Archive for the 'Volunteered' Category


A Dollar per Square Meter

Today’s effort comes to my attention via a reader, Nicholas, who wrote to me early last year and suggested that I have a look at the following piece:

Click to enlarge

Click to enlarge

It was created by Chris Dickersin-Prokopp, who posted it on his blog and on Greater Greater Washington (where it got most of its comments) in February 2013. If you read the comments, you can see that, unfortunately, the response was not quite as positive he had hoped for. Not only did Mr. Dickersin-Prokopp accept the criticism gracefully, but he said he hoped that perhaps he’d find his way onto the pages of this very blog. So, let’s have a look.

To me, the real issue here is not in the map design itself, but instead the disconnect between the author and his readers. It’s an issue in any sort of creative field: you make something, and it turns out that the people who view that something were looking for something else. Mr. Dickersin-Prokopp’s detractors seemed to be looking for details on the map:

“Wow. I find it hard to think of a worse way to visualize these data. How am I supposed to draw any conclusions from a picture showing thousands of overlapping circles?”

“I can’t tell how many overlapping circle I’m looking at in different areas. Why not just make each sale a dot and color code the dots for different sales price ranges?”

But the author, I speculate, was trying to paint a broad picture: where, generally, home sales are more concentrated and more expensive. Counting individual circles isn’t necessary here. You just need to be able to tell where there are roughly more or roughly less. I think that his map does this job just fine, giving you the 10-second quick understanding of the basic pattern. His readers, however, were looking for something else: a map that let them break these patterns down in more detail, and count the number of houses in neighborhoods (everyone always wants to know what’s going on in their neighborhood/city/state when they see a map). The map isn’t ready to do that, as is, and the second quote above suggests a reasonable alternative. But one that, probably, wouldn’t have done as good a job at telling the quick broad story.

Two different ways of reading the same set of data: the big pattern, and the individual location data. The author had one in mind, but some of his commenters had another. What to do? Well, there may well have been a good symbology to satisfy both. But, setting that aside, how do we get people on the same page as us if we’re making maps? I think this map could use a little more context, to steer people’s minds toward reading it as the author wishes. Annotations would be handy—highlight some region or other and say, “over here, we can see a dense area of inexpensive home purchases that happened because of Reason X.” Get people thinking about the patterns and pull them away from the details that the author is less focused on.

The other option is to just change the map in response to comments, trying to give the audience what they’re looking for. It’s tough, as a creator, to have to give up on part of your vision and preferences, but it’s necessary if you want anyone to look at your creation. If no one likes it, then it doesn’t matter how great you think it is: no one is looking. Of course the hard part about that is that while many don’t agree with your choices, they don’t agree with each other, either. Some folks liked Mr. Dickersin-Prokopp’s effort:

“The message the map intends to convey is clear and obvious.”

If he changes things, he may displease some folks who were previously happy. There’s not much to be done about that, except to accept that you can’t please everyone.

Bigger picture aside, there are some other interesting things to dissect here, starting with the legend. which states, “Circle size represents sale price. One square meter equals one dollar.” If I interpret this correctly, this means that, in order to divine the sale price of a home you must determine how large a circle when compared against the map scale. That is, you pretend the circle is a real area of land in DC, measure the size of that land, and then convert to dollars. This is quite a fascinatingly weird way of explaining the symbology, and I’ve never seen anything like it. For good reason: it’s difficult enough so as to be largely useless. If you want people to understand what symbols mean, show the symbols; a legend should show more than it should tell. It would be better if this particular map showed people some sample circles and the values they represent. Then readers can reasonably eyeball things.

Typical proportional symbol legend designs. Via Esri Mapping Center.

Typical proportional symbol legend designs. Via Esri Mapping Center.

It’s highly unlikely that anyone is planning on measuring the exact price of each home (and if they do, they deserve what they get). We just need to let them get a rough idea of the pattern, and the circles above do that easily, without requiring people to visually measure square meters.

Speaking of which, that brings us to the scale. The increments could be a lot more helpful. At the very least, they should be in kilometers, but miles would be likely much more intelligible to an US audience. And the scale bar probably doesn’t need to be as wide as DC itself. No one is likely to do any measurements with it. To that end, I’d argue it could be dropped altogether. It is an unnecessary map element.

One Nice Thing: Putting your work out there for criticism and comment is tough, especially if you’re opening it up to the harsh gaze of the random Internet visitor. Mr. Dickersin-Prokopp handled the experience gracefully, and listened to the feedback he received, where I’ve seen other people in similar situations try to rationalize away criticism, or strike back angrily. Kudos to him.