[OSM-legal-talk] ODbL: How obscure/inaccessible can published algorithms be?

[80n]

On Sat, Dec 12, 2009 at 12:07 PM, Frederik Ramm <frederik at remote.org> wrote:

> Hi,
>
>    OdbL has this requirement where, if you publish a produced work
> based on a derived database, you also have to publish either
>
> (a) the derived database or
> (b) a "diff" allowing someone to arrive at the derived database if he
> has the original, publicly available database or
> (c) an algorithm that does the same.
>
> Is that correct so far?
>
>
I don't think it's quite as simple as that.

Suppose, considering your WMS example, two separate companies provide
identical services:

A) The first service uses massive processor power to analyse a raw planet
dump and provides the output directly.  Perhaps a larger company such as
Google might take this approach.

B) The second uses some algorithms and optimisations that involve creating a
derived database, but results in exactly the same output as service A.
Perhaps, a smaller company such as Geofabrik might take this approach.

Let's assume that neither company voluntarily publishes any information
about their methods.

On what basis can you demand from company B that they release their
intermediate database?  You don't know (for sure) that they have an
intermediate database.  The ODbL doesn't give you any rights to ask company
A to warrant that they are not using an intermediate database.

What kind of duck test can you use to be sure that a derived database is
involved in the process?

80n



> To use a simple example, let's say I build a WMS that works with OSM
> data. To make this perform well at low zooms, I have to combine ways
> into longer bits and simplify their geometry. The result is clearly a
> derived database that falls under the above, and in practice I would
> probably choose the "a" route and simply make a weekly PostGIS dump
> available for download and be done with it.
>
> However, I wonder about the permitted ways of doing (c).
>
> I guess it would probably permitted to specify a number of PostGIS
> commands that achieve the changes. - Let us assume for a moment that
> applying these PostGIS commands would require a machine with 192 GB of
> RAM and Quad Quadcore processors and still take two weeks to complete,
> putting it out of reach of many users. Would it still be permitted to do
> that?
>
> Or, would it be allowable to say: "For simplification, a Douglas-Peucker
> algorithm <link to DP wikipedia entry> is used." (leaving open the exact
> implementation and parametrisation of DP - bear in mind that with some
> algorithms, how they work is easily explained but implementing them in a
> way that runs on standard hardware may be a hard task).
>
> Or, would it be allowed to say: "For simplification, just load the data
> set into <name of horribly expensive proprietary ESRI program> and hit
> Ctrl-S X Y, then choose Export to PostGIS"?
>
> What about: "For simplification, we did the following steps: <detailed
> instructions that are easy to follow>. These steps in this sequence are
> patented by us, so if you want to follow them, please apply to us for a
> license to use our patent."
>
> Bye
> Frederik
>
> _______________________________________________
> legal-talk mailing list
> legal-talk at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/legal-talk
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/legal-talk/attachments/20091212/9e287f0a/attachment.html>