Emily Chang - eHub Interviews

twittering: woo! awesome sunday

eHub Interviews »»

Visit Mappr, originally added to eHub on Sep 10, 05.

eHub Interviews Mappr

Thanks to Eric Rodenbeck, Mike Migurski, and Tomas Apodaca of Mappr for this email interview posted October 20, 2005.

eHub: What is your web application/service about?

Mappr: Mappr is an interactive environment for exploring place, based on the photos people take. It looks through digital photos that people have posted to the popular picture-sharing website Flickr.com, and places them on a map of the US. The act of tagging a photo with a city or state name (or both) is enough for Mappr to make a best guess as to where that photo was taken.

We launched the site in January 2005. 1,214,671 flickr photos have been mapped as of October 5, 2005. Visitors to the site use a map, built in Flash, to browse through these photos. The result is a way of exploring that can lead to new understandings of the relationships between images and locations.

Mappr anticipates a time when GPS (Global Positioning System) data will be built into every camera and attached to every photo, without having to wait for that technology to come on line. When it does, we hope that Mappr will have suggested new, location-based ways that people can share their pictures with one another.

eHub: Why did you start this project?

Mappr: We spent much of 2004 working on mapping projects with Moveon.org. Our work with them culminated in a live map application—50,000 people participated in a simultaneous online conference with Moveon staffers and one another, and this community was able to get a visual sense for how big it is that they hadn’t seen before. It was a striking example of the power that data visualization has to quickly and effectively communicate complex ideas.

With the election over, we started looking around for other datasets to apply this technology to. Flickr’s open API seemed like the best first case, because of the huge numbers of photos available to us and the fact that a great many of them were tagged with information we could use to determine geographic location. If a photo was tagged with “miami” and “florida”, we could make a pretty good guess where it was. If it’s only tagged with “miami”, it’s probably in Miami, FL, but since there are cities named Miami in Oklahoma and Arizona, we had to allow for those as well… so it became quite an interesting problem.

We mainly wanted to see if it would work - if the tags that individual people added to their images could be used to draw other kinds of conclusions about the data. The fact that the people who were tagging their photos weren’t specifically trying to put them on a map was very important to us. There are plenty of ways to be able to intentionally assign location to photographs, but the idea of being able to look at a large dataset and be able to tease out unintended patters seemed alot more interesting.

What we found was that a taxonomy (a rigid hierarchical system, namely the locations of cities inside of states) could be applied to a folksonomy (flickr’s user-generated tags, which don’t fit into any hierarchy), with some interesting results. For example, a Mappr search on “route66” shows photos from along this highway stretching across the country - and neither Flickr nor Mappr know anything about route 66. Finding these kind of ad hoc patterns emerging from a large data sets are really what got us excited.

eHub: How much time do you devote to its growth? Do you have a day job?

Mappr: We don’t devote much time to it at all, anymore - all the processes needed to find and map photos on Flickr are completely automated. From time to time, when alot of people are banging on the server at the same time, we need to throttle it back a bit to keep the server from catching fire - but other than that, we don’t need to touch it, aside from responding to emails about the project.

The project was developed by Stamen Design, and we definitely have day jobs! We’re a design and technology firm in San Francisco, and we’re working for clients all the time - but we make a real effort to carve out enough time to do investigative projects like Mappr. We’ve found that the best way to keep ourselves sharp and engaged is to allow projects like Mappr to emerge from our client work, and for the lessons that we learn doing our own projects to find their way back into our client work.

eHub: How large is your team and what are your backgrounds?

Mappr: Eric Rodenbeck, Founder and Design Lead. Rodenbeck founded Stamen in 1999, motivated by a strong desire to extend the boundaries of interactive media. As Art and Narrative Director at Quokka Sports, he crafted the design and storytelling structures for numberous online sports and adventure sites, notably the 1997-1998 Whitbread Round The World sailing race and Trango Tower. He studied architecture at The Cooper Union in New York City and has a degree in the history and philosophy of technology from The New School.

Michal Migurski, Technical Lead. With Stamen, Migurski has helped create a body of data visualization and interactive work for clients such as MoveOn, BMW, the University of Southern California, and others. Stamen is actively engaged in visual exploration of web services, and is responsible for experimental projects such as In The News and Vox Delicii. Migurski has been working with maps in interactive web-based applications since 2001. He is a graduate of UC Berkeley’s Cognitive Science program.

Tomas Apodaca, Engineer. For the past year and a half, Apodaca has been working with Eric and Michal at Stamen, building interfaces and visualizations for massive amounts of data. Before that, he was a cofounder of Angry Monkey, a company that created feature film websites and DVD menus, produced cartoons and a music video, developed prototypes for commercial project management software and spun off a research and consulting firm. During the dot com boom, he worked at Organic, where he made games and interactive pieces for that company’s many corporate clients.

eHub: What is your design philosophy?

Mappr: “Stamen thinks by doing.” We are a very hands-on kind of shop, and we like to design as we build. What this means is that we can avoid situations where “the technology won’t work with the design” - we never have problems where the comps don’t work, because the design and the implementation go hand in hand. We like to make things that exploit the inherent properties of the medium we’re working in, that stay true to the materials we’re using. It can take a little bit longer to work this way - but a relationship built on trust in the working process allows for changes that can come up midway through the act of making, and the result in work that feels right.

eHub: What technologies are you currently using?

