[OSM-talk] vandolism on OSM

Dave Stubbs osm.list at randomjunk.co.uk
Fri Oct 3 22:34:33 BST 2008


>
> I think my idea deals with non-obvious vandalism very well.
> A user of the data can choose to use data that has only certain
> tags by certain groups or individuals and therefore have an idea of how
> accurate that data might be.
>


Yes, but in a world of 65000 users you're left with a lot less data.
It may be enough for your purpose in which case that's great. For
smaller areas it probably works quite well, but I'm fairly sure it'll
hit a scalability problem. With OSM as a whole we rely on there being
more good guys than bad guys, and that's generally worked so far, but
you do have that instability problem in that it can take a while for
errors to be detected and corrected.

The subtle vandalism is hard to spot because it looks genuine. You
only determine it isn't genuine with local knowledge or on the ground
observations. That means you can't trust anything until it's been
checked out by a trusted member and the more people you bring in as
members to hit your coverage goals, the more chance you get
compromised and start letting in lower quality data.

How much of this is actually a problem depends a lot on what you are
actually trying to achieve.

Dave




More information about the talk mailing list