How do you routinely build up and make use of your collection of scientific papers ? As for me, I regularly search the web for relevant journal or arXiv publications (using feeds) and then add (links or pdf's of) the selected papers to my library. From this collection, by far the largest fraction (guess: 95%) is NEVER used again. For the small remaining part, I may occasionally re-read the title and abstract and have a quick look on the figures. Considerably less than 1% of my collection I have read thoroughly, in the sense that I followed each line and equation. Such cases usually occur when I am preparing an own manuscript and need to include references to the literature. This makes me wonder if I should not completely give up the habit of collecting papers. They are in the cloud, anyway. Would by interesting to learn how other people are handling this issue. - Claus Metzner

I very rarely read more than the abstract. However, I still find it worth archiving (at CiteULike) for many reasons: sharing with others, finding rapidly when asked a question (via tags), generating BibTeX files for certain topics (again via tags) when writing. Frankly, I find most academic papers incredibly dull. If peer review is what it's cracked up to be, someone else can decide whether the findings are sound. If it isn't, I'll hear about it somewhere. - Neil Saunders

Neil, thanks for pointing to CiteULike (I just subscribed). It's funny you think that "... someone else can decide whether the findings are sound". In general (i.e. unless I have the role of a referee), I also don't care too much if the arguments, methods and results of a paper are 100% sound or not. Mostly I am hunting for small reusable items (side remarks, ways of presentation, methods, "justifiers") which I can potentially implement into my own future work. A paper can be wrong in total, but nevertheless contain many useful ideas. This is another reason, by the way, why I am a strong advocate of the idea of micropublishing (see, Michael Nielsen: http://michaelnielsen.org/blog...). I would like to have an online service where I can publish my many "micro-results" that probably will never make it into a full paper (which not many people actually read, as it seems). - Claus Metzner

10 years ago during my PhD. It was the first paper about "RNA interference & C. Elegans". Just like Neil, I only read the abstracts, store the papers on delicious and sometimes I print the paper and I read it in the train just to have a clear conscience :-) I now more enjoy the IT papers. - Pierre Lindenbaum

I basically learn 95% from papers and so I try to extract the how/why from the what, which might not be relevant in my field. For me getting the "details" of a paper means to understand "why did they do what they did and how did they come up with such an idea" Conclusions vary;) - marcin

I only read very critically the papers that are near my own work or that discuss methods that I want to use (a small fraction of the things I bookmark). Aside from that I more or less trust what they say and try to keep the concepts. I really like to read, mostly because this is where I get ideas from and secondly because it gives me an impression that science is actually progressing at a reasonable pace (unlike my individual work ;). "workflow" goes Connotea - PDF in a folder - printed -read and I lose some along the way in each step. - Pedro Beltrao

marcin, Pedro: By extracting general concepts from a specific publication, you are clearly adding extra value to the paper. I think it would be great if those reusable concepts could somehow be put into a standardized form and be made publicly available in an online search engine. - Claus Metzner

I "read" in several modes after triaging by title and discarding ~ 90% of searches/rss. Abstract (50%); abstract + figures (40%); skim paper text (8%); read paper head-to-toe (1%); including methods (0.5%); read multiple times and extract utility/discuss with colleagues (0.5%). - Chris Cotsapas

My percentage of saved (citUlike, connotea or Mendeley) articles actually read thoroughly is probably similar to everybody else's (i.e., low). But I'd want to read more - there's just not enough time. So it'll have to be abstracts and conclusions, mainly and then some figures. That's why I like journal club: thorough discussion of one paper, either prepared by others or forcing oneself to read it thoroughly. - Björn Brembs

Björn, "I'd want to read more - there's just not enough time" - The increasing disproportion between the amount of worthwhile literature on the web and our limited reading capacities seems to call for a radically new mind set. I myself did not yet go through this mental phase transition, but there must be a way to avoid the feeling of regret, each time we realize how much interesting information we simply have to ignore. Sometimes I envy my students for their "beginner's mind": They simply have non idea how much literature is "out there" in the web that could be potentially useful for their project. They just start working, using what they already know. A very happy state of mind ... - Claus Metzner

I should be reading a lot of papers in detail because I'm still familiarising myself w the methods of my field, but usually never manage to read more than 1-2 per week properly. It's good to hear that reading at different intensities seems common practice. Here's a paper that addresses the growing information overload and how web 2.0 could/fails to help: http://www.pubmedcentral.nih.gov/article... via http://friendfeed.com/mycopha... - laura

Claus, general concepts are not equal to general knowledge. Say in a paper they used method X instead of Y to analyze D because Y has a specific quirk on D. So my general concepts will be "data-sets of type D are good for X and bad for Y". A reader has not only extract X vs Y, but also be able to define "type D". - marcin

I'll closely read certain papers, and the rest, I generally skim through and keep all of them in tagged folders on PC. I As a lay person, I have a limited capactity however when trying to understand the papers that come my way. Other than that, I like what Björn says about journal clubs and the like. I guess that also is sort of the purpose of F1000, etc.. - Graham Steel

I have journal club every other week, so last week ;) - Brian Krueger - LabSpaces

That could be one year ago, when I was struggling to study math by myself. Hungerford's abstract algebra is so difficult that I've given up through Chap. 3. - Juvenn Woo from fftogo

Claus - I don't know what that mindset should be: the literature is littered with the debris of so much crap that it's almost not worth wading through. I find it much more efficient to get drunk with people at conferences and swap updates on our work and our colleagues' than it is to actually read papers. - Chris Cotsapas

I've stopped using cite-u-like and starring search results in GReader, as I don't follow up on these things. Instead, if there is something I want to read enough to "Open in Papers" I might get round to it. - Chris Cotsapas

Last week. There are about 6 papers a day that come out in the tiny little subspecialty of adult bone marrow derived multipotent stem cells, and I think that's probably much worse in other fields, which is the whole idea behind getting recommendations via other people's shared items, and soon, via last.fm-style recommendation at Mendeley. The reason for storing/tagging them is so that you can find them more easily again, but also because if someone else goes looking for an expert via social networking, you want to be a major node in the network. - Mr. Gunn

I rather prefer to blog interesting papers, not in hope that someone else will read but of course I can read :) - Abhishek Tiwari

Wow, I'm impressed ! Andy Maloney trying to read, review, and understand 60 papers in 30 days ... http://friendfeed.com/andymal... - Claus Metzner

In an ironic twist many will appreciate, I reviewed two papers today. Bringing my "read" total for the last 30-day period to.... 2. - Chris Cotsapas