The year-end issue, like all of our special issues in general and 2010’s year-end specifically, was cruel to the graphics desk. 7-day workweeks. Sleep dep. Spiders living in my neckbeard. But it turned out nice, so I’ll walk you through the process of getting my charts on page.
Detroit
The Detroit map, showing 2011’s sparks of hope in everyone’s favorite decayfest, started out hideously complicated. Roads, census tracts, places of interest, shorelines, yadda-yadda.
The hatching on each census tract was supposed to show two factors, with three levels of severity each: right-sloping hashes for population loss (bad, worse, worst) and left-sloping for percent of people living in poverty (bad, worse, oh jaysus). The more thick lines crammed into a tract, the worse things are on both fronts. Seemed like a good idea to me, but what a mess.
Anyway I didn’t get my idea cuted up in time, so it ended up like this. With no relation to my work. Worked out for the best.
Volatility
Next was my graphic on stock market volatility, which was severe as heck in the third quarter. I did my data harvest off the Bloomberg terminal, then started on a two-tier graphic.
Needed to tie the two charts together somehow. So Jennifer got hold of it, beefed up the quarter markers and heds, and I added the pink bars to try and force the eye to associate between the two chunks.
It’s on pink now. And here comes the moire! That train’s never late. I was trying to get Q3 to pop out, with mixed results as you can see. I also started in on the desaturated/saturated look.
Fine, I’ll quiet it down. Pink’s more soothing than a bunch of zags. I tried to link the callouts to their proper place in the timeline via fat white bars. The gesture jazzed me, but didn’t last long.
After losing the highlighting bars entirely, we’re down to just a big white splotch to get you looking at Q3. The little pink numbers replaced the bars, and we’re off to the presses. I miss my bars though…
Data Leaks
Jennifer put me in touch with the erudite Jake Kouns of the Open Security Foundation, which runs the popular DataLossDB. They track data thefts and leaks through news sources, company announcements, and FOIA requests. Lovely people. I wanted to plot an entire year’s worth of data loss incidents, showing the type, date, and number of personal records lost (CC numbers, SSNs, home addresses, etc.).
At first I tried the trusty radial thingy, which we used in our kickstarter graphic to great effect; Kenton Powell was nice enough to gin it up in Processing, so I fed it my csv and got this spiky thing. Problem is there were some huuuuge outliers (70 million records stolen in one day!) that ran off the page and into the next room, so the spikes had to go. Unless I could plot these on a log scale (AKA log-jammin’), but that was kiboshed after it was made clear that I’m alone in liking that method.
So I changed the method to my specialty: gross abuse of the pie chart tool (it scales by area! Thanks for doing something right, Adobe.) After a lot of copying, pasting, and some -22 errors, I got a year’s worth of scaled-circle incidents on page. January up top, December at the bottom, each blip is one incident, scaled by number of records stolen. Lookin’ good.
Put it all together on page along with callouts detailing specific incidents, plus some choice stats and a tiny useless treemap.
Jennifer takes the wheel again, conforms the text boxes and gives the page some spark. Getting closer…
Another art director takes over, and we’re smoothed out again. Treemap’s gone. I should have seen that coming. I think it turned out nice, was fun to do all the reporting as well.
Corrigendum
Did you know Endal, the world’s most decorated dog, looked a lot like Ed Harris? Think about it.
Thanks to: Kenton Powell, Jennifer Daniel, Chandra Illick, Tracy Ma, Robert Vargas. The 2012 year-end issue can’t come soon enough. For me to hang myself.
—Evan
