Just pushed out an update to the Adobe Illustrator MultiExporter script that lets you specify if you want to export PNGs and JPGs at a different scale factor so that you can generate versions of the images at double resolution for iPhone retina displays.

The option is set in the “Scaling” field in MultiExporter dialog box: Leave it at “100%” for  normal exports, set it to “200%” for double-resolution exports. You’ll probably also want to set a prefix or suffix or the file name, too, so that you can keep the files separated from the normal resolution ones.

I don’t think the update should have broken anything else, but let me know if you run into problems. (And to read more about how the script works, here’s more information about it.)

The script is available here — MultiExporter.jsx — and also in my Illustrator Scripts repository on github.

The Society for News Design’s annual contest is one of the handful of design and infographic contests that we enter each year. They’ve got a database online that lets you search the winners, but while it’s great for looking up your entries to see if you’ve won, it’s a little harder to get the a sense of the big picture — for example, which papers are great at features page design, which ones excel at news page design and which ones win for their photography or information graphics.

So, I figured there must be a better way to display the data that lets get a sense of how different publications do in the different categories. So, a little bit of Ruby and Javascript later, I hacked together an interactive crosstab of SND winners. Check it out. (Modern web browsers only, please)

I’ve posted the slides from my presentation as a PDF. Links to the interactives I mentioned are below.

• The Unemployment Rate for People Like You
• Obama’s Budget Proposal
• Hazards of Storing Spent Fuel
• Tracking the Oil Spill in the Gulf
• One 9/11 Tally: $3.3 Trillion
• Steve Jobs’s Patents
• A Peek Into Netflix Queues
• How Mariano Rivera Dominates Hitters
• House Election Results Map
• House Election Big Board
• House Races: What to Watch, Hour by Hour
• Layers of Ownership
• What’s In Your Water?
• An Olympic Musical
• Budget Puzzle: You Fix the Budget

Often, when you get data that is organized by geography — say, for example, food stamp rates in every county, high school graduation rates in every state, election results in every House district, racial and ethnic distributions in each census tract — the impulse is since the data CAN be mapped, the best way to present the data MUST be a map. You plug the data into ArcView, join it up with a shapefile, export to Illustrator, clean up the styles and voilà! Instant graphic ready to be published.

And in many cases, that’s the right call.

For example, census maps of where whites, blacks, Hispanics and Asians live in New York City show clear geographic patterns, answering questions like “What areas of the city are more segregated?” or “Where is there more diversity?” You can see how Prospect Park is a stark dividing line, with largely white areas west of it, and largely black areas east of it. You can see the how far the Asian population stretches in the area around Chinatown in Lower Manhattan. And you can see the census tracts where there’s greater intermingling of different groups.

View full interactive map on nytimes.com »

Maps also a terrific way to let readers look up information about specific places. On election night, they answer questions like like “Which seats did the Republicans gain?” or “Who won all the seats in Oregon?” or “Who won my Congressional district?” You don’t have to remember the number of the House district you live in — you can just look at the map, zero in on the area that you’re interested in, and see if it’s shaded red or blue.

View full interactive map on nytimes.com »

And obviously, when the story is completely based on the geography — “How far has the oil spill in the Gulf spread?” — there’s nothing more effective than a map showing just that.

But sometimes the reflexive impulse to map the data can make you forget that showing the data in another form might answer other — and sometimes more important — questions.

So, when should you use a form other than than a map?

1. When the interesting patterns aren’t geographic patterns

On election night, people don’t want to know just “Who’s winning my district?,” but also “Which party is doing better than expected tonight?” And the results map, as good as it is at answering the first question, has a much harder time answering the second.

The most important trends on a House results map don’t correspond to clear geographic patterns. Sure, there’s broad trends in that there are more Democrats in the Northeast and more Republicans in the West. But those differences often pale in comparison to the differences between an urban district, a suburban district and a rural district that are all next to each other in the same state.

One may have been Democratic for years, another may have been held by a long-time Republican but is now up for grabs after a retirement, and the third may be a perpetually close swing seat. But none of that context is visible on the map. So, unless you’re a true political nerd, it’s hard to know whether a splotch of red appearing in a single district in Minnesota is a surprising result or not.

So, when we set out to design our election results package in 2008, we wanted to see if there was an additional way to organize the results to help readers see not just who was winning each seat, but also what was surprising about the results that night.

