Monthly Archives: March 2016

Maps and mappers of the 2016 calendar: Nathan Saylor

In our series “Maps and mappers of the 2016 calendar” we will present throughout 2016 the mapmakers who submitted their creations for inclusion in the 2016 GeoHipster calendar.

***

Nathan Saylor

Q: Tell us about yourself.

A: I’m the GIS coordinator for Hardin County, Ohio, where I do a variety of map projects to support and promote various county entities. I really enjoy my position there as I’m a one-person shop and there is ample opportunity to learn and experiment with my craft.

I am also the owner of Saylor Mapping. This was started to answer the many requests coming to the county GIS for cemetery mapping services that the county felt was beyond its reasonable scope to handle. Saylor Mapping is also breaking into municipal utilities as well.

I am also very involved with #GISTribe, which has a scheduled meet every Wednesday at 3pm ET on Twitter (though we’re active all the time), as well as the gistri.be archive and blog.

Personally, my wife Marti and I have been married for nearly 18 years and have six clever kids.

Q: Tell us the story behind your map (what inspired you to make it, what did you learn while making it, or any other aspects of the map or its creation you would like people to know).

A: The deadline for the Ohio GIS Conference map competition was looming, so being in Buckeye country, I pondered what the map might look like if Michigan wasn’t there. I had never really thought about it, but looking at it, I considered what the impact of its sudden absence might be and within about 30 minutes came up with some economic reasons why this might be proposed by the fictitious Ohio Consortium for Greater Lakes.

Of course this is born out of the well-known rivalry between Ohio State and Michigan, and this was totally a play at the judges. While I had a lot of humored responses and requests for copies (download available here), sadly Ohio missed the opportunity to formally recognize the genius in their midst (I say with tongue firmly in cheek).

Q: Tell us about the tools, data, etc., you used to make the map.

A: The data was from Natural Earth, and I used ArcGIS Desktop. If you fancy yourself a fontophile and an anglophile, you’ll have noticed the font used for the lakes is Blackadder which was named for a British comedy alluding to the jest in which this map was made.

Go Bucks!

'Great Lake Expansion Proposal' by Nathan Saylor
‘Great Lake Expansion Proposal’ by Nathan Saylor

Will Cadell: “People talk about ‘thinking outside the box’; I don’t think there is a box anymore”

Will Cadell
Will Cadell
Will Cadell is the founder and CEO of Sparkgeo.com, a Prince George-based business which builds geospatial technology for some of the biggest companies on Earth. Since starting Sparkgeo, he has been helping startups, large enterprises, and non-profits across North America make the most of location and geospatial technology.

Leading a team of highly specialized, deeply skilled geospatial web engineers, Will has built products, won patents, and generally broken the rules. Holding a degree in Electronic and Electrical Engineering and a Masters in Environmental Remote Sensing, Will has worked in academia, government, and in the private sector on two different continents, making things better with technology. He is on the board of Innovation Central Society, a non-profit society committed to growing and supporting technology entrepreneurs in North Central BC.

Q: Sparkgeo. What does your company do exactly? Do you have any competitors in the custom geospatial consulting field?

A: At Sparkgeo we put maps on the internet.

I try to keep this description as simple as possible. It goes back to the question of what a GIS person does, which is actually really hard to explain and terribly boring at dinner parties. Instead, I stick to maps and the internet, both of which are critical features of what Sparkgeo does. The other leg of our stool is people. Really, we work in places where maps meet people on the internet.

We find ourselves doing lots of interesting things. Things like building data pipelines, building geospatial compute engines, building UIs, undertaking broad data acquisition and analysis projects. We have found ourselves in the enviable position of only doing interesting things.

With that in mind we end up touching the “full stack”. A web mapping project is actually a full stack effort; you must consider every piece of the data flow to build a great map. The web map is the tip of the spear, but the data supporting that map is really the shaft, it’s the weight of the effort. Understanding the linkages between data and its delivery, and being somewhat flexible about how to sculpt those linkages is why Sparkgeo is useful.

