[open-science] Openness and Licensing of (Open) Data

Mon Feb 9 14:36:16 UTC 2009

2009/2/5 Neylon <cameron.neylon at stfc.ac.uk>:
[snip]
>>> Let's be clear: most open licenses whether in software, content or
>>> anywhere else aren't 'enforced' (and won't be enforced) by recourse to
>>> the courts. Like norms licenses will be primarily enforced by social
>>> pressure. Saying use licenses isn't saying: let's settle all our
>>> disputes in courts!
>
> The problem is - that as soon as you call it a license then scientists will
> largely assume that is exactly what you are saying, and that will provoke a
> strong negative response. Call it a protocol, call it a standard, and people
> will feel comfortable. Hint that it is contractual and potentially
> enforceable and we will do two things. First run in the opposite direction
> and second ignore the terms /because/ it is a license.

What's weird to me is that I know scientists who seem quite happy with
putting CC licenses (open and non-open ones) on their blog or or
applying open-source licenses to their code. I'm interested why
people's attitudes on data are so different?

[snip]

> John's view, and to a certain extent mine are summed up by:
>
>> in geneal, my instinct is that it is fine if these things *look* like
>> licenses, as long as the enforcement is not in the courts. we create
>> enormous potential for negative effects by using ip and courts, for
>> little return. if i were evil i would be al over setting up a data
>> troll business, use some easter eggs, and sue for revenue in about
>> five years. makes patent trolling look like simple street robbery.
>
> The issue here is not whether someone would actually do this - but that
> scientists will immediately see the possibility and get disturbed by it.
> Remember these are the people who assume that it is more likely that if they
> make data available then someone will steal it, write up seventeen Nature
> papers, and become a millionaire without crediting them, rather than someone
> might make some modest use of it and credit them in the accepted way that we
> generally do.

So we all agree that attribution (and perhaps share-alike) are going
to be wanted?

The question then is whether to put these 'requirements' in norms or
licenses ...

> It's an issue of control - if researchers feel they have ownership of this
> protocol/standard then there will be fewer barriers to adoption. If there is
> a perception that the protocol is "owned" by lawyers, which is what I
> believe will happen if it is called a license, then there will be very
> strong resistance.

OK that this is a naming or institutional issue. We could name the proposal:

Open Data Standard and License

Also I must emphasize: even if you want PD you're going to have to
apply that is (or looks really like) a license since there are DB
rights in a whole bunch of places ...

> In my view our best chance of promoting best practice and "enforcement" is
> through a protocol that funders choose to adopt for the research they
> support. This is a much more powerful way of driving researcher behaviour
> than through person to person contractual arrangements. I think this also
> gives us the best chance of pushing through the idea that data and
> collections of data should be explicitly placed in the public domain to
> maximise the ability for re-use.

I feel this is a related but distinct (and v. important) issue: how do
we encourage 'open' data in sciences?

I think your suggestion of driving this is via a protocol but this is
distinct of what people actually do to make their data open. The
license/norms are the way we define what is open. That is we need:

1) definition of how people make their data open (~ publish it
somewhere and apply appropriate license/norm/...)

2) Encouragement (probably from funders) for scientists to use this mechanism.

> [Do we in fact agree on that incidentally? We haven't really touched on
> whether we agree what the idea world would look like. I know Rufus feels
> that share alike is acceptable but I don't know whether that means he thinks
> it is on balance a good thing.]

I am ambiguous about share-alike. Personally, for my own stuff, I tend
to the 'MIT/BSD' end of the spectrum (i.e. attribution only with no
share-alike). However, I know people/communities where formal/informal
share-alike is pretty important.

> [Is the mailing list published to web somewhere or should I repost this
> somewhere? I'm very keen to keep the discussion out in the open as it were]

Yes:

<http://lists.okfn.org/pipermail/open-science/>

If people want rss feeds that can be arranged via our registering the
list on gmane.

Rufus