Slashdot Mirror

By the way: the interface I was thinking of is the open text mining initiative (OTMI), abandoned by Nature. Nice idea, kept the publishers relatively happy but it didn't catch on (see brief critique in comments section).

While I think that from a financial standpoint, libraries would save money publishing articles themselves, they don't have the editorial ability of a major publishing house. It's the kind of thing that doesn't de-centralize well, either, because you need a limited amount of "restriction points" to effectively filter out the crap from the submission stream. That then enables people who produce material that makes it past the editorial and peer-review filter to claim that as an accomplishment when applying for grant and tenure.

Here's the analogy for the slashdot crowd: it's like article submissions are electrons. You want to get a certain current by using editors as resistors(and believe me, they do resist). Now what happens to current as you add more resistors in parallel?

The solution to all of this would be for libraries to require all submissions in some kind of markup like LaTeX(to enable Semantic Web goodness- in the biological and medical fields, submission as MS Word files is not uncommon), not produce a printed version at all, and publish the citations with abstracts in a central index, similar to Pubmed. Now, instead of an article in Science or Nature having more weight than an article in the eskimo journal of snow science, an article would be ranked by the number of views, and by the number of non-self citations it receives from subsequent articles.

However, there are two problems with this. One is that it would take a longer time for an article's worth to be evaluated that way than with a dedicated staff of editors working on it, where the value accrues immediately upon publication(though this is all debatable). The other, more serious, problem is how to find important, relevant work with no one filtering out the crap for you. Imagine slashdot with all submissions accepted, and no moderation or comment threading, just one long stream of post after post after frist psot! So maybe moderated comments would be enabled on the citation indices, and you'd have something like the grand thing we have here, with a mix of people acting selfishly and altruistically. That's not that different from the system we have today, I don't guess. Who's going to decide which articles make the front page? A small pool of people who are expert in their field? Editors?

I subscribe to some RSS feeds of pubmed search queries for keywords important to me, and each query is updated with several articles each day. This is for articles appearing in peer-reviewed journals(again, many of which editorially triage articles before selecting ones to send out for review).

Keeping up with the reading of this pre-selected set is a big task already. It would be more than a full-time job to read an unfiltered set, then moderate or comment upon each. I know some fields are even busier than mine.

So, I would love to see the current system change as much as anyone else, but I don't have the time or funding to put my efforts where my mouth is, nor does anyone else who wants to actually do research instead of just read about it. The editors are doing a hard job, and much as I hate them sometimes, I respect what they're doing.

Personally, I like using pybliographer (as was already mentioned) for my thesis. But also check out JabRef which is written entirely in Java. So if you ever needed to go back to windows and still want to manage your BibTeX entries, JabRef may be a good option. Be careful moving back and forth between different bib managers because each one has its own convention in created keys (by default)--thus, the key for one entry in pybliographer will be different than the key referring to the same entry in JabRef. IIRC, both programs allow you to redefine how you want your keys to be configured, so if you define your own key structure, this problem is minimal.

If you're on a Mac, try out BibDesk. This user has a screencast (flash video demonstration) showing you how to export "BibTeX data and adding it to a BibDesk library, autofiling and associating a PDF file, adding the citation to a TeX file, then formatting a bibliography."

Here's where you're failing with respect to Google Scholar: Your crufty old interface(s).

Really. Just as Google Maps has superceded Mapquest, pretty much entirely due to their superior interface, Google Scholar is going to make WoS irrelevant as a literature search tool. All they have to do is implement a slick citation navigation system(tree-based or some such, for which the technology already exists, check out the TouchGraph applet) and it's done.

Currently each different database is better for a different use, but that wont last, so don't kid yourself. Modernize your interface or become irrelevant.

Have you heard of Connotea? You can grab bibliographical info from Pubmed, HubMed, and many other indices directly into your Connotea list, and output in .RIS, so you can import into RefMan or whatever. The eventual goal is to move totally away from Thompson ISI and their crufty old products, but until someone comes out with a Journal Style formatting package, we're stuck with RefMan's heinous old interface, at least for now.

So if you click through any search result, you can grab the citation info, and then pass that on to EndNote or whatever, but hopefully we'll soon not even need to do that. Give Connotea a try, you may find it more useful for at least making the list.

What I'd like to see is better cooperation with dx.doi.org and more OpenURL support, but I guess that is mostly up to the libraries. I'm going to try to talk my school into registering their resolver with Google, so it knows which library I get my access from, and hopefully Open Access continues to spread.

Since it's really all about the interface, now we need good forward and backward citation navigation. Tree based approaches, like the one they use at Hubmed, is nice, but the implementation is still a little rough. I would think Google, with their AJAX skillz, could do something much nicer, ala Google Maps.

The relevant paper came out ages ago.

But I see how the game works now:

Work really hard to discover something vaguely suggestive. Publish a colorful PDF with a description of the work that, really, makes no sense. But (don't forget this part!) explain that it's going to Change Everything. Then wait.

Sooner or later this Roland character will publish it on Slashdot and hordes of the most influential Lord of the Rings fanatics in the world will be all over it.

Sit back and let the money roll in!

Google is not quite anonymous.

This "cloud" thing is kinda nifty, kinda goofy. Some of the images are crappy but the technique is interesting. Maybe it's a good idea that just needs expanding. You need better images to start with, if you want a better end result.