MapBrief™

Geography · Economics · Visualization

The Elegant, Effective Election Data Visualization That Is Thankfully Not A Map

 

We occasionally field the suggestion, nay, insinuation, that our posture here at Mapbrief is unduly critical.  Whether it be pie charts, contiguous cartograms, or untucked shirts, our opinions are admittedly fixed. So much so that you might say “dogma”; we say “consistency”.

But in the spirit of staying positive, let’s admire this recent graphic from the Washington Post detailing campaign contributions:

 

 

Note the five dimensions of information: 1) donor name, 2) total donation,  3) state of Senate race, 4) receiving political party, and 5) relative size of state-level donation. As a bonus, the list is sorted top-to-bottom by total donation.

What makes me especially happy is that I can so clearly picture what this graphic could have been: a map of the US with the 11 states having some choropleth ramp (39 states of “no data”!), with a proportionally sized pie chart atop each state showing the party breakdown.

The horror.

Moral of the story: sometimes the best map is no map.

Great job by the Post team.

 

—Brian Timoney


 

 

Open—Wide Open—in Portland: A FOSS4G Review

 

Having been to previous FOSS4G conferences—2007 in Victoria, 2011 in Denver—I was eager to circle back to this year’s iteration in Oregon to assess what’s changed, what’s new, and what I can adapt for everyday use in my consulting business.

I. Sprawl

With up to eight tracks running simultaneously, everyone’s conference was going to be different. But knowing that all the sessions would be available on video, one could take more “chances” with the unusual-sounding and catch up later with the more familiar (PostGIS, MapServer,etc. ) later.  Be honest, how many conferences leave you eager to review another few hours of talks a week later?

II.  Open Source Is A Development Strategy, Not A Price Point

The biggest cultural change from 2011′s FOSS4G ?

GitHub.

We all know what open source is and don’t have to address the legacy stereotype of open source participants as an exotic tribe of code natives uncorrupted by the ordinary lures of the market economy.  Everyone has a GitHub account, from your favorite 3-letter acronym government agencies to your other favorite 4-letter geospatial vendor. With this new broad understanding of the open source model comes the opportunity to have a richer conversation about the efficacy, adoption, and sustainability of the tools–themes touched on by keynoter Mike Bostock.

III. Remember the Browser Wars?  The Browser Won.

By far the greatest impression was made by the WebGL and the manifold possibilities of powerful applications that run in the browser with no plugins.  The technology itself is a little trippy with shaders and meshes made up of points, lines, and triangles.  Except you actually want to render your lines as triangles too (don’t ask).

Life was easy when there were hard-and-fast distinctions between Server, Desktop, and Web.  Then along came the mobile web.

Now we have the dominance of the browser.

Database? It’s inside your browser. With spatial features.  That are spatially indexed.

Spatial operations?  Inside the browser. Hefty on-the-fly rendering?  Inside the browser.

Let me throw out three implications:

  • the “basemap” is dead—your whole map, base features and thematic overlays—is being rendered on-the-fly. If we were at the dawn of a new Golden Age of Cartography a couple of years ago, I suspect we ain’t seen nothing yet.
  • the web is no longer synonymous with “connected to the Internet”. The use-case that kept cropping up was the new user expectation that their web app would keep working when disconnected from the interwebs and seamlessly sync with a canonical datastore when connectivity was restored.
  • no install software, because the browser is the operating system.  Anyone whose business required the tender mercies of Enterprise IT to install software will appreciate a more direct connection to end users via the browser. It’s one of those nudges that amplifies open source software’s natural advantages in rapid deployment and fast iterations.

IV.  No Paint-by-Numbers Business Models

Just as diverse as the tools on offer were the ways of incorporating open source to make money.  While funded startups Mapbox and CartoDB certainly commanded a great deal of attention, the true breadth of geospatial open source was perhaps best seen towards the margins.  Cloudant, recently acquired by IBM, talked up its platform in the context of PouchDB and the burgeoning smart car market.  The Climate Corporation, largely fueled by open weather and agricultural data ( acquired last October by Monsanto for $1.1B—yes, that’s a ‘B’ ) presented a bit on better spatial indexing in the Lucene/Solr/ElasticSearch family of search technology. What’s clear is that far from stale cliche of open-source-as-hippie-altruism, FOSS4G encompasses a much richer variety of possible income models than the increasingly cramped GIS-software-and-services subset.

V.  Location, Location, Location

With the logistics of the event well organized from top-to-bottom, I’d be remiss if I didn’t compliment the host city itself.

Sized quite manageably for visitors, Portland is one of the few US cities you can show off proudly to your Progressive Urbanist friends: ubiquitous public transit, Dutch levels of citizen cycling, high taxes, and restrictive zoning (edited for politeness).

And the weather belied the cliched stereotype of the rainy northwest:  a week 80F temps and bright sunshine. Reminded me of San Diego.

 

—Brian Timoney


Bridge photo courtesy of   Aaron Hockley’s Flickr stream
Netscape photo courtesy of   Toshihiro Oimatsu’s Flickr stream
Paint-by-numbers photo courtesy of   VJ Beauchamp’s Flickr stream

 

 

Data Insight, Reputational Risk, and Bill Belichick: Sports Analytics & Business Analytics, Part II

 

Can everyone immediately cite your worst professional decision?

If you’re 3-time Super Bowl champion and coaching mastermind Bill Belichick, any moderately savvy NFL fan would blurt out “4th and 2“.

 

Go for it on 4th-and-2 on your own 29 yard line with 2:08 remaining or punt to the Colts and give Peyton Manning the opportunity to march down the field?  Given the Colts had only one timeout, a first down would almost guarantee victory.

Belichick went for it; they didn’t make it; the Colts march in from the short distance to win the game.

And by most accounts, the odds were in his favor.

Insight Is Not Cost-Free

So if widely respected championship coach can get crucified for playing the percentages in a non-traditional way (during a regular season game in November), why are you surprised when your precious analytic insights that have even a whiff of the counter-intuitive fail to gain traction?

Statistical reasoning is abstract; middle-management reputational risk is very real, very immediate.

We can all applaud the “democratization of data” and tools that empower analysis-for-everyone. But if that means the status quo is at risk, well then, that’s a whole different conversation.  A conversation that’s typically the province of well-dressed, well-paid consultants rather than, say, you the underling who is a bit brighter than the good folks you report to.

Or else it’s the rare don’t-give-a-damn outsider like “The Coach Who Never Punts“.

 

 

Probabilistic Thinking In A Binary World

The most entertaining side-show of the 2012 Presidential election was the dust-up between Nate Silver and TV host Joe Scarborough when the latter decried anyone who declared that the race was anything but a 50-50 proposition.  Silver, of course, was drawing massive page views at the New York Times with his daily calculations of election probabilities using a large dose of Bayesian methods.  The revelation was that putatively intelligent people such as Morning Joe much prefer thinking about outcomes in binary/yes-no/Manichean terms rather than as shifting set of probabilities.  Especially when it pertains to a topic in which they are emotionally invested.

So when pitching a middle-manager on workflow improvements that have  an 83% chance of success, you think you’re saying “pretty good odds” when in fact what they are hearing is “it might not work.

*  *  *  *  *  *

Moneyball” has become the fashionable pop culture short-hand for analytics-driven decision making: but recall even Brad Pitt and his good looks encountered institutional resistance.

Without Brad’s good looks and Bill’s Super Bowl rings, what are you to do?

Spend as much time analyzing your organization’s culture as its data.

 

—Brian Timoney