[OSM-legal-talk] OSM DWG tools

[andrzej zaborowski]

On 9 September 2010 06:02, Eric Jarvies <eric at csl.com.mx> wrote:
> I would like to make some suggestions, that otherwise seem obvious to me, but may not seem the same to others.  This prompted by my recent experience with identifying OSM data on a notable third party site/source/repo.
>
> I think the DWG should have some basic, yet effective tools, with which they can quickly and easily query specific ways, but more importantly... specific changesets of a given way, wherein they can then easily render it, and then easily mash it up in a map stack(with the offending sources), allowing them to easily visually deduce similarities.  If these are found, then mathematical similarities(coordinates) should be run to further substantiate the finding.  I understand this would often times mean using screenshots, wherein points of the screenshots would need to be selected/designated in order to match up with OSM(projections, etc.).  And the same applies to data sources, wherein a simple tools should exist that could take and extrapolate the data from these suspected sources, namely coordinates and tabular attributes, wherein they are run against OSM data and all the same instances are identified, as as all reasonable similarities.    Anything, really.  So long as it has a documented procedure(repeatable by third parties), then it'll suffice(for contributors/membership, and cases that may in face need/depend upon it).

The FSF has recently released an auditing tool for scanning binaries
for existence of GPL software in them, based on a database of
signatures.  I don't think this could work for map data, first of all
because you'd need the vector data and these services you talk about
mostly just show you bitmaps.  Then, it's really easy for them to
slightly modify the geometries so that no automatic scan will detect
the similarity, or, if it will, it will also generate tons of
false-positives.  And then I don't think it would be useful, there
aren't so many cases of suspected infringement and when there are,
accusing someone of misusing your data is too serious to rely on fuzzy
matching, you will always want to be sure. (compare visually and if
possible check if changes in OSM are reflected on the other map.)

That said the FSF audit tool is mainly for big companies that release
software but have too many people working on it to control everything
that they do, so they just scan everything that leaves is released to
the public to avoid getting sued by FSF or other authors.  Maybe it
would be useful for OSM to have a subscription to teleatlas and Navteq
vector data with the API matching all incoming data against it and
alerting somebody if a suspicious large upload is detected.  Google
map maker and Waze could do the same thing matching contributions
against OSM.

BTW the AMF api has a nice call that returns all versions of a way
including, where "version" means any state of the way's geometry,
rather than just tags and nodes ids.  Potlatch can use it with deleted
ways, but another client could conveniently use it to visualise the
history of any way.

Cheers