I am sure we have competitors, but there is simply so much important geospatial work to do in the technology sector presently, I don’t feel pressured by it. Really the most competition is for talent. Indeed, that talent gap is to a large extent why we exist.

In the last year or so we have been spending our spare time on maptiks.com, which is like Google Analytics for a web map. Our thinking here is that although lots of organisations spend time on mapping technology, few seem to iterate back over their maps to make them incrementally better, and fewer still inform that process with actual data.

Q: What libraries and tools does your company use? Can you provide some examples of your favorite projects?

A: These are some of our favorite things:

  • Mapnik
  • Python
    • GDAL & OGR
    • GEOS
    • Shapely
    • Fiona
    • Django
    • CherryPy
    • Pillow (PIL)
  • Javascript
    • Mapping APIs:
      • LeafletJS (inc cartodb.js & mapbox.js)
      • Openlayers
      • Google Maps
      • ESRI JavaScript API
      • Cesium.js
    • Turf.js
    • AngularJS
    • ReactJS
  • Amazon Web Services
    • All the things

However, it’s not about the tools or the library; it should be about the question and how best to answer it. Sometimes the best answer is “don’t do this thing”, sometimes the answer is “buy a bigger boat”, and sometimes the answer is “we’ll help you build a thing”. We are in the enviable position of not having to sell licenses for anyone so we can actually be objective (and opinionated) about technology choices.

Q: Tell us about your work with Nextdoor. What technology stack did you use? What lessons did you learn?  

A: Of course I can’t tell you too much about how Nextdoor works. What I can say is that we have helped them achieve a number of their business goals through the development of a custom geospatial datastore accessed through a custom python API. In essence though, we just added some focused geospatial expertise to their already talented engineering team.

We have used this model a great deal in helping technology companies achieve their geospatial business goals. By attending stand ups and taking on the “geo” tickets we can add the capacity necessary to give a typical web engineering team the geo-confidence they need to keep their velocity up. Often these kind of engagements become much longer term relationships.

Although ultimately we are a “professional services” organisation, we have become a lot more about people and relationships than we are about projects and requirements. That way we get to work with some of the biggest tech companies, hottest startups, and most interesting non-profits on Earth.

Q: You are the CEO of your company. Describe the tasks you do in a typical week.  

A: I talk to a lot of people. I write a lot of emails. I pitch ridiculous ideas. I write reports. I do a bunch of administration. I solve problems. I remove barriers. I remind clients about our invoices. I go and buy snacks for the team. I manage payroll. I ponder our future. I talk to our accountant about tax management. I go buy more sticky notes for the office. I stress about project pipelines. In fact I stress about a lot of things 🙂

…and occasionally I get to write some code or make a map. It’s actually the best job I’ve ever had. I’m always having to learn new things and solve new problems. I tell everyone, including our team, if they stop learning they should leave. That’s true across the board, there are too many interesting things to do to waste time being bored.

Q: You recently wrote an article about remote working (http://buff.ly/1OEEWte). What does the breakdown of your company look like? How many are remote and how many work from the office? How do you bring together everyone? How do you promote company culture with remote workers? Explain how you manage/check in with employees that are working remote. What are the strengths/weaknesses in the current setup?

A: I first heard the term “remote first” in terms of the workplace mid last year and I realised it fits us well. We have an office in Prince George, BC (well North of the wall). But on any given day a member of the the team could be anywhere and it’s not a big problem. We typically have a check-in meeting, our version of a stand up — except people are on different projects — at 8:30am Pacific. It lasts for 15 minutes max, plus any necessary bonus rounds. People attend it from where they happen to be. It’s “early” for the Pacific timezone because we have some people on Eastern time. Even a remote company has to figure out timezones 🙂

Remote first means that we communicate first using tools like Hangouts and Slack, it means that people are kept in the loop by default, and things don’t get decided “without the remote guy”. If a team is meeting, then everyone on that team is invited, and that meeting will happen on a common set of tools used by people in or out of the office.

