Open Data Commons — Questions and a response

Rob Myers wrote some excellent questions, and here is a response. I separated it out to a blog post to make it a bit more user-friendly, as there is a lot here to address. Thanks Rob for taking the time to write so much, and if you (or anyone else) have any further questions, don’t hesitate to contact me or post below.


If I render an image from licenced data (e.g. a map from OSM’s street data) how does the copyright on that work interact with the licence on the data? Can such a derivative work be relicenced, for example under BY-SA or as a proprietary work?

So the data is separate from the database, as much as possible, in this licence. If I understand your question correctly, then creating a map from a set of data in a database licensed under the Open Data Commons Database licence would only necessitate the notice described in Section 4.3 in regards to the database.

If I used the database licence, together with the factual information licence (covering the geodata in this case) then you could re-license the result provided that the notice provision was kept (per the DB licence), and the terms of the Factual information licence complied with.

Non-DB-Right Contract

“for circumstances and territories in which copyright and database rights do not apply, the Open Data Commons is also an agreement in contract for you to act in certain ways in return for accessing this database”

IANAL so I would be interested to know more about the basis for this. Are there precedents that show that it would be enforceable?

It depends on how the licence is used — this is related to click-wrap and browse wrap contracts in the Internet law world. If you set it up like software — i.e. you must click “I Agree” on the licence to access the database — this is the strongest case for a contract. If there is no underlying right being licensed (no copyright or database rights) then a browse wrap contract (terms and conditions somewhere on the site) might be a weak case.

Waiver Of Sharealike

The waiver of sharealike in 4.5 is interesting and fits many of the debates that have taken place within OSM.

4.5.a is unclear. I assume it is simply clarifying the scope of copyleft and aggregation.

I’m not sure what part is unclear, but 4.5a is meant to explain that the Share Alike does not extend to a Collective Work (in copyright terms). For an example from the content world, use of a Share Alike (SA) image in a newspaper (a collective work) does not make the entire newspaper SA. So too would putting a database together with other databases in a way that may attract copyright (collective databases) not be SA. For reference, see the last sentence of 4.b. in the CC unported licence.

4.5.b is a “mash up” clause. So this answers people’s concerns about mash-ups, which I think is a good thing.

There are possibly two problems with this though.

The first is that other data must come from “more than one source”. So I cannot just add (e.g.) telephone numbers to locations on OSM data, I must add telephone numbers and postcodes.

At first I wasn’t sure I understood the first point, but now I think I see what you are saying. I can see that there is an ambiguity here (or unintended “feature”) — that one must add 2 or more sources to the output. Would “one or more sources” address this?

Using this Database or a Derivative Database to produce a Substantial part of the Data (via a search or other query) that is then combined with information from one or more sources into an integrated experience does not create a Derivative Database (out of the integrated experience) for purposes of Section 4.4;

The second is that as written it is easy to create a query system which will allow the entire database to be used in a non-free way. I can provide all of OSM’s data through a query interface as long as I provide two bits of extra data with each query. This creates an “ASP Hole”, “licence laundering” or “backdoor” problem in the licence.

This is an intentional aspect of the licence, but perhaps not quite the way illustrated above. This involves not changing a database (i.e. a derivative database) but rather just getting a database to produce some information. So you don’t need to share alike because you don’t create a derivative database, nor do you need to add in extra data to be able to create this work. You still however need to let people know that the information was obtained from an open/free/libre source (per section 4.3).

I’m not sure that this is a ‘non-free’ use either. This is a great deal of freedom in being able to take the output of a database and mix it with other information and only have to let people know that you obtained some of it from a given source, and look here it is and the licence.

The flip side is that if you try to restrict what people do with the output by trying to force them to share alike, then you can only ever use the output of a database licensed this way with similarly licensed information. That results in the licence, IMO, being not very useful.

Hope that made sense — I should do a diagram and post some of my thoughts on this.

4.5.c mirrors’s the GPL’s internal use provisions. This is probably a good thing.

I think that b and c would answer many people’s use case concerns about copylefted geodata. But as currently written b both disallows the simplest mashups and allows the licence as a whole to be effectively circumvented.

I think I addressed both of these above. If not, I’ll post some more.

Parallel Distribution Language

4.6 contains the controvercial “parallel distribution” language that was not included in the 3.0 Creative Commons Licences revision. I am not sure that it is any better for data. It might be better if the licence allows private DRM-encumbering as the CC licences appear to.

So rather than allowing DRM-based distribution in 4.6, 4.5.c should allow people to add DRM or access controls in private or within organizations (where it is their problem and nobody else’s). This would allow use of licensed data by VLEs and uploading of licensed data to TomTom systems for example.

In relation to 4.6a, I agree that it could be drafted to allow for private TPMs and that this might be desirable. I must however emphasize the difference between “DRM” and “TPM”, as CC licences (and this licence) allow for rights management information (RMIs) and other DRM but do not allow for technological measures (TPM) that restrict what you can do with the work.

About parallel distribution, I’d like to err on the side of allowing these for the draft, as there may be situations unique to databases that make it better to include this language than in CC licences, which are really geared for content.

About your two examples though, we specifically included Section 4.6c for VLEs, as this was an issue that came up in the 2005 CIE study that Charlotte, myself, and many others were involved in. The goal is to allow works in VLEs but not to allow universities (or whoever) to restrict students from taking the works outside of the VLE. I’ve argued that this is the position of the CC licences, though it was a question at the time and so directly addressed here.

The second example, about TomTom. TomTom involves using the work being used outside of a company (if I understand it correctly) because the database would be on the device that they sell, so I’m not sure how this would be private or within an organisation. This is the kind of situation that the parallel distro language is intended to address. They can pop the database on a TPM restricted device with other data, but they must let you have access to it without the TPMs.

Moral Rights

This is the opposite to Creative Commons’s approach. Would a promise not to assert moral rights be binding in e.g. Germany or would it be regarded as unconscionable?

This licence is not an attribution licence, and so we didn’t see the need in asserting or reserving “the right to claim authorship of the work”, one element of moral rights.

As to the right to object to derogatory treatment, waiving this right ensures the widest possible use of the database.

Some jurisdictions do not allow the waiver of moral rights (such as many European Continental jurisdictions), while some require that they be asserted (the UK). If in a jurisdiction that does not allow waiver, these rights simply could not be waived — this part of the licence may have no effect in Germany or France (or others). But importantly it would allow for waiver in jurisdictions such as the UK. This balance is reflected by using “To the extent possible, …”

Fair Dealing

Fair Use is the international term I believe, although I did hassle CC to mention Fair Dealing as well in the International CC 3.0 licences.

I disagree. Fair use is the US term, but “fair dealing” is more common — Canada, Australia, the UK and other common law jurisdictions more closely based on the UK system (well, the E&W system) use “fair dealing”. The other concepts are the private copying exception in many civil law systems as well as the three step test in Berne and elsewhere. I should say though that I haven’t done a survey of all jurisdictions, but would like to note that “fair use” is explicitly mentioned in the licence (just not in the title of 6.0) and that the unported CC licence uses fair dealing. See 2.0 here:

Thanks again for your commments.