A Maxar satellite image of Kennedy Space Center from a recent tweet. Courtesy of Maxar
Joe Morrison is perhaps best known for his Substack, where he covers the world of satellite imagery and mapping with a thoughtful and fact-based — but altogether surprising — point of view. He’s the kind of commentator that once argued (convincingly) that Google Maps’ advantage was fading. And that’s not the only insight he’s offered that flew in the face of common wisdom.
When I caught up with Morrison earlier this month, he brought his critical eye to our discussion of spatial computing. Among other topics, we talked about why you need satellite imagery to build virtual worlds, how video games are way ahead of the curve on 3D tech, and why taking it slow is the right way to go when you jump into a new technology like spatial computing. After all, says Morrison, there’s no gold medal for being first.
Sean Higgins: Does the term spatial computing mean anything to you? If so, what?
Joe Morrison: My limited experience with “spatial computing” comes mainly out of the geospatial field. I was working for a consultant for a long time called Azavea, which builds a lot of open-source software for large-scale geospatial processing.
And the majority of that was 2D data — mapping data. But some of it, like the GeoTrellis project, was 2.5D data: elevation models, or pixels with a Z-axis. So I was exposed to spatial computing a little bit that way.
When you think about spatial computing in the context of mapping, do you ever think about 3D data?
I’ve written before about how video games are a lens into what modern spatial computing looks like, and how mapping tends to lag video games pretty dramatically. You can see what’s coming to spatial computing by just looking at video games.
There are so many ways in which this plays out in the history of web mapping in particular. For instance, the folks who created Google Earth — are you familiar with their story?
They owned a map company called Keyhole, right?
Yeah, but before that, they had a video game company that didn’t work out. And Keyhole was built on the stuff they were able to salvage from it.
You can also look at like Mapbox [a competitor of Google Maps] and their integration with Unity as one way in which these two fields are starting to merge. Another place where it starts to blend is Snapchat.
Snap’s mapping team is very impressive, and a lot of their work focuses on augmented reality. They acquired a company called Pixel8Earth, which was using phone cameras to cross-correlate with satellite imagery with 3D photogrammetric scans — from the ground — and put them in the correct place on Earth. So scans had a mapping context to them.
And that’s infrastructure for Snap. If Snap wants to do a Times Square event where everybody has their glasses on, and dragons fly over or the billboards all change, they need to know where that all happens spatially. That’s a geospatial problem.
People tend to focus on spatial computing on a much smaller scale.
I think a lot of times when people talk about spatial computing, they’re talking about 3D-constructed projections like those typically used in video games and virtual reality. But when I hear “spatial computing,” I think of geo-spatial computing. And it’s interesting when those two things intersect like they do at the Mapbox Unity SDK or Snapchat.
You make a convincing case about spatial computing being a geospatial problem. If we want to have a holistic understanding of the real challenges and possibilities of spatial computing, we need to think on a bigger scale.
Yeah, absolutely. And the world is mapped in 2D for the most part — definitely the foundational elements of the world. A lot of those maps are open now: Facebook, Google, Apple. I know that Facebook released a large data set of building footprints. People like Synergize have released data for all of the bodies of water in the world. These are the foundational features that don’t move a whole lot.
The land is covered decently well, at a reasonable resolution globally. But the best elevation models for the globe tend to hover around 30 meters to down to 10 meters for each grid cell. And that’s terrible if you’re trying to do a local-level analysis.
So we still have more work to do to get that geospatial computing data we need?
There are projects like the map of the world owned by Maxar. And the world DEM [digital elevation map] that’s owned by Airbus. One of those was derived from stereoscopic images, and arbitrarily overlapping optical images where you could do photogrammetry and arrive at an elevation model. That’s Maxar’s technique, which they built in partnership with Saab.
And then Airbus used this synthetic aperture radar mission called TanDEM-X, which I think is the most underrated mission in the history of remote sensing. It was an incredible feat of engineering: They took a pair of very expensive synthetic aperture radar satellites and flew them in extreme proximity, on the order of hundreds of meters apart.
When you’re traveling 16,000 miles an hour, to be a couple of football fields apart and not run into each other, and be tracked precisely? That’s crazy. And then they flew them in a helix pattern, which meant they were corkscrewing around each other in the sky, which is a thing you do to avoid a collision.
So, anyway, I say that because they used this mission to produce a five-meter elevation model for the whole globe. But it’s proprietary. It’s very expensive to get access to that type of data, even though DEM at five meters and below unlocks all kinds of use cases commercially.
Estimating how deep rivers will be to more accurately predict the amount of water traveling through them. When you know their width and elevation, you can guess their depth, and then if you know the width and the approximate depth, you know how much volume is moving.
Then you predict accurately how much electricity hydropower dams downstream are going to generate. You can do all kinds of interesting analyses. You can perform much more accurate flood-inundation modeling during storm events, and all these things that have relevance to insurance, agriculture, and catastrophe response.
So the potential of geospatial computing is much bigger than just helping companies like Snap locate their 3D models in space.
Yeah, a lot of the focus in spatial computing is on technology like lidar on iPhones, which is wild. To be honest, that is crazy to me, and I think it’s super underrated. But what I’m interested in, and what I focused my career on, is the global scale.
It’s the question of, how are you going to get information in a situation where you can’t bring an iPhone, or there is no internet connectivity, or you’re not allowed to fly a plane with lidar? I’m interested in spaceborne methods because they’re also immune to a lot of those geopolitical boundaries. Assuming nobody shoots them down.
A lot of spatial computing technologies at a smaller scale seem to have a lot of hype but no clear application. But spatial computing at this geographical scale seems to have a very clear set of applications.
