Increasing Public Access to the Results of Scientific Research
petitions.whitehouse.gov
petitions.whitehouse.gov
The scientific publishing industry does not write, review, or edit any of the content it publishes -- volunteer scientists do that. It literally takes a pre-formatted PDF or Word document and sticks it on a website, and maybe into a print edition. It charges multi-thousand dollar fees to authors to publish the articles, fees to institutions to access them, and still litters the website/journal with ads. And their profit margin is in the 40% area.
This industry is an absolute scourge on human progress, and a 12-month embargo does not go nearly far enough. 12 months is an eternity in science.
I will say that the publishing industry does do some very minor but useful things, like linkification of references, management of submitting the publication record to PubMed, etc. But the cost is exorbitant, and the government could provide these services for a fraction of the cost.
Edit: Oh, and journal prices are rising much faster than the CPI. While at the same time most have reduced or eliminated a print edition, and increasingly require reformatting to be done by faculty. Hmm...
The average journal profit margin is 20-30% and that's roughly the same whether you're talking about an open access journal like PLoS or a closed access one. They also have roughly the same operational (staff, etc.) costs. Most new journals tend to be run at a loss for a number of years being subsidised by the more successful ones.
Marketing is another big cost, journals don't magically build reputations, they have marketing budgets to achieve mindshare and to establish a reputation.
There are good reasons for open access, but lets not pretend that journals are cheaper to operate than they actually are.
Now that systems like ArXiV exist, researchers and taxpayers should not be paying for marketing and typography.
But the real reason journals persist is because readers and promotion/tenure committees want a credibility filter. A really good technological solution to this problem hasn't been found (AFAIK), but I don't see that it would be too hard to build some algorithm to rank articles according to author prestige, institution prestige, number of citations, reader interest, etc. If people want a "hard" metric of article impact, they would be better off using number of citations than journal impact factor anyway.
You're hand-waving away one the driving forces not only of the publishing industry, but much of the whole economy. Signals, filters, and branding account for the existence of everything from Louis Vuitton to hiring practices based on school prestige to Asian parents that push their kids to get into MIT. Signaling makes the world go around and you hand-wave it away by saying it shouldn't be hard to replace the branding with "some algorithm."
My entire point was that better and more objective ways are being developed that can signal this same message, while cutting out the cost and inconvenience of the journal middleman. The exact recipe for these better ways is still an open question, but I do know several people who are working on it, and I listed several specific examples.
Branding no doubt plays a role. But there is already a major trend towards downplaying branding in favor of more objective metrics like impact factor and the H-index. And I do hope that science can be held to a slightly higher standard of objectivity than the general economy (maybe that's too idealistic?).
Doesn't sound to idealistic to this science-oriented engineer. With the field of scientometrics, the feedback loops look like they can be closed entirely within the domain of science instead of needing to extend outside of it to be closed.
There is no reason to think that actual experimental studies can't be brought to bear on the matter of what ranking/organizing metrics of papers/research best (for various definitions of 'best') help scientists do their work. Also no a priori reason I can see why various fields would converge to the same measures; or that those most used within science would be the ones that the press, students, people external to the fields, and such would find most helpful.
This suggestion is similar to the NIH public access policy (adopted in 2008), which requires the results of NIH-funded research to be made freely available within 12 months of publication. The new memorandum gives agencies some freedom in how they respond - they don't need to adopt exactly the NIH policy - but it is clearly in the same spirit.
Here's an analysis from Peter Suber, a leading advocate of open access: https://plus.google.com/109377556796183035206/posts/8hzviMJe...
Suber focuses on the connection to FASTR, a major piece of open access legislation introduced into Congress a few days ago. Broadly, FASTR has a lot of overlap with the White House directive. FASTR would require every Federal agency with a budget over 100 million to adopt an open access policy. A significant difference - and one that I expect is of interest to HN - is that FASTR has provisions to enable text mining. That would potentially be of interest to some startups. Much more info here:
http://cyber.law.harvard.edu/hoap/Notes_on_the_Fair_Access_t...
If you'd like to support open access, take a few minutes to look at the Alliance for Taxpayer Access's (ATA) call to action on FASTR:
http://www.taxpayeraccess.org/action/FASTR_calltoaction.shtm...
It is also of interest to follow the responses from John Wilbanks and Heather Joseph, two of the authors of the petition (and long-time advocates for open access):
Wilbanks notes that the memo covers research from "NSF, Ed, EPA, NASA, USDA, HHS, Commerce, Interior, Defense, Energy, Trans, DHS, Ag, State, Smithsonian". He also implies that while this memo is great progress, it falls short of a full open access mandate enabling reuse and text-mining of content.
I don't think they could. If they've already signed a contract to fund a project, they can't just decide to change the rules. This will only count going forward.
http://www.xconomy.com/san-francisco/2013/01/15/california-t...
http://tech.fortune.cnn.com/2013/02/21/the-california-tax-th...
After 1996 you could presumably take it as read that digital rights were included when you transferred copyrights.
Honestly, what value does the publishing industry provide? Search? Google, Microsoft Bing, and any number of other search providers could do that at the fraction of the cost. Peer review? No, that is performed by independent scientists, who are either not paid, or paid a negligible fee. Editing? Nope. Scientific journals are not edited for content or even proofread; I know a few professors who found that out the hard way after discovering typos in their own articles. Typesetting? OK. I'll give them that one. So, for the cost of thousands of dollars to the original researchers, on top of the outrageously high licensing fees paid by people who wish to read the articles, we get papers that have been typeset. Hoo boy, what a great deal.
You can't compare the curation you get in a proper journal to the shitsplatter of your typical search engine results.