This remote culture is critical for us. The first employee I brought on (@gridcell) was remote, and now 40% of our company is remote. The really important bit, however, is how we interact with our clients. Being based up in the frozen wastes of the North and working for organisations in the tech sector means we must be really good at “being remote” because we are always remote to our clients. So independent of whether Sparkgeoers are in the office or not, we are still operating in a remote manner. This must be true for our clients too; they need to be ok with our periodic on-site presence, and our very present nature on IM or videoconferencing. It’s worth noting that with some clients being remote is not a great fit, and that’s fine. I am happy to say that we do have some great clients whom I have never actually met. Likewise, I also know that sometimes I have to hop on a plane and travel 2000 miles to shake a hand. Remote doesn’t mean not having a personal relationship and remote doesn’t mean distant.

Q: How does your company advertise? SEO? Content? Starting up Slack groups…?

A: Sometimes we do little ads in places across the interwebz (for instance on the GeoHipster website), seeing what sticks. Relevant content, however is the most valuable piece of advertising on the web. Good content has a long tail and brings people back and back.

More recently, our Maptiks growth guy (@julienjacques) suggested starting a Slack group. We did that, and now it’s grown to 900 users. The funny thing is it has turned into a real community, and as such we don’t really advertise on it because that would defeat the purpose that has evolved around it. If we (or others) were to advertise crassly on it, then it wouldn’t actually be a useful community 🙂

Q: You follow mapping trends and new technologies in depth. Are there any particular tech companies and/or startups that you follow? Any of them going to be the next big-bang disruptors?

A: I follow all the usual suspects (CartoDB, Mapbox, Esri, Stamen, Google, Boundless…) and do my best to catch up with contacts in each of them on my travels.

I am especially interested in the satellite space (hahaha) right now. The idea of Imagery As A Service seems to be booming. Planet Labs, Digital Globe, Astro Digital, UrtheCast, SkyBox, Spire all being players to some extent. Then there are companies like Orbital Insight who are taking remote sensing and magic-ing it into actionable broad data products. I think the days where someone would go to a website and purchase an image from a marginally navigable image library, then download an enormous file.zip via FTP are numbered… thankfully.

But there are the big players, too. Apple is very interesting right now; Google has always been in the geo sand box. But with the consumerization of geo I think more players will emerge here. Amazon has a location platform and drones…? UPS…?

Then there is the sharing economy: Uber, Lyft, AirBnB, Nextdoor. The quantified & wearable self: Fitbit, Under Armor (who bought MapMyFitness), Strava, RunKeeper. The nature of our industry and the ubiquity of smartphones & wearables is such that good, hard geospatial questions pop up everywhere. As a result of this phenomenon we’ve worked in the tech space, in hospitality, in finance, in conservation, with satellite companies, in hardware, in software, with government. The point isn’t to consider the tools or be confined to a vertical, the point is the pursuit of interesting questions and how we can use geography and technology to answer them. People talk about “thinking outside the box”, I don’t think there is a box anymore, I wonder if there ever was.

Q: Which industry do you see as needing more mapping technologies? Are there one or two fields that seem to be pretty behind the times?

A: Automotive will be the next industrial geospatial leapfrog.

Consider: A driverless car needs to know a great deal of information about its surroundings and virtually every piece of fixed knowledge (i.e., data not detected by vehicle in transit) will be geospatial in nature. Every major automotive vendor will need a data provider, and that data will be constantly updated. In this scenario the 80/20 rule will not suffice. If that vehicle cannot reach its destination because it doesn’t know the way, then the entire vehicle has failed. That failure might simply be a new subdivision not being present, but nevertheless the lack of a street or a misnamed building will result in the vehicle not being able to drive itself, thus failure. A driverless vehicle needs to have a complete and constantly updating map of navigable routes.

Automotive will drive (hahaha) efforts in open data, in data pipelines, in ETL, in base map production, in data storage, in connectivity, in routing. For the driverless future to happen geospatial needs to be a lot better.

Q: What is your current method for skiing on a mountain you do not know well? Do you use the paper maps that they provide, or a new app (Have you heard of fatmap.com?)

A: I ski a DPS Wailer 99. I love the backcountry, but with a young family I find myself on a ski hill more often these days. That said, my 7- & 9-year-olds are on double diamonds now, so we’ll be hitting some family backcountry soon. We are also lucky enough to have many kilometers of groomed & floodlit XC skiing within city limits (Prince George, BC) so that is a common after-work activity.

