[OSM-talk] Zero tolerance on imports

Daniel Sabo danielsabo at gmail.com
Sun Feb 20 09:06:50 GMT 2011


On Feb 20, 2011, at 12:32 AM, David Murn wrote:

> On Sat, 2011-02-19 at 19:40 -0800, Daniel Sabo wrote:
> 
>> Maybe you don't like it, but you are not the entire OSM community. Yes,
>> in this case someone overwritten what I presume was good surveyed data
>> with an import was stupid. But in general the fact that data was
>> gathered by a government surveyor with tools an order of magnitude
>> more accurate than ours does not, IMHO, make it less "worthy" of being
>> in OSM.
> 
> As was stated in the previous email, OSM isnt designed to simply be an
> API for people to load government data into.  Your comment about the
> 'order of magnitude more accurate than ours', shows that maybe you need
> to read about the airport import, where some airports were added many
> kilometres away from their accurate location, or several cases reported
> here in the last few weeks of deleted surveyed data and tags, replaced
> with generic un-verified tags.
> 
> Yes, some imports are good, especially when the community is consulted
> and comes up with a proper way to import/tag/maintain them, but if one
> individual just decides to use their own data sources, with their own
> tags without consultation with the community, it almost always leads to
> more damage than added value.  Its no good doing an import that 'saves'
> 20 hours of handwork, if it takes 100 hours of handwork to correct it.
> One practical example of this, was when I made a cross-country trip last
> year, and found some fuel stations were mapped upto 100km away from
> their real location.  Imagine if someone was relying on that data, in a
> remote area if could be life-endangering, or worse imagine if it was a
> hospital import. Some things Id rather not have mapped, than to have
> them mapped possibly inaccurately, or to at least have renderers show
> imported data with a warning to be wary of its accuracy.
> 
> David
> 

I don't disagree with anything you said. What I disagree with is that every time someone does a bad import there's a very vocal "ban all imports" response like the first message in this thread. If it were my choice I would have reverted the airports import immediately, but other people are more tolerant :).

When I say "an order of magnitude more accurate than ours" I'm restricting that to things things that are more accurate, not random lat,lon coordinates people pull of the internet :). I'm pretty sure that most municipal road data is much better than anything I can do with my GPS, and if someone can do proper conflation on it I'd have no problem with my roads getting aligned with their roads.

The problem isn't imports. If you personally know that the data is good, or are willing to put the time into verifying it and checking for dupes there is nothing wrong with importing data. The problem is that people either don't know how to do this due diligence or they're lazy and assume that someone else will clean it up. From most of the "oh god why did you import that" discussions I've seen here there's a lot more user error than malice.

The other issue is that the community is hard to communicate with. As far as I've seen the mailing lists are the only active OSM discussion, and IMHO mailing lists are a bit of a barrier to entry. Expecting people to join 3 mailing lists (talk, imports, talk-country at minimum) to ask if it's OK for them to map something is unreasonable.

- Daniel


More information about the talk mailing list