A very good example of how crappy search technology is for this sort of thing is to do legal research on Google Scholar versus Westlaw. The indexing, head notes, directories, and topical summaries you get from Westlaw make Google Scholar look like a toy. And why should we expect differently? We're comparing the product of human labor to the results of extremely primitive AI technology optimized not for research but for making sure that people can quickly find the latest celebrity nip slip photos.
E.g. if I type "economics of deflation" into Google, I get: wikipedia, a youtube video, "tutor2u.net", and investopedia among other things. Businessweek and the Economist show up on the bottom half of the first page. The first several pages of results are full of Austrian and Mises gibberish--lot's of "counter conventional wisdom" sites. This is why I don't go to Google to research economics. I don't trust it to pick quality research.
So where do I turn? "tutor2u.net?" No. I have no idea whether they can be trusted or not. So I use signaling and proxies. I look for things that have "Harvard" or "London School of Economics" domains. The folks at Harvard and LSE long ago figured out that basic dynamic, and that's why they use their names to publish journals. Is it a hard job other people couldn't do? No. But the journals exist because we trust Harvard to do the job, and we don't trust "tutor2u.net".
My wife actually hates the NIH open access policy for this reason. She argues a lot on the internet about pediatrics/child birth. She gets crazies who peddle crackpot theories by citing to sketchy papers that happen to be hosted on an NIH domain. They assume that by hosting those papers, NIH vouches for the content in those papers, but of course that's not the case.
What exactly do you mean by curation? Editing and peer review IS what determines what is published in the journal. If the paper passes those, it gets in-- Elsevier and Springer do not make a habit of telling editors-in-chief, "sorry, this paper you accepted doesn't cut the mustard". As for what is "featured", the order in which papers appear, etc., this is negligible even if the publishers do decide it (probably not), since journals are not like newspapers, not intended for cover-to-cover reading.
It sucks because it uses extremely primitive technology (even if it's start of the art, it's primitive in the grand scheme of things) to perform indexing. Like the web search engine, it relies heavily on things like citation counts, which makes it nearly useless for researching any complex point of law because your search results are full of highly-cited cases that might cover the relevant subject only in passing or in boilerplate. The results are shitty in precisely the way you'd expect them to be based on the kind of algorithms Google uses.
The limitations of Google's search algorithms (search algorithms generally) aren't apparent for web searches because by and large everyone is searching for the same things. Google is great at helping people converge onto popular resources like that. It falls apart in situations where you might be the only person searching for a particular issue. The basic problem with search is that computers don't understand documents. The whole idea behind Page Rank was to side-step that by incorporating human intelligence--that is the human intelligence encoded into the graph of links. Further refinements have been based on things like taking into account human behavior in response to search results, and aggregating over all the millions of people who search for the same things. Without all that human intelligence, all that context, you're left with a computer that just sees a bunch of text, instead of background, dicta, holding, etc.
But that's no reason to lock up the publicly funded research papers themselves.
There is always the argument of keeping the editorial quality high and that competition among publishers enables better scholarship. The truth is, however, that science politics have long now invaded most high-impact commercial journals and publishers to the point where this argument no longer holds.
Most scientists don't really have strong feelings about open access, since for academics access to journals is practically guaranteed through their workplace subscriptions. But it is possible that most scientists are not aware of the wonderful possibilities that open access enables, such as text mining and other sorts of meta-analysis. Not to mention that everyone would be able to read any study on their phone at any time/place.
Regarding:
"Federal agencies that directs those with more than $100 million in research and development expenditures to develop plans to make the results of federally-funded research publically available free of charge within 12 months after original publication"
I really hope they do something easy like a REST/JSON API.
No. Notice that this petition met its goal a year ago, and they never issued a response until now. It's more likely they were designing a plan and waited until they were done to respond to the petition.
I sure hope that this petition wasn't the only reason why they decided to issue these orders.
Holy newspeak. These are companies, they don't make contributions out of the good of their heart, _they extract value for their shareholders_, who are neither the taxpayers nor the scientists. The ridiculous gross margins and nonexistent competition suggests they do so in what can only be called rent-seeking.
Though I'm having some doubts(naturally!) about: - the $100 million threshold -the overall implementation (lets see on Jan 1/July 1 next year) - the effective length of which excuses are held valid, and which are not. I feel the white house should take clear stand, and declare which research groups were excused and why.
Then again, I must congratulate US govt on this. Something like this is ages away in my country.
the Journal-Paywall system?
Hence this research will have to be made available freely.
why put limits, like 100 million, federal/state, etc???
were no lessons learnt from Aaron's case?
The journals might still have paywalls on their own sites, but what's the impact of a paywall when the exact same paper is on the open web where Google can index it?
The allowance for a 12-month embargo is a bummer -- I'd rather research be available to the public immediately when it's published -- but there's a ton of research out there more than a year old, so this is a great advance.
A sibling comment linked to Peter Suber's reaction: https://plus.google.com/109377556796183035206/posts/8hzviMJe... Suber has been following this issue for a long time and written more about it than anyone, so his interpretation carries weight.
Doesn't that shift the burden on agencies rather than publishers, allowing them to publish on paid-for journals but then making them available after a year by paying some open access fee to the publisher?
Also, if publicly funded research is to be publicly available, shouldn't it be so right away?
I just feel that this measure is incomplete and tries to keep at least part of the publishers' inflow of cash.
They oughtta require that it's in the hands of the general public within twelve freakin' microseconds of original publication.
Twelve months makes no sense.
The only stakeholder that should be considered is the U.S. tax payer. We funded the research, whether we wanted to or not, under penalty of imprisonment or death, and in a manner that violates the intent of the Constitution. Why then, are we prohibited from seeing the results of our investment?