Mappr: We use open-source tools exclusively on the backend - SQL, Python, Apache, etc. They’re cheaper, more robust, and it helps us stay in tune with a continuously changing technological landscape. For the frontend we generally use Macromedia Flash - but have been experimenting with AJAX lately as a possible replacement or enhancement. Mappr uses Flash for the interactive display.

The project also allows you to subscribe to RSS feeds for photos in or near locations, something that flickr doesn’t support. You can subscribe, for example, to “photos near San Francisco” and you’ll get photos from San Jose, Sebastopol, Oakland, etc. It also provides an API for people who want to query the application directly. This is, for us, one of the core philosophical underpinnings of the project - if you’re going to use someone else’s data and add value to it, then you should pass along that data and value to others who might like to use it.

eHub: If your project is live, what are the most requested features from your users/community?

Mappr: Mainly, people want Mappr to cover the whole world. In particular there seem to be alot of people in Europe who’d like to see it able to map their photos.

eHub: Does your user base reside in a primary geographic location or is it distributed?

Mappr: They seem to be all over the place - but as it’s mostly a US-centered project (we also have northern Mexico and southern Canada included), repeat visitors are mostly from there. We have a few people who are using mappr to keep track of their road trips - one guy is driving around the US taking photos of himself in front of every state capitol, with his green Mazda - so we get alot of travelers paying attention.

eHub: Where do you see the project heading in the next 6 months? The next 2 years?

Mappr: We’d love to see the project expand to cover the whole world, for sure. We’d also like to see the ability for Mappr to be a bit smarter about the way it guesses photo location. For example, if a flickr photo is tagged with “empire state building” and “new york”, and a second one is only tagged with “empire state building”, then it seems likely that the second one is in New York City. We’d like to provide tools that allow people to auto-place their photos, based on the behavior of other users, without having to add geographic information themselves.

eHub: What is the greatest challenge to your success?

Mappr: The explosive growth of flickr users since the project was launched (flickr was bought by Yahoo a while ago) is a huge deal. Mappr is running on a small server and it’s getting increasingly difficult for the machine to handle this volume - it’s like trying to drink from a firehose!

eHub: What is the one thing you need to get to the next phase of the project?

Mappr: Resources. We’d love a few more machines, so that the current server could be primarily a server box and the location heuristics could happen on a dedicated machine. In order to take Mappr world wide, we’re going to need to increase the size of the location database by a few orders of magnitude - so in addition to these new servers, we’d love to get access to something like the Getty Thesaurus of Geographic Names or something similar. These things are all expensive though, more than we’re able to cover as a small design company.

eHub: Do you have a business model? If so, what is it?

Mappr: At this point, we’re mainly interested in getting the idea out there that large datasets can be visualized in ways that they hadn’t been intended - that there are ways to look at data that can result in more interesting views. Since we launched Mappr, we’ve had a few client projects as a result, and it’s opened up some other avenues - so it’s been a great success in that respect.

eHub: If you’re able to disclose this information, how much traffic or usage do you see on an average day?

Mappr: We see anywhere between 500 and 5000 visits a day - mashups are getting alot of press these days, so from time to time Business Week or someone similar will link to us, and we’ll see a big spike.

eHub: What is the one thing you’re most proud of about the project?

Mappr: When it launched, we heard from alot of people who told us that it was an idea that they had thought of, but hadn’t done - that was pretty satisfying, because we knew we had really tapped into something people were interested in doing and thinking about.

eHub: How would you describe the shift that’s occurring with the web right now to future generations?

Mappr: You used to hear alot about how increased access to broadband would open up a whole new web, with full-motion video and 3d and fancy graphics. It hasn’t happened yet, even though lots more people have broadband - and it’s pretty clear at this point that video-on-demand isn’t what the web is for. Mike likes to say that “text is the new video” - the fact that broadband is always on is much more important than the actual size of the pipes. What we’re seeing is that the pervasive nature of broadband, that it’s always on, is by far the most powerful force in driving new kinds of applications and experiences.

eHub: What site(s) do you visit everyday other than your own?

Mappr: kottke.org, nytimes.com, dailykos.com, flickr.com come to mind - but since we use reblog, which Mike wrote, to manage our incoming RSS feeds, we tend not to need to actually go to very many sites anymore. Most of the information just comes to us via reblog, which is much more efficient.

eHub: How many hours of sleep do you get a night?

Mappr: Sleep?

Thanks to Eric Rodenbeck, Mike Migurski, and Tomas Apodaca of Mappr for this email interview posted October 20, 2005.

Visit Mappr
Originally added to eHub on Sep 10, 05

» More eHub Interviews

eHub Interviews is a series with the creators of Web 2.0 applications and services by Emily Chang, author of eHub, designer, and co-founder and principal of Ideacodes, a strategic web consultancy in San Francisco that she co-founded with Max Kiesler.

Also see eHub Interviews in Japanese at CNET Japan

Browse eHub (2543 resources)

eHub Home
Features
Interviews
News+Events
Categories
Directory
Editors
Submit a Site
Advertise on eHub

Request an Interview

If you're the creator of a web application, service or product, you can submit your site and request an interview.

Recent Interviews »»

Aug 07, 08

eHub Interviews Brightkite
Brightkite is a location-based social network that enables people to take their online profiles with them into the real world…

Jan 15, 08

eHub Interviews Photophlow
"Photophlow is a web application that lets you share photos in real-time. It is meant to be used for all…

Oct 26, 07

eHub Interviews MakeMeSustainable
MakeMeSustainable is a catalyst for change, empowering individuals to reduce their environmental impact by connecting how we feel about our…