Authors: Xiaolin Shi, me, Belle Tseng, Gavin Clarkson

Background
Human knowledge and innovation are recorded in two media: scholarly publication and patents. These records not only document a new scientific insight or new method developed, but they also carefully cite prior work upon which the innovation is built.

Methodology
We quantify the impact of information flow across fields using two large citation dataset: one spanning over a century of scholarly work in the natural sciences, social sciences and humanities, and second spanning a quarter century of United States patents.

Conclusions
We find that a publication's citing across disciplines is tied to its subsequent impact. In the case of patents and natural science publications, those that are cited at least once are cited slightly more when they draw on research outside of their area. In contrast, in the social sciences, citing within one's own field tends to be positively correlated with impact.

=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=

The paper came out last week. PLoS One has these neat features where readers can rate the article, leave comments in general, or comment on particular parts of the text. So far... nothing. I'm a bit bummed. Either no one has noticed our potentially controversial article, or it's not as controversial as I had assumed.

Recently Prof. Hacker had a post about email productivity. Inbox 0 and such were touted as lifesavers. I've personally "implemented" the Trusted Trio system and have been using it for well over a year. The idea is that you read each email once, answer it immediately if it will take < 3 minutes, put it into the "action" folder if you need to answer it but it will take a little while, into the "hold" folder if you can't really take action, but you'll need to revisit it, and otherwise just archive/delete it. This way your inbox spends most of its time empty.

The reality for me was a bit different. First of all, the #1 email lifesaver is the "not to me" filter. If it doesn't have my email address explicitly in the to or cc field, it goes into a "mailing list" folder, which raises no "new mail" flags and in general waits for me to get to it, rather than calling attention to itself. The other time saver is the nostalgy add-on for Thunderbird that allows me to sort email with keystrokes.

Now my "inbox" is rarely 0. It usually has 50-100 messages. Most of them require some action from me. Why not just file it into the "action" folder? Well, because the "action" folder is a scary place. Instead of visiting it every day regularly like all the email self-help articles suggest, I go there only when I'm feeling very brave. It has some reviewing assignments, but also many, many emails with attachments. Students requesting feedback on projects & write-ups, researchers who want to call attention to their papers, or even worse, books. Other researchers who are proposing joint books, projects, proposals. Or helpful individuals who are sending helpful resources or papers that I should check out. All very worthwhile emails. But not ones I want to worry about right now.

Normally an unfiltered inbox is diluted. Such messages occur only once every 10 or 20. However, the action folder is nothing but this high-time commitment stuff. So I only really place messages in the action folder if I'm pretty sure I won't take action for a few days... or weeks...

What about answering email as soon as I read it? Well, I find (and this has been acknowledged by the email gurus as well), that the more you write, the more people write back! So it follows that if you reply within minutes, your incoming email will start arriving at a higher rate. And that seems self-defeating.

Not so with Prof. Tetlock. After an entire chapter motivating the problem and setting out the methodology, Tetlock first establishes that humans (experts even) are shockingly bad at predicting the future (worse than simple statistical models, in any case). He then describes how different personality types (hedgehogs and foxes) perform differently in prediction tasks, and then another several chapters dissecting the results (how do the participants themselves justify their mistakes? what other factors correlate? etc.).

I'm glad to have "read" this book. Even though I do a lot of academic reading both out of curiosity and assigned (see previous post on reviewing work), these are also bite-size pieces, little pre-packaged juicy bits with one or two results that can comfortably be summarized in a concluding paragraph.

What did it take to read a real book? I had to put myself in situations with no internet access, on vacation, with no other books I'd want to read, preferably on a train, or sitting in proximity of others who could appreciate that I was reading a serious book and therefore upgrade their opinion of my work ethic (and cook me food).

Now that I've read the book, I need to figure out how to cite it. You see, I'm afraid I might be a bit of a hedgehog. This book doesn't quite fit into a paper I'm writing on my own version of hedgehogs vs. foxes...

It all boils down to a feeling that it's the students, and not we the faculty who have the inside scoop on courses. Ages ago, while getting my PhD at Stanford, I took pretty awesome courses in CS, stats, EESOR and physics, thanks to recommendations from other students. And I would be able to recommend those courses, because I spent many hours toiling through them.

But now I take no courses. I may know that a colleague is a good researcher or a good speaker, but do I know things about their courses past what is listed on the syllabus (if that)? Sometimes, a bit, if an instructor boasts about an activity, or an advisee mentions their experience with a course. An even bigger challenge comes when students from other departments ask me about courses similar to mine, but in their department. Or students from my school asking about courses elsewhere... I then try and remember what other advisees had told me, or sneak a peek at my not-overly-useful network diagrams. But mostly I tell them "ahem... have you thought about talking to other students?".

