I’m speaking at the GIS/SIG 25th Annual Spatial/Digital Mapping Conference on April 12, 2016, see you there!
It’s been a wild two-and-a-half years at Telenav helping bring OpenStreetMap to the consumer. We shipped consumer turn-by-turn navigation in the US with Scout which was for me a big first – a turning point of showing OSMs true potential.
As the OSM project at Telenav has grown the need for the visionary founder has shifted and I’m stepping back from full-time work at Telenav. I’ll be still helping part-time and helping with new projects going forward. 2016 is going to be a fun year with a great team at Telenav (including all the bright folks we brought in from Skobbler) and I know they’ll continue to push out more OSM goodness.
I have a new kickstarter live now: Every Road! Every Road is a unique poster print of every road leading away from your house (or any other point). Above you see the bay area, driving away from the Ferry Building at the NE of San Francisco. Here’s the same thing driving away from a point in the Sunset:
The roads get thinner as you go and lead to a tree-like structure. Each print is totally unique to you. Here’s driving away from Buckingham Palace in London:
Here’s walking everywhere from Wall Street in Manhattan. Notice most route on Manhattan end up walking North and branch out across each side:
Here’s the same thing, but driving:
Notice how it leads to a totally different map because driving leads to quicker routes along the edge of Manhattan and then driving inward to each point. As opposed to walking, where your maximum speed doesn’t change depending on what road you’re walking on.
The data of course comes from OpenStreetMap, more details are at the Every Road kickstarter!
This book contains 15 interviews conducted by OSM founder Steve Coast with the people who were there as the project began and grew. Starting in 2004, the interviews trace how a rag tag collection of volunteers was able to produce a map which compares in quality to maps produced by multi-billion dollar corporations. Learn how such an ambitious project got started and then succeeded at mapping the world, for free!
The book was the result of a kickstarter that raised just under $10k.
The Book of OSM is shipping very soon. Fifteen interviews with key people around OSM from when it started to today – you can sign up to be notified when it ships here. Here’s the physical review proof:
My kickstarter project for the book of OSM is nearly done. Kickstarter backers already have the PDF of the book: The contents are final and the cover has some bugs being fixed:
Notice the white lines on the back cover. Once that is done, it’ll be available on Amazon.com and Kindle. If you want to stay up to date, there is a mailing list on the website for the book.
Tesla are making maps, and why not? It used to cost billions of dollars to go collect the data, now the customers do it for free. The sensor packages in phones, let alone in the car, have put the power of geography in everyone’s hands.
We used to do this with volunteers in OpenStreetMap but now even that will be superseded. Armed with cameras, GPS, radar and sonar, a car can just capture all the data and (pretty much) make a map automatically, for free. It’s happening for two reasons:
- The costs are now so low. The old model of paying people to drive around expensive vehicles to capture data is too expensive and takes too long. Even if you did drive every road, every year, it still isn’t good enough for autonomous driving.
- The incentives are so high. If you were Tesla would you want a dependency in Google for maps? Really, would you want a dependency on anyone now that the cost is zero?
I was talking to someone recently who’s 18 or so. They literally thought “encyclopedia” meant “wikipedia”. Digging in their memory they remembered that there used to be sets of books (Britannica) too, but not anything else. That’s how maps are going to go, in a sense.
When proprietary map companies drive around expensive cars to collect data they had to pay for the car, the equipment and the people. Tesla has their customers to pay for all of that. They’ll have a fixed cost in building software to capture the data and then amortize that very quickly so the effective cost for them to build one of the best maps of the world is going to be… zero dollars.
StreetView cars are going to go away, it will just be every car taking photos all the time.
The number of maps is going to explode. It used to be that pretty much just two companies held map data for the US and EU; NavTeq and TeleAtlas. Now there are five: Google, Waze, OSM, TomTom and Nokia. Correction, now there are six with Tesla. We can argue over their relative qualities, OSM lacks geocoding, Tesla probably does too and so on. But it’s coming.
And those are just the public ones. There are a bunch of logistics companies and governments who have their own maps.
The list of people who might want their own map, and be in a position to build it, is large. Every car company, every decent-sized tech company, every location company (Garmin to Strava). The arms race is now on.
As the number of maps goes up, the value per-map is going down. That means that you should probably short anyone reliant on proprietary map revenue for their future. OpenStreetMap is pretty comfortable because the dollar cost of OSM is already zero so there aren’t any adjustments to be made, except perhaps that the incentives to help complete a free map of the world are going away, if everyone has maps and the value is dropping so precipitiously.
You could argue that price of maps won’t change since Google, Waze and Tesla will keep their maps to themselves. And yes, maybe. But it’s more likely not since they are removing themselves (and their dollars) from the customer pool. One of the new maps will be made available at some point too, they won’t all be locked up.
