Proposal: consider scientific research papers like open source software

http://lemire.me/blog/archives/2011/02/11/taking-scientific-publishing-to-the-next-level/

by elehack • 15 years ago

40 16 15 years ago

impendia• 15 years ago

Along related lines, Tim Gowers (a Fields medalist in mathematics) has started the Polymath Project, which has produced several excellent, massively collaborative, research papers in math.

http://polymathprojects.org/

_corbett• 15 years ago

as a researcher I am completely on board with this idea-but to work it has to be built into the culture, the grants, and the way hiring is done. even demanding code be open source is an uphill battle, no major grants in my field provide any money for support, maintenance, hosting etc. or demand code be open-sourced for funding.

evgenit• 15 years ago

This is a wonderful idea, particularly since even a partial implementation would do good. Just having a license for derivative works and the latex source would do wonders.

A typical situation is that you notice an obvious extension to a result in a paper, which does not warrant publication. Being able to make an updated version of a paper with that easily would be a good thing.

jessriedel2• 15 years ago

Nothing wrong with the idea per se. But, as usual, no explanation for how the incentives and cultural changes needed to induce people would be implemented. Sorry, but no one is going to reward a grad student, career-wise, for rewriting other scientist's papers and catching bugs. Heck, the only reason scientists proofread their stuff at all is to avoid embarrassment. If there were an army of students eager to fix your mistakes, scientists would be even less careful.

maxbRuns2• 15 years ago

I agree. In my mind, one of the cultural changes that we need is for universities to abandon the tenure system. I think if that happened, there'd be more hiring of new professors, so there would be more of an incentive for academics to maintain a strong reputation in their field.

Academic versions of github might arise, as ways for researchers to show their prospective employers what they can do.

ylem• 15 years ago

I doubt that removing tenure would result in the hiring of new professors. One recent trend has been to hire adjuncts as much as possible and to only slowly hire tenured faculty. Let's suppose that you have a professor (say a theorist) who works on a few ideas--but deep ones--she can take her time and focus on them instead of worrying about losing her job because she hasn't produced recently (funding is another story). Similarly, if she does something controversial--then again, her job is safe.

The incentive for most academics (say science/engineering) to maintain a strong reputation is personal. I do research because: 1) I want to know the answer to a question 2) I'm selfish and I want to know the answer first ;>

This leads me to putting in 100 hr weeks (sometimes you have to grind it out--other times, you have to step back and work less to be more creative) at times and being rather productive. Do you really think that fear of termination would result in the same output? Do you think that the average institution could pay researchers to put in the hours that they do--weekends, holidays, evenings, etc.?

neutronicus• 15 years ago

Tenure is the one big incentive academia has going for it. Un-tenured academics work like dogs for peanuts (given their skill sets) to get tenure. You take that away, and it becomes "work like a dog for peanuts your entire career". Who's going to take that deal without the carrot of "do what interests you without fearing for your job security" at the end of the tunnel?

Streams of irrelevant publications are only going to get worse when "publish or perish" extends to your entire career.

(Not to mention that tenure is a good idea for, say, climate change researchers, or people doing potentially provocative demographic research, or stem-cell researchers, or anyone else someone influential somehwere might want to see silenced)

Cushman• 15 years ago

I think the subtext here is that academic culture should become more like OSS culture, where your reputation is built on the content of your commits rather than just having your name on a project somewhere.

nocipher• 15 years ago

I fail to see how it changes things in a fundamental way. Having your name on a paper is supposed to mean that you contributed to the results of that paper. This is how you build up your reputation: you publish lots of relevant results.

This is only possible if you continue to work and publish, which in turn makes you a more sought after partner for collaboration. In the same vein, if your contributions are consistently minimal, then you will be less sought after which will lower your frequency of publication. In essence, the current system already correlates rewards with the quality of one's contributions.

Thus, the proposed changes do not affect how one garners reputation. Instead, I think the subtext is that the current body of research papers could be significantly improved, but isn't because there is no system in place to revise them in a distributed manner.

jessriedel2• 15 years ago

To elaborate on what nocipher said: The problem is not that people are getting their names on projects to which they didn't contribute much[1]. The problem is that hiring committees only care about new scientific works you are producing and not about other ways to contribute to science such as fixing typos, pedagogy, etc. If you want to incentivize such work, you either need to change the criteria upon which universities hire (unlikely) or give supplementary rewards such as cash prizes (possible).

[1] Yes, there are times when it is hard to tell who contributed what to a scientific work. And sometimes people get too much/little credit. But this isn't a massive issue, nor is it something that can be fixed by adopting an OSS culture. Software is often much more modular than science is, so it is often possible to put individual names on commits. But at those times when software is more like science--such as when several smart people are sitting around designing over-arching ideas/structure/strategy for a large project--then software faces a similar difficulty in teasing out individual contribution.

zipdog• 15 years ago

I can't even access half the published papers I want to read, so I suspect the scientific community will need to solve that problem (ie how to throttle the publishers) before any thought of allowing versioning and revisions to papers.

As well, how would references work? Citations would have to reference a specific version of every paper, which could potentially create a huge mess.

But I totally agree that change needs to come to the way scientific papers are published and accessed, and this is one of the better ideas I've seen out there.

brainid• 15 years ago

I know of one existing journal that is pretty close to this, the insight journal: http://www.insight-journal.org/

rubidium• 15 years ago

>We spend much time perfecting irrelevant papers to get them through peer review.

Maybe stop publishing irrelevant papers? Admitted, academia has serious problems. The solution isn't open source though.

Building a collaborative framework around research would be great, but solidarity and small group research will continue to be the primary drivers of innovative research.

jorleif• 15 years ago

Academics can't stop publishing irrelevant papers, since funding is granted based on publication history. To get really novel things done researchers must keep a steady stream of publications. This is easiest achieved by making somewhat easy, incremental research to a large part. This incentive, or even rather selection, problem is, I think the core reason for this ancient publishing model.

pixcavator• 15 years ago

Good idea but why make thing so complicated? Research papers isn’t code -- they don’t have to "compile". Just write what you want in your blog/site/etc and link. If you want it all in one place, set up a wiki.

Proposal: consider scientific research papers like open source software

Related Stories