I get asked all the time about the history of OpenStreetMap – the wiki-style map of the world that I have been contributing to for 10 years now. An interesting way to show how far the project has come in those 10 years is to view the map from then and now side by side. But the internet forgets quickly and those old maps are not easily available. So I built a viewer that lets you see for yourself.

osmthenandnow

Some background.

Getting a hold of old OSM data is not hard. OSM keeps a good archive of ‘planet files’. These are copies of the OSM map database from a specific point in time. To go back in time before 2012 we need to look in the cc-by-sa folder. CC-BY-SA is the Creative Commons license OSM data was distributed under until some time in 2012.

I chose to use 2007 data for a number reasons. The most important one is that 2007 represents an inflection point of sorts. It is when we are starting to see bigger data flowing into OSM. AND donated their road network data that year, and mappers in the US started tinkering with Census TIGER road data. Both lead to large scale data imports, disrupting the organic, human driven growth of OSM. (The debate around data imports lingers on into the present time.) The data in the Then and Now tool reflects some of these profound developements. Shown below: rich, imported data in the Netherlands on the left, much sparser data in Germany to the right. (If you visit this location in the tool, you will notice that the Germans have more than caught up.)

importvscraft

Similar contrasts can be observed in parts of the United States, where experiments with TIGER imports had started in some area. Below, Michigan.

michigan

The second reason for choosing 2007 for the historical map is much more mundane. I started contributing to OSM in mid-2007, and some of my early contributions show up on the map.

The third reason is more technical. OSM went through a series of fairly rapid technical developments in the early years. The OSM data model and API evolved between 2004 and 2007, when it started to settle. Choosing data from October 2007 greatly reduced the complexity of setting up a map server. All of the tools and processes used to produce the visible map from the raw data have evolved significantly. Data from earlier in time is exceedingly hard to process with current tools.