The interesting thing about backcountry skiing around central British Columbia is the lack of documentation; every trip is a little bit exploratory. That, combined with relatively poor and out-of-date maps (Canada is big and largely empty) leaves me doing a lot of navigation by feel. It can get pretty cold, too; devices and batteries tend to become less reliable below -20.

I do, of course, appreciate the irony in the mapping guy navigating largely by instinct.  

Q: It must be fascinating to compare the world you grew up in with the one that your daughters are growing up in. Do you mind sharing a little insight you have as a parent and geographer/technologists. Can they read maps? What routing technology do they use to get to a new place?

A: My girls love maps. They have been completely brainwashed by me; they know exactly how important good maps are. Their navigational abilities are somewhat untested, but they do have a good sense of direction; we test that on the trails a lot. Their use of technology is interesting; we keep screen time to a minimum, but the way they interact with touch interfaces is fearless. I think we will see great advances in industrial design as interface designers embrace touch and haptic technologies. We are also trying to expose our girls to what it means to write code; their lives will take them in many different directions, but having some exposure to the discipline of code is valuable.

Which do you prefer when it comes to maps?
Data or design - Both
Functionality or beauty - Again, both. But wait, “functionality” doesn’t mean lots of buttons -- it means fit for purpose. As a community we need to de-couple features from functionality.
Historical or futuristic - Neither; it’s the story that compels 
Markers or pins - These are the same thing 🙂
Clusters or heatmaps - Clusters (unless it’s a weather map) 
Markdown or Handlebars  - Markdown
And other things…?
Black and local coffee or pour-over with butter - Black Americano, no pollution, and lots of it
Fitbit or Strava - Fitbit & MapMyFitness (Fitbit have an interesting geo conundrum presently - Strides or GPS for distance https://www.linkedin.com/pulse/your-device-can-too-smart-will-cadell ). Strava has done an amazing job of socializing athletic pursuit; I started using MMF first though, and most of my data now gets piped into Fitbit.
Twitter or Facebook - Twitter
Commuter or road bike - Both & MTB too
Nordic, alpine, or telemark skiing - Mountain Touring, Skate Skiing, Classic Skiing. Tele is cool but you have to be really talented to ski anything big, and I’m not 🙂

Q: Any closing comments for the GeoHipster readers?

A: Thanks for the opportunity to tell you a little about Sparkgeo. Also, thanks to the geohipster community for keeping things sufficiently geo-weird.

 

Maps and mappers of the 2016 calendar: Kenneth Field

In our series “Maps and mappers of the 2016 calendar” we will present throughout 2016 the mapmakers who submitted their creations for inclusion in the 2016 GeoHipster calendar.

***

Kenneth Field

Q: Tell us about yourself.

A: I tend to call myself a professional cartonerd having never had a job with the word ‘cartographer’ in it. I have a Bachelors in cartography and PhD in GIS and spent 20 years in academia the UK. I was Course Director for GIS programmes at Kingston University in London and did all the usual academic stuff of research, teaching, supervising students, publishing etc. I’ve been privileged to have won a few awards for my maps, writings about maps and interior design (kitchen tiles!). I recently ended a 9-year stint as Editor of The Cartographic Journal and I’m currently Chair of the ICA Map Design Commission. I also co-founded The Journal of Maps, and am on the advisory board for the International Journal of Cartography.

I got totally frustrated by the admin-heavy bureaucratic nonsense of University life and moved to the dArc Side in 2011 to work with Esri to support high quality cartography and help develop the next generation of tools to support more intuitive, better map-making. That involved moving from the UK to California which is a switch I can heartily recommend. I’ve been called a ‘cartographer in residence’ though that implies some sort of temporary job which I hope isn’t the case. It’s a terrific place to work and I have so much freedom to experiment and push the boundaries of what’s possible in cartography.

