Tom MacWright

tom@macwright.com

Hivemapper impressions

Hivemapper is a company that recently pivoted from drone mapping to dashcam mapping, and has added a crypto element as a way of compensating people who contribute imagery. They teased a dashcam hardware in 2020, and announced the crypto angle in 2022.

The idea of individuals contributing streetview-like imagery with a dashcam is not new. Mapillary did exactly that starting in 2013, established a large dataset and active community in a few years, and sold to Facebook in 2020.

What follows is a quick review of what is currently available about Hivemapper’s new phase:

What is a “map” in this case?

A map is streetview-like imagery. The type and breadth of data collected is really similar to what Mapillary collected: you’re getting front-facing camera imagery with GPS. Sure, you can collect a few other things, maybe, but at least initially, you aren’t. These images are very useful for helping people map OpenStreetMap, possibly useful for companies working on machine-learning tech for self-driving cars, and useful for fun niche projects like analyzing tree cover in cities and whatever.

What it isn’t building is “a map” in most senses one would understand. Unfortunately, TechCrunch and other sources just breathlessly say that, sure, this is competing with Google Maps.

Maybe in the far, far future they will be, but it’s technically impossible that Hivemapper provides directions, geocoding, or street maps based on this type of imagery within any rational timeframe. Mapillary’s example is instructive here: they have a top-tier machine learning & image recognition team. That field has advanced in the last decade, but reconstructing street maps from street view is still nearly impossible, and reconstructing a geocoding or directions database is, in my opinion, strictly impossible. You just can’t get accurate addresses from street imagery alone.

“A map that belongs to the people who build it”

This is on the homepage, and the meaning of this statement is still TBD. It’s referred to as “community-owned.” According to Ariel, the license will be chosen by the Hivemapper Foundation, and according to their how it works,

No Hivemapper Inc. executives will hold positions in the Hivemapper Foundation entity.

The devil’s in the details here. There’s basically three levels here:

  1. First, raw data that people contribute, which will be images and gps coordinates. Unprocessed, unstructured, and not useful to the vast majority of people. This will probably be open source via some license, TBD.
  2. Processed data. There is a lot of secret sauce turning that raw data into anything else, whether that’s streetview-like spinny globes or extracted ML stuff. To process the data you need expensive employees and server time and other capital costs.
  3. Services providing processed data. At scale, you’re going to be providing lots and lots of data transfer, very fast, with server uptime requirements - all of the stuff that mapping companies have to do. Access to services, which requires someone to pay for bandwidth and servers and engineers, is a very different than than ownership of the data underneath the service. OpenStreetMap, for example, is free data, but if you want to put OpenStreetMap maps or use a geocoder with OpenStreetMap data, you’re going to have to pay either a service provider like Mapbox or a web host like AWS for that service. The cost of underlying data has only a mild connection with the cost of services.

How much of this is going to “belong to the community”? I’m not sure, it’s pretty up in the air. There are some examples already, like OpenStreetMap, which is very explicitly owned by the community and purely run by a foundation. It’s hard to imagine Hivemapper being even more “owned by the community” than OpenStreetMap, which has a history of fearing and rejecting any perceived corporate influence.

Decentralized

Speaking of up in the air, we’ve also got “decentralization” in the marketing literature. I’m not sure how this shakes out, either. I could guess at a few ways things might be decentralized:

  1. People collecting the imagery are decentralized, as in they’re in different geographical areas and they aren’t employees. This is the same sense as Wikipedia, OpenStreetMap, Mapillary, etc contributors are decentralized.
  2. The raw map data might be stored in a decentralized way, like IPFS? There’s no mention of this in the how-it-works, and I expect this would make everything harder for no reason, but it could be the case. As I’ve written and revisited, IPFS is a dramatically overpromised technology that would be extremely difficult to use for anything.
  3. Will the map data be on the blockchain, in Solana? You could read that into the “how it works” but it’s never explicitly said, and I’d guess that that would, like IPFS, make everything harder and worse. I haven’t seen any blockchain-related project use an actual blockchain like a database: projects like Axie Infinity use sidechains and bridges, which aren’t decentralized, reliable, or anything, and constantly end up hacked.