So the value that used to be in the map data will move up the stack, right?
Actually I’m not so sure. Charlie Munger has this nice story about textiles where some factory making fabric bought some machine that made them 20% cheaper, or whatever. They were proud of this. But of course, every other factory does the same thing and the value doesn’t go to the factories but to the consumer. The consumer wins not the factory.
The primary way map service companies got their valuations was waving around the Nokia acquisition as a bellwether for value. But Nokia actually owned something – map data. If you’re providing services then you need something else, like large long-term customer deals to provide map services to anchor you.
The cost for anyone to start doing basic map service work is… zero. So we should expect a bunch of competition in that sector. And of course, thanks to Google making maps free to the consumer, everyone thinks maps are free so the actual dollars you get from developers or whatever are pretty thin.
So at the top end, your customers are going to go make their own maps and at the bottom end, Google, Apple and so on are making maps free. It’s a tough spot, but luckily there’s probably a decade in there to figure it out.
Further Out, 10-30 years
The interesting implication is that creating a map will cost zero for Tesla, doesn’t mean it’s zero for everyone. It’s very beneficial to have a convenient fleet of sensors out on the roads. But that too is changing. Everyone and their brother is building self-driving cars.
But will there be a market for them?
It’s deeply interesting. A lot of the models seem to be predicated upon everyone buying two self-driving cars like today every household buys two cars. That just seems really unlikely. It looks more likely that instead of two cars I buy one self-driving car and rearrange life a tiny bit, since when I’m done with it, it can drive home and help someone else.
So the market could easily drop 50%. But wait, there’s more!
Why would I have one at all? Pizza Hut will have them. They’ll send a car to pick me up and run an errand so long as I stop in for pizza. Or if I want to visit a friend, they can just send their car to pick me up. Or, one of the ride-sharing services. So, really, would I buy a self-driving car at all? So if it’s mostly fleets buying the things, you can kiss goodbye to the network of dealers. And since most of these things will be electric, the current network of maintenance shops is going to go away.
At minimum the landscape around driving is going to change dramatically. It’s hard to see drive-thrus working out like they do today, or taxi companies… or mapping companies.
Because in a world of self-driving cars, why do we need traffic lights? Why would we need stop signs or lane markings? How much paint do we use every year to mark lines on roads? Will we need roads? No – we won’t need roads covered in asphalt, we only need the parts the wheels go on covered and can leave a gap in the middle of weaker, or no, material. So the cost of building or maintaining a freeway is going to drop off a cliff too.
And in that world, maps themselves are going to be completely different. Why would you need street signs? In fact, why name anything at all? The name of the highway you’re on is completely meaningless since you don’t use that information, the car does.
Copyright when first envisaged granted a limited-term monopoly on a work which then later fell in to the public domain, or PD. This would give the author some amount of time to make money and pay the mortgage, balanced with allowing people later on to take the work and build upon it. So you write a book, you can sell it but nobody else can, and then some number of years later everybody can do as they please with your book.
This is no longer the case. Copyright is effectively infinite. This means that while we can take old works and build upon them (Pinocchio) we cannot do the same with even pretty old works (Mickey Mouse). Edge cases exist of course, for example you can in many places use old works for parody.
Some people wish to make their work available under less restrictive terms than owning it forever. For them, there are a set of licenses which they can use to release their works.
- They can claim attribution. Broadly, this means you can use my work but you have to say where it came from.
- They can claim share-alike. Broadly, this means you can use my work, but any derivative works need to also be sharable. So you can’t take my book and then rewrite portions and claim it for yourself.
- They can claim commercial rights. Essentially this means you can use my work for anything but profit.
- You can use some combination of the above.
Thus instead of claiming copyright forever for your new book, photograph or software, you could instead for example say “use it however you wish but all changes must be shared-alike and you can’t use it commercially”. This allows individuals and companies to put works out there and allow them to spread more easily than if they retained all the copyrights.
Two basic methods of making money have emerged while using these open licenses:
- The intellectual piece is free, but any physical product costs money. For example, 3D Robotics software for drones is freely downloadable and you can change it. But if you want a physical, flying drone then that costs money. Very similar is this: the basic software is free but some critical piece required for some use case requires payment.
- The work is available publicly under a difficult open license, but privately under some commercial agreement. This is known as dual licensing. The downside is that to encourage commercial usage, the open license tends to be as painful as possible. This way, a student at home is unaffected but a company might find a license difficult. Perhaps it requires a legal review, or places burdens on the company like open sourcing everything they do. To avoid this pain, they pay for the commercial license.
The trouble here is we still don’t have things leaking in to the public domain over time. It’s seen that once a work is licensed under some license that it’s stuck there until the end of man.
What if we changed that?
