On Mon, Jan 7, 2013 at 9:34 PM, Jeff Meyer <j...@gwhat.org> wrote: > Isn't that true of all data in the database?
OSM is built on surveyors doing surveys. That is we have people who go out and walk around with GPSes, or maps, and manually survey what's on the ground. Then when a second person goes to the same area, they are validatidating the original data. Maybe the second person has more accurate data in some part, or maybe there's been a change, etc. We've shown in studies that the number of mappers increases both the data density and the data accuracy over time. But this only works with ground observable data. Land owership isn't ground observable. Sometimes a feature, such as a fence is, but the actual land owership isn't. Therefore, it's not possible for a second observer to come in and provide either validation or updates to the data. Additionally, land ownership changes frequently. Lastly, there is only one authoritative source for this data. To recap: Land ownership data is only available from the government, which is the one authoritative source of this data. It's not something that the "crowdsourcing" model lends itself well to. And it changes rapidly. So what Ian has suggested, and I agree with him on, is that this data is a poor candidate for inclusion into the crowdsourced OSM data. That doesn't mean it can't be used alongside it. This land ownership data (assuming it's licensed properly) can be rendered on the same map as OSM data (there are many examples of using TileMill to mix data sources in just this way) and if the data is imported into a database, there can be queries made against the two sets, so it would be possible to see the land owner for a given POI, for example. This is the best of both worlds. It keeps the OSM focusing on its strength, and makes it easy to stay current and precise on the land ownership dataset. Someone else brought up boundries, so let's discuss boundries. Boundaries in OSM, especially in the US, have been an ongoing and constant problem. Boundaries are places where people are fiddling all the time, trying to get the exact "levels" right. In addition, much of the US has duplicate boundaries (places represented by areas, and nodes), arguments about the definition of spaces, disagreements in the data between municipal and census data, etc. And this data changes, and we have not (even after years of working on the problem) found a good way to conflate and update. Finally, on top of that, the information Flickr has collected is telling us that our idea of neighborhoods needs to be rethought,and really does not lend itself well to the OSM model. So there too, is a potential win for OSM. We could rely on current, highly accurate public domain boundry data and use that for rendering, geolocation and other places, while keeping it out of the OSM dataset. The result of this would be: 1. More up to date maps 2. More accurate maps 3. Better geolocation (forward and reverse) 4. Reduction in errors caused by flawed data in OSM 5. Less editing wars due to differences of opinion between mappers and the authoritative data sources 6. Allowing OSM to focus on its core strength This seems like a win for everyone. - Serge _______________________________________________ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us