Each election, our reporters handicap the all the Congressional races, classifying them on a five category scale that ranges from Strong Democrat to toss-up to Strong Republican. We use those rankings to organize results on a page that we call the Big Board, which divides the districts into a five-column table based on those categories. The Strong Democratic seats are in a column on the left, the Strong Republican seats are on the right, with the leaning and toss-ups seats in-between. So, as results come in throughout the night, you should expect to see mostly blue on the left, mostly red on the right, and a mix of colors in the middle.

View full Big Board on nytimes.com »

But if you see red creeping onto the left side of the page, or blue onto the right, it’s a visual indication that one party is doing better than expected. If you look at the Big Board from 2010, when Republicans gained control of the House, that’s exactly what you see. There’s no Democratic blue in the rightmost two columns, but you see a number of seats being flipped Republican red on the left side of the page, and twice as much red as blue in the center toss-up column.

And if you take and scroll down the page, about halfway through the states, you’ll see a clear signal of one of the biggest upsets of the night. The single red square popping out of the column of blue Strong Democratic seats indicates the Republican upset in Minn. 8, where Jim Oberstar, a Democrat who had been in the House since the 1970s, was defeated by newcomer Chip Cravaack.

View full Big Board on nytimes.com »

In a similar vein, a graphic by Alicia Parlapiano and Amanda Cox that accompanies an article about how the Obama campaign views the 2012 election map does the same thing, eschewing a map for a visualization form that is a better fit for the story.

The article explains that while the conventional wisdom is that Ohio and other industrial states are the keys to a Democratic presidential victory, the Obama campaign thinks it may be able to win without those if it can capture states like Colorado and Virginia, which have a greater share of educated and higher-income voters.

Normally, the first thought upon hearing of a story about the electoral map is that you should make a map to go with it. (And in this case itself, the story was even slugged “MAP” in our content management system, reinforcing that notion.)

But the reason that the Obama campaign was focusing on those states was not because of their geographic location, which is what a map is most effective at showing, but rather because of their voter demographics. So Alicia and Amanda thought that there should be a better way than a map to show why those states were being targeted.

View interactive graphic on nytimes.com »

The result: a scatterplot that compares the percentage of people in the state with college degrees to the margin of victory for Obama or McCain in the last election. The chart clearly shows how those two states are well ahead of others in the education level of their residents, and how the education levels correspond with other states won by Obama and McCain.

2. When the geographic data is more effective for analysis

A second case where it’s more effective to use a form other than a map is when you should be analyzing the geographic data, rather than just displaying it.

After Hurricane Katrina struck New Orleans back in 2005, we produced dozens of graphics in the following weeks and months, including many maps that showed things like how the levee system encircled the city, the extent of the flooding, the demographics of New Orleans and the locations of damaged buildings.

But one of my favorite graphics — or actually, part of a graphic, as you’ll see below — was generated in ArcView, a mapping program, but didn’t actually result in a map.

A couple weeks after the storm struck New Orleans, one of the questions that we wanted to answer was “Who in New Orleans had been most severly affected by the flooding?”

At this point we’d collected a ton of data, including detailed images from the Dartmouth Flood Observatory that showed the areas of the city that were still flooded on Sept. 2, several days after the pumps and levees had failed. Using that image as a base, I traced the flooded areas in ArcMap to create a shapefile that showed the maximum extent of the flooding.

I’d also pulled demographic data from the 2000 Census for every block group in the city, including racial demographics, median household income, poverty rates and educational attainment.

So, at this point, you could take that flood extent and overlay it on different demographic maps to show the characteristics of the flooded areas and the dry areas. Here, the flood extent is overlaid with a map showing the median income in each census block group.

But while maps like that are interesting to look at, it also forces readers who want to figure out the correlation between income and flooding to try and visually sum up all the colors on the map in their head. The map shows there’s low-income areas in the flooded areas and there’s also low-income areas outside the flooded areas. There’s middle- and upper-income areas in each, too. Unless the pattern is super clearcut, trying to figure out how much of a relationship exists is a tricky task.

So, instead, I used ArcView to select all the block groups that fell inside the flooded areas and calculate aggregate statistics for the area as a whole. Repeat the process for the block groups that hadn’t been flooded, and you have data for a simple table that clearly and effectively shows the difference between the two areas.

