I’ve read a few stories over the last year about the need for a back up to GPS. The New Yorker has highlighted the vulnerabilities of the current GPS system. Scientific American wrote about the lack of a ground based back up system, like eLoran, if GPS fails. Since we spend a lot of our day at Pixel8 working on global localization problems my mind often wanders to the potential for “visual positioning systems” (VPS) as a compliment to GPS. The part I always get stuck on with this idea is the fact that GPS is a public good. Alternatively, the global 3D map for localization is at the opposite end of the economic spectrum — a private good. It might be worth while to take a quick tour through economic goods in the context of an augmented and autonomous future. Are there potentially better economic constructs for a 3D map of the globe than our collective current trajectory?
One of the best classes I took in grad school was a macro class in the Department of Food and Resource Economics. The course was a dive into public goods, externalities and failures — both market and government. The concept of public goods was a fascinating one. A good could be non-excludable and non-rivalrous. In short this means you can’t restrict access to the good and one person enjoying the good does not limit its availability to another person. A classic example is a sunset.
You can’t practically exclude me from enjoying the sunset, and also my enjoyment does not decrease your enjoyment of the sunset. The opposite of a public good is a private good, which is excludable and rivalrous. For instance an apple can have access to it restricted and when you take a bite there is less apple now for me to enjoy.
The reason I start with public goods is because GPS is a rare example of a human produced pure public good. Access to GPS is open to everyone and my use of GPS doesn’t diminish your use of GPS. The beauty of a public good is that they generate positive externalities. You can think of positive externalities as the indirect benefits of a “good” or in this case a “technology”. GPS was developed for military purposes, but when it was opened to the public it created additional benefits for everything from in-car navigation to playing PokemonGo to ranking your bike ride on Strava. RTI estimated that GPS has generated $1.4 trillion in economic value since it was made available to the privates sector in the 1980’s. That is pretty damn positive.
What Kind of Good is Global Localization/3D/VPS
Currently, the efforts by at least a half dozen companies (Google, Apple, Facebook, Microsoft, Amazon, Snap and Niantic) to create a 3D map of the world are squarely private goods. Each firm is creating their own 3D maps of the world, which are private property and excludable. By definition of there being at least six worlds there are rivalrous elements. On the continuum of economic goods we are currently on the opposite end from a public good like GPS. Yet, there are so many parallels between VPS and GPS in both purpose and structure. Like many digital goods the cost of replication for a 3D feature database is low and perpetually getting lower. So, it is difficult for one user to decrease the value of the good for another user. This means it is arguable that the excludable gates placed around a 3D map of the globe are artificial. While this is the case for many digital goods it opens up some interesting opportunities.
Before visiting these we should probably put to bed the hope that a 3D map of the globe can be a pure public good. In spite of my click bait title the private market has already invested billions in creating a 3D map of the world. This makes the possibility of it being a pure public good slim to none. This is in direct contrast to GPS where the government had paid the full freight for developing GPS before there was ever a private market application. This left the decision to structure GPS as a pure public good up to the government. There were certainly other options — like auctions used for spectrum. The private market players who’ve already made investments certainly aren’t going to place those in the public domain. Also, it is very unlikely the government is going to fund a competing effort to create a duplicate map for the public. As such we can reasonably rule out a public good future for the 3D map of the globe.
Could Club Goods be the Future
Fortunately, we aren’t limited to just private and public goods. There are also common goods and club goods. Common goods are non-excludable goods that are still rivalrous like forests or fisheries. They are open but exhaustable, so not a good fit. Club goods on the other hand are closed but inexhaustible. Classic examples are cable television, cinema and copyrighted works. Club goods are often associated with the concept of artificial scarcity since they have zero marginal costs once built. Like GPS or cable TV, a 3D map of the globe certainly has a big infrastructure investment to make, but once up and running additional incremental user cost is relatively negligible.
It is this upfront cost of building and maintaining a perpetually accurate 3D map of the globe where there could be efficiencies to gain. Does it make sense for six plus companies to make duplicate maps of the globe and perpetually maintain them? For the augmented reality use case the updating requirements are far more frequent than traditional 3D mapping. Looking even further ahead to autonomy the burden only increases.
Given the billions of dollars each firm will need to continually invest it seems forming a “club” could be a reasonable path forward. While the augmented reality firms we’ve mentioned (Google, Apple, Facebook etc.) are certainly rivals from different perspectives, the 3D feature databases that power them are not meaningful competitive differentiators. You can think of the 3D map of the world as table stakes and the biggest differentiator is cost. The best way to compete with cost is through a shared map that minimizes investment from all parties. There is also no reason to think that government could not also be a part of this “club” since they’ll have their own need for a 3D map of the globe and importantly a backup to GPS.
In many ways this “club” concept has already been proven. OpenStreetMap (OSM) operates tacitly as an unfunded club when it come to the big tech players we’ve discussed. With the exception of Google they’ve all paid mappers to contribute to OSM or leveraged the collective work for their commercial offerings. The problem is this approach has been fraught with political strife within the OSM community and also perpetuates the project’s chronic under funding. This all creates ongoing challenges that could be avoided in parallel projects.
How do You Start a Club?
I immediately think of dimly lit backroom deals involving cigars and big tech companies. The reality is likely far more pedestrian. Even with the success of OSM creating a 2D base map — a shared 3D feature database of the world isn’t going to happen overnight. We can start with a shared language. There is no standard for a global feature database that would make this data interoperable. Creating a standard would be an excellent place to start. This is not my observation or idea, and has been discussed independent of this post/concept. A standard would bring interested parties together and start to establish trust and a common cause.
Next, starting to open more data in alignment with the standard allows interoperability to begin being tested. This allows real world benefit to be demonstrated, which in turn makes the case for further investment. Investment is what makes the “club” viable and long term sustainable. I believe the “club” construct also makes collaboration with government far simpler. The potential for leveraging data collected by public sensors is vast. Instead of each firm trying to strike partnerships at the local, state, national and international level there is a single entity to collaborate with. Many data products and derivatives could be open. Also, all the challenges for setting privacy and ethics standards are inherently more manageable in a “club” construct. In short the positive externalities make a super compelling case. Taking small steps starting with standards could start the ball rolling. Whether a “club good” emerges is debatable, but standards and collaboration by groups building next generations 3D maps of globes certainly seems like a prudent path to explore.