Incentives

Mappers are paid in HONEY, a token on Solana. Solana is a Proof-of-Stake blockchain, which has a much better track record for environmental impact than Bitcoin or Ethereum, albeit in exchange for being notoriously unreliable because it isn’t decentralized or secure and keeps going down. But anyway, fair trade.

The division of tokens is heavily weighted toward the company and investors: only 40% will go to the contributors:

Of the 10 billion tokens, 20% will be allocated to investors who provide startup capital, 20% to employees of Hivemapper, Inc. who build the technical and operational systems required, 15% to Hivemapper, Inc. for providing ongoing security and technology, and 5% to the foundation that will be established to govern and facilitate the ongoing success of the network. The largest portion, 40%, will be allocated to token rewards for ongoing contributors to the map ecosystem as described in this document.

This may change in the long-term as people sell, but this is emblematic of the trend that crypto insider ownership is becoming the same as classic startups. The idea that crypto cut out centralized entities and intermediaries ended up being more a “vibe” than a real statement.

The economics themselves have some similarities to other “utility coins” like Helium and Filecoin - there’s quite a lot of explicit citation to Helium. Which - fine.

In short, what these are doing is trying to bootstrap supply by paying early participants by “minting” tokens. In the Helium and Filecoin cases, there’s a sort of unproductive “proof of coverage / storage” state that’s rewarded by the network and leads to tons of cheating in Helium’s case. It doesn’t seem like this problem exists in Hivemapper’s plans: you don’t get rewarded for just having a camera or being “on the network”, it seems.

This is sort of like early-stage startups like Uber, in which cab fares were subsidized with investor money. But it’s taken to the extreme - instead of a 50% subsidy, these networks are subsidizing the vast, vast majority of “rewards” that people receive, making the prospect of being an access point or storage provider or whatnot seem like a real gold-rush opportunity.

This sort of creates a weird two-stage economy in which “markets” have almost nothing to do with the initial phases, which are dominated almost entirely by speculation, and then you transition into a “normal” economy, theoretically.

None of these projects, as far as I know, have ever transitioned into stage 2, the one in which people pay for things and people providing those things get paid by the consumer. I have doubts about whether they can, because when it really comes down to it, being a small storage provider or network access point or streetview-collecting driver is not a highly-compensated position in the rest of the economy and there is no reason to think that it will be worth more (to buyers, who are paying you) in the crypto-economy.

But that is all supposition, I guess.

The team

There are a lot of crypto startups that are hyped-up crypto zealots who want to apply their secret sauce to some problem area that they don’t understand at all - see Toucan for example, an idea that must have been conceived and implemented while high.

Anyway, thankfully, Hivemapper’s team has actual geo people with actual experience in mapping and the constraints thereof. Dashcam mapping is different from drone mapping but there’s enough in common that they’ll probably be able to build something cool. I think this is actually a mapping company gesturing in the crypto direction rather than a crypto company gesturing at maps.


Summary

  • A lot of Hivemapper’s plan is up in the air - the license structure, the outputs, much of the technology.
  • It’s good that Hivemapper is a real company with actual geo cred. If they create something like Mapillary, that’ll be a positive for the world.
  • It’s great that they chose to build on a Proof-of-Stake chain, so that the environment impact may be lower.
  • A lot of what I’ve written is the result of Taking Marketing Seriously and Thinking That Words Have Meanings on websites, which is probably grating and annoying to crypto people who want vibes or people who are okay with vagueness overall, but my intention is to dig a bit deeper.

The role of crypto here is two things:

  • Doing payments, which probably entails some international regulatory arbitrage. You could do this in the traditional finance world with something like Stripe Connect, but sure, it’s arguably simpler, from a business and legal perspective, to do it with crypto.
  • Bootstrapping the economy by paying early contributors with money that you create yourself, and thus also making the $450 cost of a camera seem like less of a hurdle because you’ll expect to win it back. On the other hand, if the economics fail, you’ll have a lot of angry users who are out $450 and 100% of their motivation was self-interest/gambling/capitalism - see Helium for many examples of this situation.

Anyway, Hivemapper: it’s interesting. We’ll see how it goes once things get more concrete.

April 16, 2022  @tmcw