The OpenStreetMap Way

It’s very interesting observing and participating in the OpenStreetMap community .. it’s one of the most frustrating and inspiring groups I’ve ever come across. The past couple days have seen a long thread on the use of OGC standards and compliant tools in OSM, something that comes up with remarkable periodicity and some unnecessary hostility from both “sides”. OSM is not anti-standards at all, in fact uses PostGIS in a limited way right now, but it’s such a unique and new idea that most of the infrastructure has been built with agility, under quick iterations of home grown tools. Traditional GIS never conceived of a system where map creation was open to everyone. There are complete non-geographers (neo- if you wish) and some folks with a good deal of history in GIS .. and as a community we rediscover the conventional wisdom just as often as we overturn it. I’d personally love to have a WFS-T interface to OSM, opening up to whole new suites of tools .. but actually very few find it workable, even the Ordnance Survey has been looking at alternatives (something I hope to talk about soon). Back to thread .. towards the end it calmed down and several people articulated well the “OSM approach”. Bravo.

In the mean time, the commercial data providers continue to move towards OSM approaches. Navteq has launched Map Reporter for users to feedback map errors (it’s the number one customer service request with online maps and navigation systems). It’s definitely a more usable tool than the similar TeleAltas MapInsight. I was easily able to report one of the dozen or so errors within a half kilometer, a missing alley, and can even share my report with everyone. This alley is of course accurately, and the whole area in my opinion more beautifully, represented in OSM for the area (try the osmarender layer for the most detailed view). But for Navteq to be successful, it really needs much more transparency, otherwise these submissions enter a blackhole, without engagement .. what is Navteq’s schedule for looking at these submissions; when is the next surveyor coming around; how many other people have reported the same error; and what’s the benefit for contributing. For OpenStreetMap, I think we can learn that in the future we’ll need multiple levels of engagement .. some people won’t want to invest the time to edit the map itself, but are happy to report errors, and we’ll need an interface for that.

Btw, I’ll be talking about these things and more tomorrow night at the Sussex Geek Dinner.

3 thoughts on “The OpenStreetMap Way”

  1. That’s great the OSM is moving towards PostGIS, that would be a great step towards WFS-T.

    Is Ordnance Survey actually using WFS-Transactional? Or are they just looking for alternatives to WFS for output? I do wish WFS-T didn’t get all lumped in with GML 3.1.1, which is what people have a hard time working with. It’s actually quite simple, indeed on openlayers shows that it really doesn’t take many lines of code to persist a feature to WFS-T. The thing that most people find unworkable about WFS is the huge volume of GML – producing and parsing it in a timely manner, as well as doing the on the fly figuring out of XML Schemas. I’ve got JSON output from WFS instead of the default GML, which after this month of craziness ends I hope to post and show how WFS itself isn’t so bad, it’s really more the GML.

  2. OSM is using PostGIS for tile rendering. The data is preprocessed, so not appropriate for any editing operations. But I mentioned just to show that’s in not a aversion to any particular tool.

    WFS.js does look quite simple. I could see how those operations could be mapped to OSM API operations. One useful thing I think would be a proxy, which translates between WFS-T and OSM API .. this would show that it’s feasible to do OSM over WFS-T, and demonstrate the benefit through additional suite of tools available for OSM.

Comments are closed.