Justin O’Beirne of the blog 41Latitude (which I hadn’t heard of before) has a great series of posts critiquing the user interface and cartographic styling of OpenStreetMap. It is an unfortunate truism that Free/Libre Open Source Software (FLOSS) projects tend to have poor user interfaces, so perhaps we should expect the same from OpenStreetMap?
It’s true that some of the problems O’Beirne observes could be due to flaws in the software that generates the maps. This is the cause of overlapping city labels and perhaps accounts for the differing apparent thicknesses of some country borders (if the linework isn’t sufficiently simplified for the given zoom level). Poor cartographic choices (a different sort of design problem) explain the excessive profusion of colors on the map, although some argue that this is due to importing cartographic conventions from Britain to North America. Looking at OpenStreetMap in London, however, it still seems like there are too many colors for major roads. Isn’t it a cognitive challenge to read green, orange and red lines all as the same grid of major streets? Rather than differences in coloring conventions, is this more of a cultural difference in what road maps are expected to look like? I note that Google Maps for Britain uses a similar variety of color, compared to the more restrained palette for North American cities.
Most of O’Beirne’s critique, however, stems from inconsistencies in what appears on the map and at which zoom level. This is due much more to the inherently distributed nature of participation in OpenStreetMap (although some may be caused by inconsistency in the original TIGER data). First, groups working in different parts of the United States may have developed different tagging standards, a problem which will likely resolve itself at some point as tagging practices diffuse across the map and disputes are settled on the OSM mailing lists and on the wiki. Second, there isn’t an even distribution of effort across the country. O’Beirne notices a few small towns that appear more prominently than neighboring larger cities, and he suggests this is due to OSM users or vandals (what is the difference, exactly, in this case?) inappropriately promoting their hometowns at the expense of other cities. While this is possible, it seems more likely to me that this is the natural result of users starting to improve the map near their hometown and working outward. Many towns haven’t (yet) found their local volunteer(s) to take on responsibility for cleaning up a particular patch of turf. The problem of spatially varying map quality is one that will continue to be a problem into the foreseeable future. Which is not to say that some areas will always have “bad” data, but rather that as the bad areas improve, the “good” areas will also continue to improve in accuracy and detail. New techniques for rendering crowdsourced map data will need to be developed that can gracefully handle differences in the level of detail present in the database. Exactly what these techniques will look like, I have no idea.
Overall, though, O’Beirne’s observations are spot on, and OpenStreetMap needs more of this constructive criticism. It also needs more volunteer efforts from people with graphic design and UI experience. That said, OpenStreetMap’s default “Mapnik” style is amazingly good, all things considered, and will likely continue to improve, as will the variety of other rendering styles available to choose from (the cycle map being the best example so far).