Is the Future Open?
Given the amazing success of OpenStreetMap in creating a free map of the world, you’d think it is a foregone conclusion that “open” is the future of mapping. Yet, the vast majority of the emerging augmented reality (AR) and autonomy maps are proprietary. In the AR space the proprietary aspect is further compounded by the dominance of massively vertically integrated players like Apple, Google, Facebook, Microsoft, Snap and Niantic. With the exception of Facebook’s acquisition/integration of Mapillary the competing AR maps are all built on proprietary data. The question is will the future stay proprietary or will we see a similar trajectory to 2D maps where “open” wins the day?
Haven’t We Been Here Before
Looking back at the history of online mapping, twelve or so years ago, the big tech players were all building their own web mapping applications and fueling them with third party private data — all sounds similar to AR mapping today. Some of the company names have changed, but the corporate strategies are all familiar. Fast forward to today and from our big tech “gang of six” only Google doesn’t currently use OpenStreetMap extensively for web mapping. Does this mean history will repeat itself for AR? Is “open” a manifest destiny? Not necessarily.
While we often call the data to power augmented reality 3D worlds “maps”, they are considerably different from traditional web mapping and cartography. OpenStreetMap originally was powered by GPS tracks and later by tiles and tracing of satellite imagery. While putting these elements together into a cohesive technology and community was and is a massive effort it did have the benefit of being relatively lightweight. The OpenStreetMap project ran on a server or two in an UCL closet for several years. The video to map Pixel8's small, two block, test area in Boulder is several gigabytes of imagery and generates 11,159,680 data points. Unfortunately, the scale of compute and storage for 3D worlds is on another level and so is the cost.
The “gang of six” not only have the concentration of compute resources, but also hundreds of engineers to solve the challenging photogrammetry, localization, devops and myriad other technical challenges that arise in the creation of 3D worlds. Compute and talent both have considerable overhead costs associated with them, which make creating an open alternative challenging.
AR-Insider has an excellent series of posts on “Making the Physical World Clickable”. In the articles they also discuss the challenges for third party developers competing in AR with the big players. Specifically, Matt Miesnieks makes the case that the big incumbent players have an early powerful strategic advantage because of their ability to vertically integrate hardware (glasses+phones) and software. We can also add 3D “map” data to that vertical stack and each player we’ve discussed is investing heavily to do so. Matt, also points out that
“The only prevailing counter force that blows in favor of the web is that the real world is really big and really messy,” he said. “Even a $2 trillion market cap company (Apple) isn’t big enough to manage everything that’s going on in the real world. We’re going to need some way for this mess to be constantly evolving and adapting, and the web is the way to do that. So I’m kind of curious to see how those two forces play out when it comes to AR.”
In Matt’s assessment my bias reads “the web” as open crowdsourcing outside of the walled gardens provided the big AR providers. While I feel certain that “crowdsourcing” will definitely play a key role; whether that is open or closed remains to be seen. Today I’d lean towards that crowdsourcing being proprietary. Niantic is already using their Ingress users to crowdsource the scanning of “points of interest” (POIs) for Pokemon Go. Facebook, Snap, Apple and Google all have huge communities of users collecting GPS tagged photos and videos — the raw photogrammetry ingredients for creating 3D worlds. They all have the components for leveraging their own proprietary crowds to make 3D worlds. So, how can a complimentary “open” community flourish?
Building an Open Future
While Pixel8 has been building tooling and infrastructure to an enable an “open” future we are just a small piece behind the scene. They key is community. While the big tech players all have crowds of users they can tap — their participation isn’t purposeful or directive. To steal from Andrew Crooks it is ambient crowdsourcing a.k.a. the data exhaust of users doing a totally different task not intended for mapping. Their user aren’t crafting a 3D world together as a community. In my opinion this is what makes OpenStreetMap a massive success. The community’s purpose is the world being built and they benefit from it.
We think a very similar opportunity is waiting for the next generation of maps. A shared 3D world we craft and curate together. Potentially an even bigger community of gamers, photographers, videographers, VFX-ers and mappers. We don’t think the current proprietary approaches are bad — quite the opposite. They are a logical path to follow and there aren’t fully baked alternatives. That said if the community emerges there is a wonderful opportunity to create another symbiotic relationship that further evolves the structure for collaborative mapping. Our team still has work to do, but hope to add contributions to the cause this Fall. Looking forward to the feedback and seeing new geographies mapped in 3D.