MapBrief™

End-of-year summaries and next-year predictions are the web’s way of helping you pass time during the most unproductive work week on the calendar. Or save you from continued contrived conversation among those with whom you share little except a similar genetic imprint. Rather than go the solipsistic blogger route and explain why The Decemberists put out the best album or that Incendies was my movie of the year, I’ve chosen a tack in which I’m more heavily invested. For “hotness” here refers not to PR buzz but tools that can solve both my clients’ current problems and their soon-to-be problems.

The next great GIS isn’t a “GIS”–but rather the statistical package R.  It’s the nexus where modelling, statistics, and graphics meet.  An open-source project with a large community and big developer momentum, there’s a critical mass of know-how such that you’d be hard-pressed to come up with a quantitative challenge that hasn’t already been tackled by the R community.  For mapping, the obvious starting point is the maptools package, but there are also hooks to familiar tools such as GDAL (RGDAL) and PostGIS, as well as the recently released GUI DeducerSpatial.

Let’s be clear, it isn’t about trying to replace your trusty GIS with statistical software. It’s about acquiring a more robust quantitative toolset to wrestle with a multi-variate world. Statistical clustering (spatial and non-spatial), principal components, multi-dimensional scaling, etc. will all be go-to techniques in a world that can no longer be explained by a single variable displayed on a map.  We all laugh at “red-dot fever” where lazy analysts overwhelm a map by displaying every coffeeshop, every bank, every whatever, creating visual confusion instead of anything approaching meaning. But the crashing of the tides of “Big Data”, the “sensor web”, and the “Internet of Things” upon our shores is imminent, and wrestling with those datasets with advanced statistical techniques will be the prerequisite for making meaningful maps.  A small taste of what’s possible in R is one of this year’s most compelling maps:  the Facebook map.

A lot of mapping shops will be scratching their collective heads this year figuring out how to serve a public that uses everything from Internet Explorer 6 to the iPad, as well as your preferred smartphone. With the mobile web leaving the worlds of Flash and Silverlight behind, where to turn for interactive vectors in the browser. Why not plunge into the future and go pure HTML5/SVG? Because the mapping community, with the large presence of government agencies at all levels has a disproportionately high use of older versions of Internet Explorer. Recent stats from ESRI suggest that visitors to ESRI.com use IE 6-8 at a rate roughly double that of the overall web user population.

Raphael is a javascript library that bridges the gap by rendering vectors natively as VML in Internet Explorer 6-8, and as SVG in the newer web browsers. Hence you get “live vectors”: rollovers, tool tips, click events, etc. without requiring plugins such as Silverlight or Flash.  Of course, rendering tens of thousands of vertices won’t go so well in older browsers, but for everyday thematic maps, such as this example of US States, it is perfectly serviceable. With 2012 being a Presidential election year in the US, expect many news organizations to move away from Flash to Raphael for their choropleth-ing of results.

Some of the most innovative work of 2011 came from Washington DC-based Development Seed. First, their open source TileMill cartographic studio is a much-needed tool that enables the making of visually compelling maps without GIS. This is a huge boon to the geospatial sector where too many of us, present company included, have been all too content to crank out always utilitarian, sometimes ugly, maps for our clients. Better still, the styling specifications use the CSS-inspired Carto language, making for easy re-use and sharing of styles.

And that’s not all. Using the very clever UTF grid approach, maps created with TileMill and served up via TileStream (also open source), feature interactivity that is also cross-browser–from Internet Explorer 6 to the iPad.  And because the interactivity is pixel-based, it can handle many, many thousands of features without killing your browser.

The coup de grace is the MapBox iPad app. All your great cartography created in TileMill with the full experience available in disconnected settings. In important ways, I see the iPad (and, hopefully, future tablets that can match the user experience) as a great second chance for mapping on the web. Because the first time around we as an industry failed our users by insisting on a desktop-GIS-inside-the-browser metaphor that was utterly foreign to anyone except fellow professionals. I’d love to see the default standard be a well-designed, informative basemap plus once “clickable” layer: enough information for 85% of your users without introducing the confusion of dozens of layers (tucked within layer “groups”!). And my observation tells me that there’s an intimacy users have with their iPad that is very different that their relationship to their desktop machines. Simple to use, informative, and aesthetically thoughtful is the big win here.

So the excitement about these new tools and capabilities would naturally lead to a conclusion that “it’s never been a more exciting time to be in geospatial.” But there’s also an underlying lesson in highlighting projects that don’t come out of the traditional group of GIS vendors: geospatial is attracting significant outside attention and people are getting things done using tools and methods that are unfamiliar to many industry veterans. Combine that with the contraction of the public sector that is a huge component of GIS employment, and the more sober conclusion is that our little niche traditionally off-to-the-side is more mainstream and much more competitive. So think about and spend time with some of these new technologies not for the “cool” factor, but to ensure the continued relevance of your skill set.

—Brian Timoney