I propose we engage in some kind of license ascent over time. Perhaps descent would be better. Under this scheme, some work starts out under a restrictive and painful license and over time makes it’s way in to the public domain. For example:
- I write a book. For the first year, it is available under a attribution, share-alike non-commercial license.
- After the first year, it is available attribution, non-commercial.
- After the second year, it is available under attribution.
- After the third year, it is available public-domain.
We are reintroducing the concept of the work leaking in to the public domain gradually. So when I first create some piece of work I own it outright and then over time it becomes less and less burdened.
For static works like a book, the timelines may be longer. Say, two or five years per step. For works which changed all the time, like datasets about the world, perhaps each step lasts a year. Why would we want to do this? Two reasons: Because otherwise it’s really hard to make money from open source, and otherwise open projects don’t benefit the public domain.
There are classes of works which require “giving back” like OpenStreetMap in order to attract people to contribute. That is, why would you contribute to OSM if you couldn’t access the data? OSM has now existed for 11 years and the state of mapping in the public domain is still essentially the same as it was 11 years ago. But what if OSM data dropped in to a more liberal license, or the public domain, over time? Perhaps we could have a PD version of OSM but it was 5 years old. It wouldn’t compete with OSM itself, but it would enrich what people could build on without restrictions.
Put another way, do we want OSM to be perfect in another 10 years and the public domain still be essentially unusable? Wouldn’t it be nice to improve both OSM and (for free!) the public domain maps available?
Now imagine you have some new project which requires crowdsourcing to succeed. Dual licensing has the downside that picking the open license has many difficulties. You want to pick something that encourages people to contribute yet allows you to retain space to sell things, and this isn’t easy. If instead you practiced license ascent then everybody gets the data at some point in the future. Perhaps if you are a PD person you wait 3 years, and a share-alike person you have to wait two years. But either way, it’s better than never getting the data under a license that you would consider useful.
And, it does this whilst allowing the project or company to make money off the freshest data. It also creates an incentive to make the data fresher, all the time, because otherwise the old data will be good enough for people.
Now you could argue that any project should be open from the start, but open projects tend to have significant downsides. Open projects are terrible at user interaction and experience. They’re terrible at design. They tend to be incoherent. But they are great at innovation and collecting data. At the other end, private companies which collect data tend to be great at design and so on, but terrible at innovation and collecting data because they don’t have volunteers. I posit that license ascent is a way to achieve both and that it’s better than just picking a license or selling widgets on the side.
Last month I ran a mapping party in Castle Rock, Colorado at the new Philip S. Miller Park:
The park was challenging for a few reasons:
- Nothing on the map before we did the party
- No up-to-date aerial imagery
- Lots of footpaths in winter mud conditions
Luckily we had a bunch of enthusiastic people at the event. The footpaths were easily captured using GPS units but the new buildings, football field and other macroscopic features were harder to do.
Drones to the rescue!
Luckily I own a Phantom Vision 2+ drone which looks like this:
So I sent it up to 500 feet or so and took some pictures with the HD camera which looked like this:
The image shows part of the car park, internal access roads and the new sports building (red) and swimming pool (beige). Having some pictures is great, but what we needed to do was patch the images together to be able to map on top of them. You take these warped images from some height, location, yaw, pitch and roll and stitch them in to something flat and usable .
Enter MapWarper. This web-based tool will help you spit out that map:
You’re looking at multiple images stitched together, click it for a bigger interactive version. MapWarper is a little clunky in the work flow as it stands today. Each image is stitched to OSM as a ground truth and then you use multiple of those in to a layer. The problem here is when you have no ground reference to stitch to, which is the issue we had. It would be super useful to be able to stitch images to each other, and to the ground rather than having multiple images in free space. Still, the thing basically works but is best used (and apparently intended for) single high altitude images, or old maps. Not multiple images like I did.
One solution would be to send the drone higher and cross fingers it doesn’t decide to fly away or something, and take a single image that way. The downside is lower resolution of the imagery. Upside it (hopefully) less distortion from the fairly wide angle lens the Phantom Vision 2+ has.
You can go from MapWarper to editing using iD on the OSM website pretty trivially, and anyone can now use the imagery to help improve the map. So one person can go through all the imagery pain, but then everyone else can use the imagery as if it was just any other layer. Big savings there.
So what’s important here? I think it’s a new tool in the belt for use at mapping parties (and a new set of toys to play with). You’re no longer (and haven’t been for a while) restricted to existing imagery and GPS units. For a fairly modest cost you can collect your own live imagery and make maps better, all by yourself.
I’ll be on Science Friday around 12:40 pacific / 3:40 eastern talking about OpenStreetMap. Here’s the page about the segment which should have the audio later. You should be able to listen to it on your local NPR station though it’s a PRI show. Colorado Public Radio has it, for example.