I’m an advocate for high quality cartography and deliver keynotes, workshops, training and research support internationally and on the conference circuit. I blog about the good, the bad and the ugly under various guises (blogs.esri.com, cartonerd.com and mapdesign.icaci.org), tweet far too much (@kennethfield) and sometimes I make maps (carto.maps.arcgis.com). I’m currently writing a book on cartography with Damien Saunder which we hope will be out by the end of the year– the more I publicize that, the more I am committed to getting it finished! More than anything I’m passionate about encouraging and helping others make better maps through identifying and sharing best practices (and explaining cartofails).

I can also be found on a snowboard, in a pair of hiking boots, behind a drum kit, or supporting my once great football team Nottingham Forest.

Q: Tell us the story behind your map (what inspired you to make it, what did you learn while making it, or any other aspects of the map or its creation you would like people to know).

A: I tend to make thematic maps. I’ve always been skeptical about 3D thematic cartography and often struggle to find a compelling reason to switch from a planimetric map. The key, for me, is to use that third z dimension for something really useful and not just for the sake of making a glitzy map. As technology has improved to overcome many of the failings of static 3D cartography (fixed point of view, occlusions, labelling, etc.) I figured it was time to experiment. The 2015 General Election in the United Kingdom gave me the excuse I needed. It’s an online map, called Political Causeway, which you can view here (use Firefox or Chrome).

The map shows the results of the UK election in 3D on an interactive virtual globe and positions it as a new development of the tradition of using cartograms to represent election results.
The map shows the results of the UK election in 3D on an interactive virtual globe and positions it as a new development of the tradition of using cartograms to represent election results.

The challenge of trying to display results for 650 irregularly shaped and sized constituencies in the UK has spawned many different outcomes. Maps that show geographical constituency boundaries face the challenge of trying to accommodate distortions arising from the simultaneous display of visually incomparable areas. People are disproportionately distributed across space, and while constituencies attempt to iron this out for voting it creates a cartographic problem. After Danny Dorling’s innovative work on the development of cartograms for elections, we’re seeing increased use, particularly of hexagons, for political cartography. Of course, using hexagons as a data-binning technique can be traced back to the mid-1800s, but they are the very essence of carto-hipsterism. Equal-area and tessellated hexagons provide a good visual structure that also allows a reasonable amount of adjacency topology to be incorporated. They are abstract but ultimately a good visual way to display election results.

And why 3D? Partly the technical challenge, but also the structure of the voting meant I could use separate layers on the map to encode different aspects of the vote — with winners sitting on top and other political parties being represented in a second-place layer, third-place layer and so on.

For the 2015 UK election, virtually every media organisation used some sort of hexagonal cartogram (I wrote a blog on it here). As a fan of cartograms I also wanted to use hexagons, but I also wanted a challenge: to develop a three-dimensional hexagonal cartogram on a spherical virtual globe… and make it make sense. The map was inspired by a picture of David Cameron visiting the Giant’s Causeway in Northern Ireland taken in 2013. Hexagonal mapping and political photo-opportunities collided and my map idea was born. The map went on to win the Google Award for UK election mapping at the 2015 British Cartographic Society Annual Symposium.

Q: Tell us about the tools, data, etc., you used to make the map.

A: The map was designed and produced as part of my work at Esri. I always attempt to challenge conventional thinking while building maps that are heavily informed by cartographic concept and theory. Sometimes this requires us to think outside the box, break rules, and bring new ideas to the table. It’s an exciting area to be involved with — marrying creativity with new technology to create innovative and interesting cartographic products.

In two-dimensional space the creation of equal area tessellated hexagons is relatively simple. Creating hexagons that curve across a surface is more complicated due to the geometry involved. The initial challenge was to build a set of hexagonal regions that partition the Earth’s surface. This was achieved computationally by creating an icosahedral discrete global grid. A number of grids of different resolutions were built and used in different ways in the final map. I used Kevin Sahr’s excellent DGGRID program to generate the grids — the key being that to wrap a hexagonal mesh around the globe the overall mesh has to include occasional pentagons else the tessellation wouldn’t work. Think of a football (a.k.a. soccer ball).

