Intellectual property rights?

[Rob Seaman]

Hi Andrew,

Thanks for the thoughtful reply.

> Astronomical IP is an area that has not really been much of an issue  
> to date - unlike many sciences, most astro researchers are quite  
> open to sharing data, after a nominal prioprietary period has  
> passed. This traditional method is starting to change, though, with  
> various large projects adopting extended proprietary periods,  
> although primarily to ensure quality control rather than out of a  
> sense of ownership or copyright. Ultimately the data are only going  
> to be used by the community if access to it is open.

I'm a big supporter of open source software and open access to data.   
I think we're talking apples and oranges, however.

Proprietary access restrictions - whether they expire after some  
period or not - represent a contract (in effect) between the original  
investigators and the observatory.  For this contract to have any  
point, one or the other must "own" the data.  In most, but not all,  
cases it is the observatory that retains ownership.  This is also  
implicit in the observatory being able to build an archive of holdings  
that ranges across multiple investigators.  And in the VO being able  
to federate those distinct archives.

In short, the proprietary period is a private agreement permitting  
investigator access to observatory data, while copyright (or  
equivalent) is the public agreement limiting community access to those  
same observatory assets during (at least) the same proprietary  
period.  They are different pieces of the same mechanism.

> I think that copyright is a particularly bad model for astronomical  
> data access control, similarly for patent law - the main reason for  
> copyright or patents is to allow the inventors/authors of a saleable  
> commodity to have time to profit from their work/invention.

A good analogy might be to open source software.  AURA retains  
copyright in the IRAF source code, specifically to provide it to the  
community via a (loose fitting) "copyleft" style license agreement.   
It may be that copyright is a good fit to software, but a poor fit to  
data - but is there a better model of ownership for scientific data?

An observatory could treat data as a trade secret, for instance.  As  
you point out, this is undesirable.  In any event, Coca Cola's lawyers  
will aggressively assert ownership of its secret formula whether or  
not the company chooses to publish it.  "Open access" may be another  
name for "controlled publication".  Open source is not the same thing  
as public domain.

> In astronomy there is no presumption of "creation" in the same way -  
> the sky is there and can be observed by anyone (with the technical  
> know-how to build the instruments to do so), so (in principle at  
> least) the data cannot or at least *should* not be "owned" by anyone  
> in this way.

A specific exposure with an astronomical camera represents a unique  
snapshot of the universe at a particular epoch from a particular point  
of view with particular hardware and observing conditions.  This is  
not different from photography (digital or otherwise) for other  
purposes.  Stieglitz's “Flatiron Building” is not the same as mine.

> I don't think I've heard of a case of an observatory citing  
> copyright to prevent people getting hold of data that is outside the  
> proprietary period (or even inside - they would simply cite the  
> proprietary period). Can you fill me in if there are well-known  
> examples?

I agree that it is hard to get observatories to focus on such issues.   
This is very similar to other issues of curation and preservation of  
data :-)

Let's try inverting the question.  It is commonplace for large (and  
small) observatories to form partnerships for different purposes.  The  
NOAO archive, for instance, contains data "belonging" to at least  
three separate countries (well, scientific organizations representing  
these) and five U.S. universities.  In addition, the observing  
resources pertaining to acquiring these data are allocated by an  
overlapping, but not identical, set of several different entities.

In the absence of copyright, how are these data to be kept separate?   
The rules (much more complex than simply a proprietary period) for  
disseminating each class of data differ.  We can simply tag each  
image, etc., with the name of the entity who "owns" it, but what is  
the foundational legal model for this tag, if not copyright?  Else, I  
could attach tags to "your" data and assert a completely different set  
of owners.

If you don't own your data, who does?

> In any case, I think all major survey projects these days are moving  
> towards a model of making data public with shorter and shorter  
> proprietary periods (within the limits required for quality  
> control). I think a simple time-limit model for access control would  
> be sufficient for almost all cases.

There are prominent counter examples that have nothing to do with  
quality control, per se.  There is certainly room in astronomy for  
multiple data policies.

I agree, however, that a time-limit is appropriate in most cases.  The  
question is what underlies that.  Again from the opposite direction -  
what is it that VO single sign-on is protecting?  Simply denying  
access isn't sufficient.  We can certainly expect SSO to be breached  
occasionally, if only via social engineering attacks.  (I've seen  
publicly accessible terminals at astronomical institutions with the  
root password prominently affixed.)  Are we to say to a PI whose data  
are compromised that there is nothing to be done because all  
astronomical data are public domain from the moment of their creation?

If data are not in the public domain, what are they?

> I can imagine that some teams may withhold data *products*  
> (catalogues, processed images/spectra) for more extended periods, to  
> maximise their scientific return, but then they are unlikely to host  
> them in a distributed access infrastructure like the VO in that  
> case, or if they are could simply apply an extended time-limit for  
> access control (with "extended" potentially being infinite, and  
> perhaps being modifiable after the fact).

Withholding access is certainly part of the implementation of an  
astronomical IP model.  This cannot be the fundamental legal  
protection, however.  Most astronomical research is publicly funded.   
What is to stop a curious citizen (let alone an overly aggressive  
academic competitor) from pursuing an FOIA (Freedom Of Information  
Act) request during the proprietary period?

More examples.  Observatories produce EPO materials from their  
scientific data.  EPO assets (like other educational curricular  
materials) are placed under copyright to protect the rights of the  
observatory while permitting public distribution.  Are we to think  
that the original scientific data is left unprotected in the public  
domain, while the wholly derived materials are worthy of protection?

Observatories may occasionally seek license fees for commercial access  
to their archival holdings.  In the absence of IP, how does this work?

Examples like these (hypothetical or not) underscore that there is  
some theory of IP already being followed in the astronomical  
community, even if such is rarely explicitly asserted.

Rob Seaman
NOAO
---

Homage to Stieglitz (and "Monopoly"):

-------------- next part --------------
A non-text attachment was scrubbed...
Name: IMG_7760.JPG
Type: image/jpeg
Size: 63473 bytes
Desc: not available
URL: <http://www.ivoa.net/pipermail/datacp/attachments/20081202/d108dcf4/attachment-0001.jpe>