifi mode and connected to a hotspot or cellphone for data uploads, without the need to purchase an LTE plan. Additionally the wifi model, which doesn’t support LTE, could still upload immediately via hotspot or direct phone connection.

As such Hivemapper has decided to apply this speed score to all cameras. Essentially the delay for uploading reduces your rewards, normalized to the time in a day, per second. If you via LTE or phone immediately then you essentially get full rewards. If you wait 12 hours to upload then your rewards are reduced by 50%.

It is important to note that this affects the HDC and HDC-S as well. For the HDC, which is Raspberry Pi based, the privacy processing time on an image is very long, especially when you’re actively mapping. During mapping hours and while driving the HDC is barely processing any data and it can take hours, even more than half a day, to get through all of the images stored on the camera. This essentially means when this MIP gets enabled that the HDC rewards are going to be cut by 25%-50%, depending on the driver’s behavior….and there’s nothing really that can be done since you can’t speed up the HDC processing time.

Also note that there are a couple conflicts for this speed score topic in the MIP. The MIP clearly states the scores are derated over a day, with it being possible to get a score of 0, however later in the MIP it says late images are still accepted (but not rewarded?) based on the existing cutoff times, which are weekly. So this raises the quesiton of how you can wait for more than a day to upload and still get rewarded. The question has been asked but we are waiting for a reply from Hivemapper.

https://medium.com/hivemapper-foundation/map-improvement-proposal-24-mip-24-1e8bdb131e8b

One final note, a couple of these MIPs did make reference to sunsetting the HDC and HDC-S. Although discontinuing the older cameras isn’t included in any of these MIPs it does seem likely that the next round of MIPs are going to finally kill them off and make the ecosystem entirely Bee-based.

So that’s the “quick” summary. How do people feel about the changes? Will it go well for HM? Or are they dead?