And the key thing there with all this, as you mentioned when we were talking earlier, it’s not hype. Synthetic aperture radar was invented in the ‘50s. The TandDEM-X mission was conceptualized and funded in 2004. This is not new. It’s not like any of this stuff is newfangled. It’s just sitting there waiting to be applied.
And that’s a great shame because there’s a ton of value to be unlocked. So, when I hear about the Metaverse and 3D and all that, I’m excited by it, because I feel like it’s not new technology. It’s just the investment and marketing and education are required to apply it in a more broad, commercial context.
The people that are doing the best work are often not doing anything novel. They’re simply making a very expensive thing cheap. They’re not inventing lidar. They’re making a very expensive lidar unit cheap enough to put on every car.
So that’s where I tend to focus, not on the cutting edge but 20, 30, 40 years ago. The physics has been proven. Now, it’s just a logistical question of how to make this cheap, affordable, and accessible.
Is the journey from vision to reality just a matter of figuring out the mundane details? Maybe you could answer this in the context of commoditizing satellite data, which you’ve written about in your Substack.
It’s a lot more complicated. I don’t think there’s any guarantee that it’s going to happen, that satellite imagery will ever become accessible to a broad audience. I think the arrow of progress points in that direction, but I don’t think you can take progress for granted.
I don’t claim to understand the many interconnected forces that create the complex system that will result in whatever changes happen over the next decade.
Even zooming all the way out from these things, it’s a tremendous capital expenditure to build satellites. Even though people always talk about how launch costs are coming way down. But a one and a half meter lens in a cleanroom didn’t get any cheaper.
There’s Maxar bragging about how each of the Legion satellites only cost $100 million. The physics for the optics still requires really big lenses, and so it’s always going to be a huge capital expenditure. And SAR has its own totally different set of physical limitations that have to be accounted for.
The point I’m making is that we’ve been in an exuberant bull market for a long time that resulted in a lot of venture dollars pouring into the space sector in particular, and into the Earth observation niche as well. Companies going public and raising hundreds of millions — we’re talking billions of dollars in aggregate.
And some people look at that and say very cynically, “That’s all hype and over-promise and under-deliver.” But to me, that’s a regular technology cycle.
We’ve seen this happen with technologies like the internet. There was a boom in the ‘90s, things fizzled out, then picked back up again with intensity.
I’m getting internet installed at this cabin that I’m in today. Gigabit speed fiber cable. But the companies that created that infrastructure? A lot of them didn’t make it. And it was exuberance that funded it. Undeniably, it created a tremendous amount of value.
And I think Earth observation is going through a similar phase where that window is starting to close. It feels like the capital is there to build the infrastructure, and really great companies could become huge infrastructure players, like Verizon.
So, I’ll take the hype. Sometimes it’s annoying, and sometimes it’s frustrating to see the sponsoring finance guys. For the satellite boom, the stuff they would go on TV and say was unbelievable. It was lies. They were lying and it wasn’t clear to me that they knew that they were lying, but someone with any sort of intellectual rigor would realize that they were lying.
But by the same token, they were able to raise money that will be used for good purposes. Some of them will probably get sued to high heaven for representations that they made. And some of them will get acquired and whatever, but the point is that the financial aspect has to be there.
And then on top of that, there have to be enough people that survive and the forces of competition have to be such that it does get cheaper and more accessible.
Do you have any other takeaways from your time in the industry? Anything you’ve learned that might help people understand what to do when faced with a new technology?
The thing that I’ve learned over the last 18 months since joining a satellite company is that this is a small world, there are like 250 people that control the entire Earth observation industry. For every important decision that’s ever made, you could fit all the people in a small conference room. And it’s people that make the decisions.
You have to get to those people and convince them that it’s in their interest to grow the market. But people are very unpredictable. Anyone that says oh, definitely, change is coming? Unless it’s directly within their control, they have no idea. It’s not a guarantee.
How do you recommend people navigate these markets, then? For instance, if I wanted to start a company in spatial computing, or even adopt some new spatial computing technologies at my existing company.
If you’re an entrepreneur, for instance, and thinking about starting a business in satellite imagery, you’re wondering about this right now, today. You’re asking, “Is it going to get cheaper? My whole business model depends on understanding where things are moving.”
And the answer is pretty simple: Just be more conservative than you think you should be.
It’s a lesson that has played out in every technology forum ever. The first mover? That’s fake, that’s not a real advantage. There is no such thing as a first-mover advantage. Facebook is not the first social media platform, and Lyft is not the first ride-sharing platform, but they did alright for themselves. Just because somebody’s doing it doesn’t mean you can’t do it.
Versus something that you can’t find a single example of anyone doing currently. Or something no one has tried before. My view is perhaps cynical, but my experience dictates that it’s not a good idea. Most really great ideas, someone’s tried them. They may have failed, the timing may have been wrong, and you may have better timing.
But if no one’s even trying, and you don’t understand that market deeply, and it’s not where you come from? It’s a terrible idea. I can tell you right now that it’s a bad idea.
If it’s your problem that you’re struggling with and no one’s ever tried it, then you have a shot. But even then, instead of starting that business, try to find somebody to build it for you or try to find somebody who’s doing it and hire them. And then if they’re terrible, it’s time to start looking.
Everybody puts so much emphasis on innovation, but it sounds like hanging back and watching what other businesses do is a great learning tool.
Yeah, and the bigger the technical challenge, the better it is to have that extra information. There’s no gold medal for being first.
What about people who are afraid to be late to the game?
It’s often a lot later than you would expect. If you’re paying close attention to a sector because you understand it really well, you’re living way farther in the future than you realize. You just have to catch the wave that’s happening in the present.