Once general grids had been developed, they were further processed to build a grid of 650 tessellating hexagons covering the UK. The Election results were manually collated during the live television coverage and entered into a spreadsheet, then processed into formats to support the creation of different layers in the final map. ArcGIS Pro was used to build a layer of three-dimensional extruded hexagonal polygons for each of four layers representing the winners, runners-up, third place, and also-rans (others). I’m calling them hexstones. Each hexstone in each layer was extruded proportionally to the candidate’s number of votes giving a causeway-like 3D surface and volumetric blocks. The base heights for each layer were modified so each layer sits relative to the layer beneath to build up the final three-dimensional political causeway. The model ultimately looks like a way of viewing the stratification of the election results a little like we might cut away layers of geological structure to see beneath. I liked the way this supports the causeway metaphor beyond simply the hexagonal shapes.

One of the limitations of any three-dimensional map of prisms (of whatever shape) is the inability to look ‘inside’. This is overcome through interactivity in the final map, but I also created a capstone for each constituency that shows the share of vote for the same four layers of results but in a single layer. This provides a way of seeing the political pattern of voting across all candidates across the top of the hexstones. It also allows the map to be viewed from above and reveal more than just the winners. At least — that’s the idea.

A range of supporting datasets (a custom 2D base map of hexagonal patterns, 3D leader lines, 3D labels, a 3D legend and pop-ups) were produced to support the final map, so it becomes something to explore rather than just look at and expect everything to magically happen for you. Interaction in a 3D environment is absolutely critical.

The map was published from ArcGIS Pro to Portal for ArcGIS into a 3D Web Scene which takes advantage of the 3D capabilities of WebGL browsers. The map must therefore be viewed in either Google Chrome or Firefox. The published web scene layers were configured to build the final map. The icosahedral global grid data was used to create a custom hexagonal basemap of different resolutions to create an abstract world map. Using a standard map of real global geography would not have suited the abstract 3D cartogram. A little transparency allows an imagery layer to bleed through, giving just a hint of a real world. The UK is outlined in hexagons of the same grid used for the 3D symbols. This illustrates the bloated England 3D cartogram compared to the fewer constituencies in Scotland and Northern Ireland. It provides an explicit comparison of the size and shape of the cartogram compared to the hexagonal representation of the real geography of the UK.

The 3D hexstone and capstone layers can be turned on and off in the legend. This supports the viewing of not only the winners, but the landscape of the runners-up, third place and the also rans. Party colours across the map give a recognisable link to the political affiliations. The undulating nature of the hexstones shows total voter turnout across the map… a small but subtle illustration of where the electorate were motivated to vote to a greater or lesser extent.

A layer of labels can be added to the map. These are scale-dependent so as you zoom in, pan and rotate the globe they update to give a reasonable amount of labelling in the immediate view atop the causeway. Simply adding all labels at once would swamp the map. More are revealed as you zoom in, and vertical leader lines anchor the labels to each constituency hexstone/capstone, and pop-ups can be revealed by clicking the label. This gives detailed results for each constituency, and reveals the full statistical makeup of the results.

A legend is viewable, again built from 3D hexstones that shows the party affiliated colours (to aid map interpretation), this time proportionally scaled in height by the total number of constituencies each party gained. Legend labels can be queried to get further details of the map and the overall results.

The map is fully interactive so you can zoom, pan and rotate. This gives the map user an ability to zoom across the landscape and position the view camera to any desired location and angle. This partly overcomes the limitation of a static 3D map that some features are inevitably occluded and foreshortened. Some pre-fixed positions are available to provide quick navigation. Of course, foreshortening does occur because it’s draped across a virtual globe with a curved surface. I’d usually use an isometric projection, but given each hexstone is broadly the same height (scaled by voter turnout) the need to visually assess the height of a hexstone and compare to another isn’t a crucial cognitive task. In addition, because the UK is relatively small, the curvature of Earth has a minimal impact.

This map exhibits some degree of technical and conceptual innovation. It has certainly pushed our ability to develop 3D products that support ease of use, clarity and interpretation, and it pushes election mapping and the use of hexagonal cartograms as a way of representing and reporting results. It was an hex-periment certainly, and I’m calling it helecxagon mapping.