In theory all the content has been cleared for copyright and now has a creative commons license, meaning that anyone can use and adapt it for their own purpose. It has slides, labs, datasets, demos, student projects, everything.

I talked about it briefly earlier this week at the SocialNets in Education Project workshop @ Duke. I also found out that there was interest in my DRAT course materials (SI601), but it will be another while yet before I undertake another OER conversion :p.

Just last year, I started using Matt Jackson's text, "Social and Economic Networks" in the PhD-level networks course I teach. As the title suggests, it is heavily econ flavored, which puts a nice emphasis on game theoretic models of network formation, and games on networks. But it also includes excellent treatments of other topics, such as diffusion and search. And the problems at the end of the chapters have had both me and the students scratching our heads and more importantly tinkering in Mathematica.

Two news books are about to appear.

The one I've been reading of late is by Kleinberg and Easley on "Networks and Strategic Behavior", and it is 1/2 about networks, 1/2 about game theory, info markets, and other neat topics. Aimed at undergraduates, it explains the subject matter so clearly, so eloquently, so seductively, that it brings tears to one's eyes (tears of joy, but also of envy that someone is able to write like this). It should be available by this fall.

Mark Newman's long awaited textbook is also supposed to hit the shelves sometime soon. I don't know when exactly (some people don't like being asked how their book is coming along), but he will be using it or a preprint version when he teaches CSCS 535 again this fall. It will likely be aimed at physics graduate students or advanced physics undergraduates (or students with a similarly strong mathematical background). I expect it will become the definitive volume on the topic.

Is that fair? 7:1? Could it be more like 3:1? I already seem to say no to 1/3-1/2 of the requests I get. Why do I get so many requests? Maybe it's that as a generic "networks" person I'm of use as a reviewer on interdisciplinary panels, in CS venues, Physics venues, even some stats and social science venues. Some swear by farming our reviews to students. I've done that for approx 10 reviews this year (on top of the 81 I did myself). But it seems like a drop in the bucket compared to the total load.

I used to look forward to some reviewing. Getting a paper authored by a respected colleague in the field. Now I often turn those down. About a year ago, I had been asked by a prestigious journal to review a paper by someone whose work I admired and whose sense of humor I really enjoyed. But I felt that this particular paper did not pack the punch that I expected of papers in that journal, and I did not recommend it for publication (on both rounds). My name and comments to the editor ("this would be fine for a lesser journal..." (ouch!)) were accidentally released to the authors. Although I know the probability of such an event occurring again are small, even the thought of rejecting a paper by someone I chat with at conferences, or who is likely to tell me about the work even before I review it, is somehow uncomfortable.

I consider myself a middle-of-the-road reviewer. Especially with the conference reviewing systems these days where you are encouraged to look at others' reviews of the same paper for purposes of discussion, I find myself on the accept side of the middle. Yet there are plenty of papers that I don't recommend for publication. And it makes for sad work.

At this point I'm doing so much reviewing, that I find myself asking the authors of one paper why they didn't cite another paper I had reviewed a short while previously. At 7 papers/proposals to review a month, I honestly don't think I read a comparable number of papers I "want" to read or "should" read for my research. Not to mention papers given to me by colleagues, that I put on the back of my reviewing queue and often don't get to.

We posed the question of how resources could be allocated in a coordinated way to maximally stem the spread of infection. Umanka and I tried several stategies - targeting hospitals with the highest degree (number of hospitals they trade patients with), highest betweenness (they are on the "path" between other hospitals), and a greedy allocation based on the number of beds infected at each hospital and downstream from that hospital.

The results are here. Both figures show hospitals as nodes sized by the number of ICU beds they have.

This shows the number of resources allocated by hospital (gray = none, blue = few, red = many).

This shows the relative benefit of a random allocation vs. targeting particular hospitals. (blue = hospital unlikely to become infected, red = likely to be infected)

Info on her thesis:

THE STRUCTURE AND DYNAMICS OF INFORMATION SHARING NETWORKS

Information flows are produced, carried, and directed by information sharing networks. And the evolution of the structure of such networks and the way information diffuses are affected by one another. This thesis studies structural features of information networks and their relationships to information diffusion...

* easily subsetting nodes according to attributes
* getting node labels to jiggle around until they no longer overlap
* drawing curved arcs (and controlling the curvature)

Available as a preprint: "On the informational properties of trading networks"