More than twice as many people lived in the significantly flooded areas, and they were more likely to be black, have more children, be less educated and earn less money than those in the rest of the city. There were more families with single parents and they were less likely to own a car. And there were twice as many housing units in the flooded areas, but the homes were newer and more likely to occupied by the homeowner.

View full graphic on nytimes.com »

In the end, the table ended up getting paired with a large map of the city that showed the flooded areas and the locations of police stations, schools, hospitals and courthouses, and described the neighborhoods in the city.

But while the table was a much smaller portion of the graphic in terms of physical space, I think it was the far more effective portion when it came to helping readers understand who in New Orleans had been most affected by the storm.

I’ve posted the slides from my presentation as a PDF. The interactives for each section of the presentation are below.

Explain the Geography

• Oil Spill Tracker

Reveal Patterns

• Taxi Patterns in New York
• Danger of Digging Deeper

Provide Context

• Evacuation Zones
• Radiation Levels
• Japan Before/After

Describe Processes

• Flight 1549

Report and Research

• Sea Ice in Retreat
• Food Stamps
• Egypt Protests
• Tahrir Square

What’s the Story?

• Mariano Rivera
• What’s in Your Water?

Sketch with Data

• Netflix Rentals

Experiment with Forms

• Election Results
• 2008 Olympic Medal Count
• Bill Rankin’s Chicago
• Eric Fischer’s Map
• ACS Map

When Maps Shouldn’t Be Maps

• House Election Results Map
• House Big Board

Data + Story > Story

• House: What to Watch
• Layers of Ownership

The Information is What’s Important

• Election Results Map
• Haiti Before/After
• Joplin Before/After
• Olympic Symphony

Engage the User

• Budget Puzzle
• Census Maps: Tweets

So, a couple train rides later, I’ve finally combined the two scripts into a single one with a couple more added features: you can choose to export as PDFs in addition to PNGs, and you can also export the product of artboards and layers.

To use the script, download MultiExporter.jsx and put in your Illustrator scripts folder (usually in Applications/Adobe Illustrator/Presets/Scripts/). Restart Illustrator, and run the script by going to “File > Scripts > MultiExporter”.

A few notes:

• You can choose whether you want to export all the artboards in the document with the currently visible layers showing, or if you want to export files for each of the layers in a document on the currently active artboard, or if you want to export a combination of all the artboards multiplied by all the layers.
• Files are named based on their layer name. It only exports layers where their name has been changed from the default “Layer 1″, “Layer 2″ or “Artboard 1″, “Artboard 2″, etc.
• If you put a minus sign (-) in front of a layer name or artboard name, it will skip that layer or artboard. (Useful for when you no longer decide you like a particular mockup, but don’t want to delete it from the master Illustrator document.)
• For layers only: If you put a plus sign (+) in front of a layer name, that layer will always be visible. Useful for if you want to have a layer that always appears in the background of each exported image.
• It stores its settings in a nonvisible, nonprinting layer named “nyt_exporter_info”
• It has an option for transparency, and lets you choose between PNG8, PNG24 and PDF.

You can try it out on the sample file artboard-layer-test.ai.

The sample file contains artwork for buttons on three artboards — “Red”, “Blue” and “-Green” — and has seven layers.

• If you run the script and choose “Export Artboards (with currently visible layers)” it will generate two files: Red.png and Blue.png that include whatever layers are currently showing at the time you run the script. (It skips the “-Green” artboard because the artboard name begins with a minus sign.
• If you chose “Export Layers (on currently selected artboard)” it will generate two files: Play.png and Pause.png. The “+Button” and “+Gradient” layers will appear in the background of both exported files, and it will skip the “-Pause 2″ and “Layer 5″ layers.
• And if you choose the “Artboard + Layers”, it will do a combination of the two other options and generate four files: Red-Play.png, Red-Pause.png, Blue-Play.png and Blue-Pause.png.

Try it out, and leave a note in the comments if you have any problems or suggestions.

