Openness
| 20 April, 2011 | Richard P. Grant |
8
|
|
Open source, open access, open posters even–but open science? Is that a step too far?
The arguments over whether open data–publishing experimental results on the web, making datasets available, etc.–is a good thing, for science as a whole or individual careers, are likely to rage for quite a long time to come. That hasn’t stopped people working hard to make it happen (you might want to check out my three-part interview with Peter Murray-Rust for example), and last week Sage Bionetworks held its second annual ‘Commons Congress on tools for doing open science.
Sage plan to improve human health by making it easier to share data and new discoveries. They face resistance from the research community itself, both because some are reluctant to share information for fear of being scooped, and because the most prestigious journals are still not set up to aid data sharing (quite aside from the open access argument). However, “If these folks help speed the discovery of new treatments, they could help save our lives someday” says Ryan McBride.
But there’s another argument the ‘open everything’ movement needs to consider: that Open Science just won’t work. This is because of ‘credit’:
Credit gets jobs, keeps jobs, gets funding, attracts students and bestows respect and international standing in the community. If there is no way to ensure that scientists get credit for their ideas and intellectual contributions, then they will not contribute to the Open Science movement.
Now, I’m not sure about this. If you can find a way to make sure that scientists do get credit, then why shouldn’t it work? Indeed, there are moves afoot to make sure this does happen (and I’ve been involved in some discussions about how it could work): with a reliable author identifier linked to a robust system of tracking each quantum of discovery, or datum, or idea (such as, oh how about the DOI?) it shouldn’t be beyond the ken of open science advocates to build a system of microcredits that might altogether do away with one of reasons for the traditional, article-based publishing system.
That’d be interesting.
[poll id=”8″]
|
|
previous post
If the idea is for scientists to post all of their raw data online before publishing, I agree with the critics. For many fields of research, the results of a project may be a single dataset which took years to obtain. In crystallography, for instance, when would be the appropriate time to release the data, if not upon acceptance of a paper describing it? Should the diffraction images be immediately uploaded to a public server? Or the final refined structure? How do you ensure that competitors won’t use the results as a clue to solve their own competing structures? Or perform all of the experiments required to validate the biological interpretation before the original crystallographers are able to? These aren’t merely hypotheticals; everyone knows stories where this has happened. I’m skeptical of the microcredits idea; being thanked in the acknowledgment section of a Nature paper (or, perhaps, being listed as 8th author) won’t get you a faculty position.
Once a paper is published, of course, there is no excuse not to immediately make all data freely available, and we’d be far worse off if journals (and granting agencies) hadn’t started to mandate PDB or GenBank depositions. But immediate release is only possible for projects where an individual scientist has no hope of – or desire for – obtaining an independent position as a result of his or her efforts. This works well enough for genome sequencing or structural genomics (which do release data very quickly), but why should a postdoc spend five years alone at the bench battling an uncooperative membrane protein, if not for exclusive publication rights? They’re certainly not doing it for the money…
These are good points. However, I’m not sure that without the infrastructure and the experience that our example author has, anybody else would be able to scoop them given what data are sharable in this way. It’s hard enough trying to replicate an experiment that has been published in the usual way…
The microcredits idea doesn’t mean getting one 8th author Nature paper. It means getting lots and lots of n-numbered papers, or even lots and lots of data all over the place. Yes, it’ll mean changing the way faculty positions and grants are awarded, but that’s no bad thing.
(And if our putative post-doc spends five years battling a membrane protein, and somebody else manages to do it faster and publish first, is that a bad thing overall?)
“The microcredits idea doesn’t mean getting one 8th author Nature paper. It means getting lots and lots of n-numbered papers, or even lots and lots of data all over the place. Yes, it’ll mean changing the way faculty positions and grants are awarded, but that’s no bad thing.”
I certainly don’t oppose changes to the current system, but the ability to produce large amounts of data does not necessarily correlate with the ability to design an independent research program, mentor students, etc. (I speak from experience – I’m very good at generating data, less so at analyzing it.) Obviously the ability to produce a single Nature paper during a postdoc isn’t always a good benchmark either, but we should find a system that’s genuinely an improvement before throwing out the current one.
“(And if our putative post-doc spends five years battling a membrane protein, and somebody else manages to do it faster and publish first, is that a bad thing overall?)”
No, which is why the NIH already funds competing projects. But my scenario assumes that the project is successful and the first one to yield results. Why shouldn’t the scientists who obtained the data be the first to publish it? Or, to repeat my original point, why should they even bother doing the work in the first place if they won’t have some (short-term) exclusive rights?
Collecting data makes one a technician. Understanding data makes one a scientist. There’s a big career difference between the two, so releasing your data before you understand it and allowing others to make the intellectual leap you have not yet reached is a risky move. Just ask Rosalind Franklin.
Then, of course, there’s the issue of IP and patents. I’m sure you’ll see most of the universities who bring in huge amounts of money from these (see Stanford’s share of Google) pressuring researchers to keep their discoveries private until patented.
Also the problem of same data being collected from different labs being different.
Also the problem of same data being collected from different labs being the same but of hugely inferior quality.
Also the problem of no explicit association of the time in which data was collected.
Also the problem of inherent fluctuations in data that can lead to accurate measurements but disparate results.
Also the problem…
Grrl is right as far as she goes, and the DOI is a mostly toothless wombat; DOI never was a tiger.
In physics there isn’t this problem, because data can be published, and the “mere technicians” are respected.
But I’m for it because it will help two problems. A.) PIs who pressure for results and don’t care if the data is faked unless it bites them. B.) PIs who pile up data and don’t interpret it – ever. There is more of this than people think.
The former can be improved because base data can be examine by others, and in some cases it can be proven to be falsified if it must be available.
It is amazing how much you can get done if you don’t care who gets the credit or who makes the profits.
There is already open dialogue among people in the blogs. Sometimes the lack of formalism facilitates creativity and collaborations. Somewhere between blogs and white papers is a path of constructive collaborations where dialogue is encouraged and IP protected….. How can we get there?
Cameron Neylon has written a response to Grrl. Go have a look, and say hi from me.