Slashdot Mirror
Peer-Reviewed Research Over The Web
bhoman writes "The San Francisco Chronicle (sfgate.com) has an article today about Stanford biochemist Patrick O. Brown, who helped develop low-cost DNA microarrays for gene research. He is seeking $20M to start a foundation that would fund peer-review of research papers and then make them available for free over the web, thereby avoiding the high-cost of subscriptions common in existing research publications. Predictably, some publishers seem to be warning that their publishing model is hard to improve upon.
The article mentions that a previous effort by Brown and others, The Public Library of Science garnered the signatures of 30,000 supporters, but then implies that it basically failed, suggesting that academics need the journals more than vice versa.
Sounds like Brown's idea is exactly what the web is made for."
Speaking as someone who just submitted an article to Science for review, I would have to say the primary hurdle that Brown will have to overcome is pedigree. Journals like Science and Nature have a history and an editorial board that ensure a rigorous review process that ultimately presents the best of the best and the most significant science to the scientific community. Publication of ones work in journals of this caliber are important to your career, and given their wide distribution can be critical in obtaining funding.
The implications of this are far more than simple "peer mating" and "copy editing" as one other poster suggested. Granted, there is nothing that can keep an online journal from eventually becoming the place to publish, but it will take time and a commitment to excellence that will have to be maintained for scientists to become comfortable in submitting their hard earned results to. Publication of observational science will not cut it. The implication of this is that since most scientists view Science and Nature (among a select few) as the pre-eminent journals, they will be concerned about submitting the most significant scientific results to a new online journal. Typically from what I have seen, when one gets rejected from the more prestigious journals, you start moving down your ladder of preference until somebody accepts your article. Of course results targeted for specific journals with a readership that would be interested in your results always matters and this is where online journals stand the best chance of making it as opposed to large pre-eminent general interest scientific journals such as Science and Nature.
Visit Jonesblog and say hello.
Anyway, I wish Brown all the best success, but as others have mentioned, it's a somewhat harder problem than it first seems (at least he's asking for $20 million, which is somewhat realistic for handling real peer review for a substantial number of articles - 10's of thousands at least).
What's behind this nebulous "peer review" concept, at least for us, is a complex and historically based system of checks and balances involving communications between authors, editors, and (anonymous and non-anonymous) reviewers; we're essentially a legal/court system for scientific articles. There's a lot of information-related issues in there, and information technology helps a lot (that's the part I'm involved in). But fundamentally, at least the way we do it, there needs to be a paid, responsible human being reading most communications and monitoring the process, and as far as we've been able to work out, you can't get the cost under about $500 or so per article.
Now, just distributing the papers can be done essentially for free (to as many people as would want to read for about $1-5 per article, for hardware, software, disk, network, etc.) which is what the famous physics e-print archive does so well. Of course it doesn't cost publishers any more than that to distribute articles online either - the costs are in the review part (and whatever copyediting they do), not in distribution.
You'll hear about journals now that are essentially free - this is almost always for one of two reasons:
Given the $500/article cost, the other question is does science really need this level of peer review, or can it get by with less? Well, we've already seen a couple of instances of scientific fraud that slipped by in physics in the last few months even with the current level of review - is skimping really a good idea? And is the $500 minimal cost or even $1000-$2000 typical cost per article now all that bad, compared to the typical $50,000-$100,000 research grant that generally funded such research?
Yet another proposed solution has been to publish fewer papers in those journals that receive the full peer-review treatment. Unless authors miraculously constrain themselves somehow, the only way that would save us money would be to reject a lot of things without review (because the costs are in the review process itself) - but then you've thrown out the whole "peer" process you're using to determine what's published!
So, maybe Brown has found a way through this morass - but the scientific system has a complex, little studied dynamic in which peer review as it currently stands plays an important role... if we really can't afford it (the old way) any more, we're headed into some uncharted waters...
Energy: time to change the picture.
I am one of the few graduate students who published in a journal as an undergrad, and I think you should know what goes into the publishing process (for IEEE publications, which covers almost all respected peer-reviewed computing journals).
Traditionally, you first have to have the article published in a conference, which requires for you to
1) Write the article
2) Go to the conference and present your work
Submission to a conference usually happens approximately six months before the actual conference. You get acceptance about a month after submission (or you get rejection). Most conferences have an acceptance rate of 50% or worse, meaning that they turn away HALF of the applicants.
The process of selection is done by assigning reviews to major professors in the field who are not submitting to the conference. These professors sometimes pass the review work along to some of their best grad students (this happened several times in a lab that I worked in).
After you are accepted, you send the final version, which includes any changes you may have made to the rough draft, and then go to the conference.
The next step is a journal article. This usually includes some additional fleshing out of the article. Most conference procedings are between 4 and 8 pages; journal articles can be as long as you can get it to be. You want it longer, because the longer it is, the more likely that the people reading it will understand and want to use your idea, because you can explain it upteen ways and provide numerous examples of why your [whatever it is] works well, which always leads to good things for a journal article writer's career.
That often takes a while as well. Once you submit the journal article, you get a preliminary acceptance contingent upon making changes after three months of review. Of course, once again acceptance is less than 50%, usually, but if you publish in a conference first, your chances are significantly higher than if you don't. You have an incredibly powerful idea to make it without a conference first.
The reviewers in this case are required to make a very careful inspection of the article to ensure that
1) the theories presented are useful
2) the theories make sense
3) the paper is written well enough to be readable
Reviewers are also required to find ALL spelling and grammer mistakes, and they have to understand the methods presented within the paper well enough to make a summary of the journal article. Also, reviewers are the same as before - experts in the field (college professors) who are not submitting to the journal at the time of the review.
These reviewers give you a report, accepting contingent upon meeting their requirements (or defending why you can't).
You then have to submit again and your article is once again edited for approximately 6 months.
If you REALLY rush, this entire process takes one year, however realistically, it usually takes two. (Yeah, I started the game as a junior in college).
Now, I don't care if we do this online. IEEE has a research engine called IEEE Xplore, which is often purchased by research institutions such as Universities. It has the whole database of IEEE publications within it.
But I don't know how to get a much better peer-review process than this; its pretty darn strict. So professors can't just whip of papers like nobody's business - they really have to put some work into it. If they have a lot of papers, it means they've done a lot of stuff that at least six other experts (for each paper - sometimes more) believe to have merit.
Mod me down and I will become more powerful than you can possibly imagine!