So, I wrap HTTP requests in a tiny little class that saves the responses to the file system, so if you request the same URL again, it will load the cached data, eliminating the need for an HTTP request:

class HTTPCacher 
  def initialize( base_dir )
    @base_dir = base_dir
  end

  def get( url, key )
    
    cached_path = @base_dir + '/' + key
    if File.exists?( cached_path  )
      puts "Getting file #{key} from cache"
      return IO.read( cached_path )
    else
      puts "Getting file #{key} from URL #{url}"
      resp = Net::HTTP.get_response(URI.parse(url))
      data = resp.body  
      
      File.open( cached_path, 'w' ) do |f|
        f.puts data
      end
    
      return data
    end

  end
end

Usage is pretty simple. Create a new HTTPCacher object

getter = HTTPCacher.new( '/path/to/data/dir/here' )

and then make a get request, passing two parameters: 1. a URL, and 2. the key that you want to cache it under. Any further requests with that cache key will load the file straight from the filesystem.

data = getter.get( 'http://otter.topsy.com/search.json?q=ipad&window=auto', 'ipad.html' )

Note that making sure your keys are unique between URLs is entirely up to you. If you try to request two different URLs but pass the same key, it won’t be able to tell them apart and it will return the cached data on the second request.

What Happens in a Nuclear Meltdown

The operating reactors at Fukushima Daiichi power station automatically shut down during the earthquake. But after cooling failures, two of them went into partial meltdown.

Provide Context

Satellite Photos - Japan Before and After Tsunami

Compare satellite images of areas of Japan before and after the disaster.

Assessing the Radiation Danger, Near and Far

Assessments of the radiation from the Fukushima Daiichi nuclear power plant by the Japanese authorities, the International Atomic Energy Agency and others.

Map of Radiation Readings Near Fukushima Daiichi

The Jobless Rate for People Like You - Interactive Graphic - NYTimes.com

Not all groups have felt the recession equally.

Obama’s 2012 Budget Proposal: How It’s Spent - NYTimes.com

Explore every nook and cranny of President Obama's budget proposal.

Reveal Patterns

Interactive Map Showing Immigration Data Since 1880

See how foreign-born groups settled in your area and across the United States from 1880 to 2000.

The Shifts in the Map - How Obama Won

Find breaking news, multimedia, reviews & opinion on Washington, business, sports, movies, travel, books, jobs, education, real estate, cars & more.

The Danger of Digging Deeper

A project financed by the Energy Department aims to capture geothermal energy from hot bedrock — a process that can cause earthquakes.

Describe Processes

What Happens in a Nuclear Meltdown

The operating reactors at Fukushima Daiichi power station automatically shut down during the earthquake. But after cooling failures, two of them went into partial meltdown.

Dissecting a Dance

Alastair Macaulay, The Times’s chief dance critic, analyzes Merce Cunningham’s “Biped.”

Explain the Geography

Map of the Damage From the Japanese Earthquake

Map and Estimates of the Oil Spill in the Gulf of Mexico - Interactive Map - NYTimes.com

The spreading slick, day by day, and a chart of how much oil has been spilled.

Report and Research

Virginia Tech Shooting - New York Times

A recounting of the events on the day of the deadliest shooting rampage in American history. From official and survivor accounts of the shooting at Virginia Tech.

Map of Evacuation Zones Around Japan Nuclear Plant - Interactive Feature - NYTimes.com

Map of the evacuations zones recommended by the American Embassy in Tokyo and Japanese officials.

What's the Story?

How Mariano Rivera Dominates Hitters

The closer has confounded hitters with mostly one pitch: his signature cutter.

House Map - Election Results 2010

House Big Board - Election Results 2010

Iowa City Water Department Water System - Interactive Database - The New York Times

Sketch with Data

A Peek Into Netflix Queues - Interactive Graphic - NYTimes.com

Story+Data > Data

Preview of Key House Races - Election Results 2010

Inaugural Words - 1789 to the Present

Turning a Corner?

Layers of Ownership

The Information Is What's Important

New York, NY 10017 Today's Weather Forecast - AccuWeather.com

weather.ericson.net: New York, NY Weather

Satellite Photos - Japan Before and After Tsunami

Compare satellite images of areas of Japan before and after the disaster.

Map of Radiation Readings Near Fukushima Daiichi

Fractions of a Second: An Olympic Musical - Interactive Graphic - NYTimes.com

Engage the User

Budget Puzzle: You Fix the Budget

How Your Views Compare With the Court

Mapping America — Census Bureau 2005-9 American Community Survey

Tools

Google Fusion Tables Example from The Guardian

Tableau Public: Power Hitters

Google Charts from New York State Test Scores - The New York Times

HTML, CSS and Javascript: 2010 World Cup Rankings

jQuery: The Write Less, Do More, JavaScript Library

jQuery UI - Home

Protovis

Raphaël—JavaScript Library

The R Project for Statistical Computing

Processing.org