Interview with Dair Grant - OpenStreetmap vs Google Maps

OSM’s map of Haywards Heath was literally just two main roads, so I thought it’d be a good opportunity to do a “first principles” look at what OSM involved and the kind of results it produced.

The first step was to build the map, after which I planned to go back and see how many mistakes I’d made (quite a few!) as well as measure how much effort it took to correct them.

After that, I thought it’d be useful to do a street-level comparison with another map to give some context to the OSM results.

Although comparing maps can open you up to charges of copying data, I’m absolutely behind the OSM model whereby if you ever upload data from a tainted source then your account and all its contributions are simply erased.

In a previous life I ported Windows games over to the Mac, so I am completely paranoid regarding IP ownership - if you’re responsible for source code that someone else spent a lot of time and money to create, you have to be!

Once I had completed the initial pass, I forced myself to go back with a printed OSM map to test just how accurate a single pass actually was. I honestly felt it would be a waste of time, as the map did look pretty complete.

However I kept finding the same kind of errors I subsequently noticed on Google Maps - streets would be in the wrong place, bad GPS reception meant I drew the wrong shape, names were swapped, stubs missed, etc.

The second thing that surprised me was that, apart from those minor errors, there weren’t really any major flaws.

The road network has pretty much the same shape as every other map, so a bike and a consumer-level GPS were easily accurate enough to capture it.

Fixing errors was also easier than I expected, although I think JOSM/Potlatch are probably still too technical for genuine “end user” changes.

There are other towns nearby which aren’t so well mapped, so there’s enough to keep me busy for a while.

I am also interested in the “completeness” aspect of OSM, as I think that’s one of the challenges the project faces. All maps are an approximation, but at what point do you say part of OSM is “done”?

I’ve no idea - but it would be nice to think that someone will find it useful in the future.

For example, there’s a large town map at the train station which a commercial cartographer supplied to the council. That map states it was derived from out of copyright material and direct surveys, and it has a couple of errors relative to the OSM map.

Given sufficient resources, you can of course map anything.

Although OSM doesn’t currently capture all the things commercial data providers do (e.g., house numbers or turn restrictions), I don’t think there’s any fundamental reason to think it couldn’t.

OSM’s tagging model can capture most concepts, you can always buy a bigger disk/server, and the road network is stable enough that individuals can easily track changes (as I noticed doing Haywards Heath, some of the errors on Google Maps are at least 4 years old).

The limiting factor is going to be that it’s not very interesting to go out and write down house numbers or parking restrictions, so beyond the basic road network there may be some tasks that require more effort than you can expect from a voluntary project.

But those are all solvable problems, so I think the only areas that will be inaccessible in the future are things like nautical/aviation data. Simply because the resources you need to capture it are still out of reach of individuals, although that may not be the case forever.