Publishing source attribution at a static URL

[riordan]

Lately copyright has been on my mind quite a bit*, particularly with respect to making it easy for downstream users to do proper citation. And it would make the project a lot easier for downstream users to cite if there were a single URL that had all the appropriate citation/attribution information.

It dawned on me and @migurski a few weeks back that many geodata projects these days have some sort of a copyright page that makes it easy for downstream users to link back to (think openstreetmap.org/copyright, the quattroshapes license, even the geonames sources.

In a sense we publish this data already, it's included in each source's configuration json and the LICENSE.TXT file distributed with each collected.zip. But having it in a single URL, with all the attribution requirements and licenses specified for each source, would make the citing the project (optional), and the sources (required) as easy as:

Data from OpenAddresses. <a href="https://openaddresses.io/attribution">Attribution</a>

or

Data from OpenAddresses. <a href="https://openaddresses.io/attribution/share-alike">Attribution with Share-Alike</a>

We're building a new citation system into soon for Pelias that'll do this automatically whenever we rebuild, but it'd be substantially easier to just drop a link in where we know it'll always have the most up-to-date data.

Right now, its up to an individual user of openaddresses to go ahead and

* As is usually the case.

[iandees]

Two things come to mind here:

1. Would the URL need to be versioned, depending on which version of the OA data is referenced? What happens when a data source is no longer included in OA, but the data dump you downloaded still includes it?
2. I'm mildly worried about this making OA a little more liable for any legal problems that might come up from someone incorrectly using or misattributing data from OA.

[migurski]

I would reference a date in the document, e.g. “license information for OA data as of YYYY-MM-DD.” 

Say more about your worry—are you thinking that we should be staying quiet about licenses?

[NelsonMinar]

I have some legal worries too, but in the absence of a lawyer no idea how to resolve them. Just a feeling that the more we say about copyright and license, the more obligation we have to do this stuff correctly. OTOH I think it's fine for us to make a good faith best effort and be proactive about removing any sources if we get a complaint.

Perhaps there should be some boilerplate text about how OA can't guarantee any license terms and it's up to the user of the data to clear license themselves? A contact address for copyright complaints would be a good thing too.

[sbma44]

This puts OA in a position where it looks a lot more like it's issuing sublicenses rather than simply organizing data (we really don't want people to get the idea that they just have to attribute OA -- in some cases we have the right to do that, but in many others we won't). I'm in the same boat as @NelsonMinar on this one.

[migurski]

Would there be wording that you’d be more comfortable with, @sbma44?

[sbma44]

It's probably worth asking an actual lawyer. My instinct is that aggregating disclosure terms in a way that makes it easy for the end user to print out is great, but that taking responsibility for the display of those terms on openaddresses.io & encouraging users to link to them is more dangerous. It could easily be that a lawyer will disagree though.

[migurski]

I’d love to hear an opinion on this; it’s new territory for me. I believe we’re going to be running a bunch of this by Mapzen’s lawyers as well.