Postopolis!: Eric Rodenbeck, Stamen

Note: This is a summary of a talk given at Postopolis!, taken in real-time, with minimal editing. Reader beware! Postopolis! was organised by BLDDBLOG, City of Sound, Inhabitat, Subtopia, and the Storefront for Art and Architecture in New York, and ran from May 29th-June 2nd 2007. Flickr group for photos here. YouTube videos uploaded here. All Postopolis! posts here.

In which Eric Rodenbeck returns to New York, with a little bit of history to clear up. He starts by revealing how he got kicked out of Cooper Union a few years ago, saying he just couldn't make those "grand architectural gestures" that seemed to be part of the deal. (Laura Kurgan, who follows later, assures him he wouldn't get thrown out of any architecture school these days.) Not that it matters either way, as Rodenbeck is a founder of Stamen, one of the more interesting new media companies around. Based in San Francisco, Stamen work with flows of data, ideally massive amounts of data in real-time. They introduce a mapping and information design sensibility to this, and construct meaning from these flows.

Central to their approach is their belief that, as Rodenbeck puts it, "data visualisation is a medium - not a technology". That is, it has conventions, gestures, constraints, techniques, even jokes. Rodenbeck runs through a few of these, including the percentage of pie chart that resembles Pacman. By this point, he's warming to his theme, an irrepressible New Yorker unfurling words and concepts as if he were his own real-time data flow. Fortunately, he assembles meaning from himself in real-time too.

So their medium is data; live, fast and deep data. Rodenbeck runs through a few early projects, such as Mappr, in which they deconstruct the Flickr photo-sharing site along an axis of location. Around 40% of the tags describing photos on Flickr are location-based, so it's a rich set. They realise that a visualisation of place can effectively draw itself, if they place photos onto a map based on the tag description. Sure enough Route 66 emerges, based on photos tagged 'Route 66'. There are a few anomalies but these are interesting in themselves. He also runs us quickly through their work with Digg Labs, which is wonderfully visually and also exposes the horror of most content at Digg, as well as their interesting Oakland crime maps [technical note here].

A favourite project of mine is Cabspotting, derived vast amounts of GPS data from cabs in San Francisco. The tracking system knew where the cabs were, whether they'd picked up and dropped off, and the time inbetween these events. Thus, it could draw a map of San Francisco, without having a map of San Francisco to reference. Rodenbeck shows a movie of 4 hours of taxi activity, with white lines indicating high speed traffic – principally on the airport run and over the bridges – and a downtown full of red lines, indicating slower progress and more pickups. The topography of the city emerges from the data beautifully. At one point, Rodenbeck seems to merge with the data himself here.

As the movie speeds up, Rodenbeck notes that it almost has "a capillery action"; that the city had same kind of mechanisms in it that, say, a heart would. This is fascinating, and links to the numerous analogies of the city as body (Ackroyd's London etc.). But it really just gives a sense of the malleability of this medium; that you can choose to map it in different ways, for different effect.

Rodenbeck also suggests that "broken isn't always bad", in terms of this data. As with the Route 66 anomalies previously, the Bay Bridge from SF to Oakland tells an interesting story based on erroneous data. The bridge is on two decks, which most conventional maps wouldn't tell you. However, as GPS receivers work less well on the lower deck, the map indicates fainter lines there; both decks of the bridge are visible. To Rodenbeck, these "seams of data", combined with places where data was missing, can be mined to show you more of the real, physical world than a conventional map would.

Finally, we see Stamen's latest project, with the real estate company Trulia – Trulia Hindsight [more here]. This quite wonderful visualisation had only been launched a few days before, so it was great to see Rodenbeck demonstrate it. It's based on the data gold mine of the real-estate company's vast amount of data from all properties sales in the United States, mapped to location and over time back to the early 20th century.

Again, with the biological metaphors. We look at San Francisco and Las Vegas growing over time – the latter rapidly – and Rodenbeck notes the similarity to mould (recall Julia Solis talking about landscapes of mould first thing that morning?). Las Vegas actually looks more like a rash suddenly flaring up (which is perhaps an appropriate representation.)

He says it needs a new vocabulary accordingly, and claims to be struggling for words to describe these patterns of data. The biological metaphors seem entirely apposite, to my mind, though Rodenbeck also suggests the terminology of film-making. Seeing a zoomed-in Levittown New Jersey, neighbourhood suddenly flare out of the map is "like a jump cut in a movie relative to a slow pan", perhaps.

It's an absolutely stunning tool for watching the development of cities, whether zoomed right out to watch a city expand its boundaries, or zoomed right in to see individual streets emerge in grids, or follow the topography of the landscape, as in San Francisco. They've collected some great examples of visualisation emerging from Trulia Hindsight here. (It's getting a lot of traffic at the moment, so be patient. Plus Flash 9 is best for this I think. I could play with this for hours, though I'm not entirely sure what business Trulia hope to derive from it, other than via brand extension. It's close to what I was getting at with my proposed Google Earth extension a while back, though that was based on historical maps rather than other data. In a sense, the data is better, having its own integrity and purpose, not having been collected or designed to draw maps – and thus portrays a more impressionistic sense of the city, which allows more room for interpretation.)

There's a brief discussion of the pros and cons of Flash 9 versus Processing – ultimately they both have their qualities and uses, and Stamen use both accordingly. Truliia's system is built on the Modest Maps framework, by Mike Migurski, Stamen's CTO.

I asked about whether they'd explored some integration with physical architecture. Rodenbeck replies that they're beginning to explore things in that area. Strikes me that there's huge potential for those who understand 'data viz' as a medium, in terms of working with both cities and buildings, given how these live, deep, fast flows of data are beginning to emerge from those things. I personally think making this behaviour visible, but in interesting expressive or impressionistic fashion, may be incredibly powerful. Laura Kurgan asked about the graphical choices made in their Oakland crime maps, and opened up a whole area of debate about representation, particularly of people (which she'd return to in her talk later.) Rodenbeck noted that you can choose to denote, say, prostitution and murder in different ways, visually, implying the difference in their 'seriousness' as a crime. Tricky stuff that, but surely one of the most important aspects to think through.

There's a character and flavour to this work which entirely backs up Rodenbeck's assertion that this is a medium of its own. Janet Abrams, responding to a review of her book with Peter Hall, 'Else/Where Mapping' in the excellent 'Urban Design Review: Spring 2007' says that "I truly hope that mappings become recognised as expressive media artifacts, rather than just useful instruments, and that their intent is more than just simply inform, but also to intrigue and convey ... hopefully they will produce emotional reactions as well as intellectual ones." In the best work of Stamen, and others, across this fertile, emerging terrain we're beginning to see 'artifacts' which delight, amaze, shock and inform. With Trulia and Cabspotting, we're seeing products which are entirely of their medium – data visualisation – and it was a particular delight to share the excitement which Rodenbeck and other pioneers in this field are experiencing.

(Thanks also to Eric for putting up with his presentation being cropped by the projector. For further reading, there's a write-up of talk Stamen did in London recently, by Rod McLaren.)