[open-science] Making science more accountable and efficient

[Sören Auer]

Dear Cameron, others,

Thanks for your comments so far.

> There have been a lot of efforts along these lines. Many have come
> from different angles but the idea of "levels" of certification for
> openness has been a common theme. There is a message inherent in the
> fact that there have been a bunch of efforts and most of us have
> never heard of them.

Can you give me some pointers to these related initiatives?

> The message is that this is hard because you're trying to tackle a
> bunch of things at the same time. First get some agreement on what
> standards of openess are.

I think for software (Open Source), creative works (CC) and data (Open 
Data) we have these standards already, what other scientific artifacts 
are there? Maybe models (like UML models, ontologies, taxonomies, 
experimental architecture models) - I guess one of the other categories 
might fit for them too, e.g. ontologies/taxonomies could be considered 
as Open Data or creative works in terms of CC.

Also, openness (as I said in my initial email) is not the most important 
issue here - if an artifact is not open for good reasons, but it is at 
least made available to the scientific community and public for 
peer-review evaluation purposes, this is already a big step forward to 
what we have now.

> Second problem is that you need to go from zero adoption to wide
> adoption before you can have an effect.

I think you have already an effect if one conference or journal requires 
authors to classify their work according to the suggested criteria. All 
authors submitting to this journal or conference will have to think 
about it and since the rule applies to all submission to this particular 
science outlets its also fair. In fact if we would manage to get a few 
good science outlets in a certain field on board this could be the start 
of a paradigm shift in that field.

> My suggestion would be that it is easier and probably more effective
> to bite of small chunks of the problem.
> This is what we tried to do with the Panton principles, and various
> people are looking at generalising them beyond data (not without some
> disagreement).

I think this fragmentation actually is counter productive. Ideally this 
would be tackled with one set of principles and/or manifesto and I would 
be more than happy if my ideas, for example, could be integrated into 
the Panton principles (although I doubt this would be possible, since it 
would require quite some substantial changes).

> As John pointed out Victoria Stodden has essentially described what
> the gold standard should be.

I think making openness the gold standard and requiring scientists to 
release their data under CC0 can not be a model for science in general. 
There are many cases where its actually acceptable that data, software 
etc. to be released under different terms and this should not be 
neglected. However, the terms should be made explicit and there should 
be always a publicly available version for peer-review and evaluation 
purposes. At least for computer science, I think this would in the end 
indirectly dramatically facilitate openness, currently I have the 
impression many computer science researchers "hide" their artifacts more 
because their implementations are not robust and usable enough than 
because they are against openness. If there is preassure (from journals 
and conferences) to make it publicly available, they have to improve 
usability and robustness anyway and then they will also make it 
available in truly open ways.

Sören