Archive for April, 2013

Map Critique, How to Avoid Bad Advice

While I often espouse the need for critique in this blog, I don’t usually go into detail on how to conduct a successful critique of your work. First, how do we determine if a critique has been successful? In simple terms, critical feedback must make the map better.

And, the more people you have critiquing the work, up to a certain point, the better. Something like this:

But you also want to make sure that the critique does no harm. You could be on the receiving end of bad ideas as well as good ideas. How do you identify them so they can be eliminated?

Eliminating errors is especially difficult when you already have a favorable opinion of the critic’s ideas and/or the critic has already given you an idea you consider to be quite good. This is the halo effect.

The halo effect applied to a cartography critique goes like this: (1) If a known expert, perhaps your cartography professor, gives you a bad idea, you are likely to think that it is a good idea. (2) If an unknown critiques your work and you consider her first idea to be really good, then you are much more likely to think her second idea is really good irrespective of its merit.

To prevent the halo effect from biasing the results of your critique, you want to decorrelate error. Decorrelating error is achieved by getting a large amount of critical feedback from many people. This will tend to move you toward a good design because the good ideas will be suggested by multiple critics while the bad ideas will be outliers, on an idea vs. quantity chart.

This only works if your critics are giving you feedback independent of one another! If, like in a typical architecture critique, the critics are grouped together in a room, shouting out their likes and dislikes, they will most certainly influence each other and you will not have uncorrelated errors.

In summary: For the best map design, get as many critics as you can, and ask them for their feedback independently of the others.*

*A good description of the halo effect and decorrelating errors is found in Daniel Kahneman’s “Thinking Fast and Slow.”

No Comments

The Watercolor Illusion

*Note: From time to time old posts will be resurfaced on the blog. This one is from Sept. 2010. The watercolor illusion is similar to, but not exactly the same as, the vignette concept. A vignette would create a subtle illusion as well. A simple example would be this banding effect created in TileMill:

I was digging around the cartography literature yesterday and came across something called the watercolor illusion.*

The illusion is thus: if you have a dark line (the more squiggly the better) next to a light colored line, your eyes will fill in the missing white space with the lighter color, albeit in a lighter tone, and give a washed-out effect. In this example your eye may think that the outside rectangle is green but in fact it is white. It is just the light green fine line next to the purple that is causing that illusion:

Perhaps you all have some better examples of how this plays out on a map, but here is my attempt. You can see I am trying to create a major definition between the water and the land (these are geology polygons). The geology polygons are in dark purple and I’ve used a bright, light blue next to the outer edge by first creating a buffer of the polygon and then symbolizing the buffer:

There are other ways to produce a gradual fading of a line in your GIS and in your graphics programs, of course. In ArcGIS, for example,  you can use a gradient fill, though sometimes, especially with large datasets, this taxes the renderer quite a bit. However, this is a fairly easy and handy way to create the effect within the GIS and without exporting to graphics software.

*Pinna, Baingio and Gavino Mariotti, “Old Maps and the Watercolor Illusion: Cartography, Vision Science and Figure-Ground Segregation Principles,” Systemics of Emergence: Research and Development. 2006, 3, 261-278.

No Comments

Workflows Past, Present, and Future

As recently as 3 years ago, these were the dominant cartography-software workflows, as garnered from a survey of CartoTalk posts:

With the exception of a few of these products, many cartographers still use one or more of those paths on a daily basis.

Might the following compilation of the most-tagged terms on GIS Stack Exchange give us a glimpse into what the most common workflows of the future will be?* The number of references is listed on the right.

* Two caveats: (1) QGIS uses stackexchange as their primary forum for questions and answers (but that doesn’t preclude its applicability). (2) There are quite a few new tools out there that people are having success with that aren’t in the top tags.


Five Common Map Afflictions, Causes, and Solutions

(1) Map Measles: Too many points at too low of a zoom level. Caused by lack of zoom-level specific coding and/or lack of simplification pre-processing. The solution is to pre-process the data to rid it of the least important points, or to on-the-fly select by attribute (e.g., “population > 500,000”), or on-the-fly binning (e.g., hexagonal binning). Rx: TileMill’s Styling For Zoom Levels tutorial, Hexbin discussion on indiemaps.

(2) Blotchy Skin: Jarring colors interrupt an otherwise flawless complexion. Caused by lack of pre-determined color palette. The solution is to choose a color palette prior to beginning the project and adjust it as needed. Rx: Color Brewer, Cartographer’s Toolkit, Colour Lovers, your own art collection.

(3) Abnormal Cell Walls: Differentiation of polygonal features via varying polygon line-widths and colors, cataclysmically colliding at shared boundaries. Caused by rapid output needs and resulting sloppiness. The solution is to use inner-buffering to provide a continuous color for each polygon while allowing for neighboring polygons to have a different color. Rx: view how it’s done, buffer with QGIS using negative distance.

(4) ADHD: No clear hierarchy of elements, features jumbled in an inappropriate order. Caused by misuse or ignorance of established standards. Also caused by human error and not acquiring critical feedback from outside sources prior to publication. The solution is to be aware of appropriate layer order (e.g., state borders under interstates, park shading under local roads). Rx: study Google Maps, seek critique.

(5) Over Accessorizing: Putting everything on the map that can be put on the map. Caused by GIS analysts, who have a multitude of information at their fingertips, wanting to squeeze every potential drop of wisdom possible out of the data by presenting it all at the same time, resulting in less information transfer rather than more. Also caused by meddling bosses. The solution is to drill down to only the salient details, to separate data into comprehensible segments, to present different maps in different tiles, or to put in a huge effort in getting the data massaged together (e.g., road atlas). Rx: study road atlases, NYTimes map graphics, MIT Senseable City Lab products, winning maps from various map competitions (e.g., NACIS 2012 narrative, first place GISCI 2012).


Map Misery

About 11 years ago I had my first child. There was a point during the birth that things went horribly wrong and the only thing I could do or think was to hope that someone would put me out of my misery.

Your goal is to not have people want to be put out of their misery when they look at your map.


Writing and Mapping

When people write articles for ArcNews or other similar publications you can tell that, for the most part, they put 100% into the written word and maybe 20% into the graphics, usually maps, that go along with the article.

I was just talking to someone who does a lot of expert witness work, and he mentioned the oft-heard statement that a picture is worth a thousand words. He’ll create a lengthy document explaining the specific facts in a case and what does everyone home in on? The graphics.

Maps are so important to the conveyance of understanding. They need to be thought out with the same care and perfection that we strive to achieve with the written word.

1 Comment