Re: Open Data liability issues

Hi Makx,

On Wed, May 21, 2014 at 6:36 PM, Makx Dekkers <mail@makxdekkers.com> wrote:
> Good points, Leigh.
>
> I agree that there is always a need for re-users to assess the risk of using the data.
>
> However, I think it would be good practice for publishers to give the re-user more information than a blanket disclaimer that says "we don't guarantee the quality of the data" and
> "we may stop providing the data at any time" which for example the UK OGL does.

I agree. Those guarantees don't need to be in a legal agreement.

> In my view, there a two different issues:
>
> 1. Data may not be complete, correct or timely
>
> Ideally, a publisher could give some more information on which parts of the data may be affected and how. Legislation.gov.uk provides such information in its FAQ (e.g. 'How up to date is the revised content on this website?' http://www.legislation.gov.uk/help#aboutRevDate). Maybe the Quality and Granularity vocabulary could look at ways to make such statements a bit more machine-readable.
>
> In some of those cases, re-users, e.g. in-car navigation systems, will just pass on the disclaimer -- and they do: a screen pops up every once in a while to warn you that you yourself are responsible if you drive off the cliff. Also, the Scout navigation based on OpenStreetMap that you refer to has such a blanket disclaimer in its product information.
>
> 2. Data may be moved or disappear without warning
>
> Re-users may be confronted with situations like when a publisher decides that they can no longer afford to maintain the data, or when a publisher decides to move the data to another URI/URL. Our best practice should say something about this. It's partly related to the URI persistence but it should also cover the persistence and maintenance over time of the data itself. It would be useful if publishers provided information on the types of circumstances that might lead to temporary or permanent disruptions in availability of data (including termination of maintenance or removal), and how re-users could find out about such disruptions.
>
> We could include your suggestion in one of the BP documents that a publisher might consider to provide data on two levels: zero-charge data without guarantees and fee-based data with something like a Service-Level Agreement.

Both of these are important to capture. Have you and the rest of the
working group looked through the questions in the ODI Open Data
Certificates? Both of these issues are addressed. Publishers are asked
to provide a quality statement for their datasets as well as pointers
to support information. Datasets can only achieve the higher grades if
the publisher is committing to regular supply.

Here's an example certificate:

https://certificates.theodi.org/datasets/353/certificates/11801

Cheers,

L.

-- 
Leigh Dodds
Freelance Technologist
Open Data, Linked Data Geek
t: @ldodds
w: ldodds.com
e: leigh@ldodds.com

Received on Thursday, 22 May 2014 09:33:06 UTC