'Political Causeway' by Kenneth Field
‘Political Causeway’ by Kenneth Field

Jan Erik Solem: “Only start something if you are willing to spend at least five years of your life on it”

 

Jan Erik Solem
Jan Erik Solem
Jan Erik Solem is the CEO of Mapillary and is passionate about all things computer vision-related. Prior to co-founding Mapillary in 2013, he founded Polar Rose -- a face recognition software for mobile and web, which Apple bought in 2010. Solem has published over 15 patents and applications, and is the author of a best-selling computer vision book called Programming Computer Vision with Python. Solem resides in Sweden, and is an associate professor at Lund University.

Q: Each city has its own “flavor” of geo-scene, what is the geo-scene like in Copenhagen / Malmö?

A: It is a mix of community groups, GIS companies, and location services startups. There are regular meetups for GIS geeks in a forum called SamGIS, and more-entrepreneurial events like the LocationDay conference. Wayfinder was an important node in Malmö before Vodafone shut that down. If I were to add a “flavor”, I’d say that it is mobile-focused. Traditionally, a lot has revolved around mobile because of the strong presence of Sony Mobile (formerly Ericsson, then Sony-Ericsson, now Sony).

Q: Your background is in applied mathematics, and you built a company around facial recognition. What brought you to geo?

A: I’m a computer vision person. My work from undergrad to PhD all revolved around reconstructing scenes and recognizing objects from images. I started a face recognition company during my PhD days and worked on that for about 6 years, then at Apple for a few years.

Since my PhD days I have always been interested in building a system for collaboratively reconstructing places from photos, and over the years I saw the connection between that and geo grow in importance. Me and the team at Mapillary are all fairly new to geo and are constantly learning. We’re approaching things from a computer vision angle and have no past technology choices or biases in geo.

Q: Describe your “A-HA” moment when you conceptualized Mapillary?

A: I built a prototype iOS app and persuaded two friends to bring their phones and map a newly-built area in the Malmö harbor. We spent one hour on a cold Sunday morning collecting data. When I looked at the results, that was the real a-ha moment. Things became very concrete once I saw that data from phones would be good enough. One of the friends who helped me that Sunday morning was my PhD student Yubin who also became a co-founder of Mapillary.

Q: What do you see as Mapillary’s biggest challenge in the geospatial business space?

A: Delivering a great experience worldwide. We’re a global company, anyone can contribute anywhere, and our community is in over 170 countries. What we’re seeing is that in developing countries devices are just catching up now, and making the app run well is a challenge sometimes. Also connectivity is poor in many areas where we have a challenge in effectively serving and collecting content. This goes for both our community and for our customers.

Q: The niche that Mapillary is filling is exciting and is applicable in a number of areas. Describe for our readers some of the swank things your users have done with your technology?

A:

Q: Judging from your Twitter feed and blog, all you do is work. What do you do when you aren’t grinding on a keyboard?

A: Honestly, it is mostly work. I have three young children. Besides work and family there is little room for other activities. I don’t watch TV, play games, follow sports, or other “normal” everyday activities. I ride bikes when I have the time, and I love to do longer rides and races.

Q: Where do you see Mapillary in 10 years?

A: We’re aiming to be the best solution for visualizing places and for exploring and visiting places through photos. As a side effect, if successful in that, we are also a great provider of data for mapping, navigation, and automotive. We want to build an independent entity that can be integrated anywhere — in apps, in mapping platforms, websites etc.

Q: As the founder of two successful startups, what advice do you have for the geohipsters out there?

A:

  • Only start something if you are willing to spend at least five years of your life on it. Building something great takes time and commitment. Even fast-growing companies take years to develop.
  • You are not your education. Most skills can be learned. The engineer can learn basic business skills, the business graduate can learn to code, etc. Pick up what skills you need to do the necessary work and continue learning, always.
  • Small teams can do amazing things. It is entirely doable to have the best solution without having a large organization.
  • There is no shortcut to hard work. The “work smarter” advocates have it all wrong, the hours you put in are crucial.

Q: Lastly, do you consider yourself a geohipster?

A: Um, probably not 😉 .