One of my collaborators, Mark Newman, has done some work on cartograms over the last couple years. (This is a way to take a map, such as one of the United States, and squish the sizes of, say, the states to reflect their relative population or whatever other category you want.) He is involved in a project called Worldmapper, which is doing this for the countries of the world can give a nice, easyily interpreted visualization of which countries are using the most energy, etc. Some of Mark's other cartograms are also available here.
Some of them actually showed up in media coverage of the 2004 Presidential elections and subsequently on CafePress t-shirts. For example, you might have seen some of these.
Mark's talk on this topic is especially amusing. At a talk on random networks, I gave a talk on Congress and I stated at the beginning I was going to pretend I didn't have any political opinions during the talk. (Mark is one of the coauthors of those papers.) The next day, Mark started his talk by saying that unlike me, he was going to not pretend that he didn't care about the politics involved, and then he proceeded to slam the Republicans several times during his talk. Mark also recognized my stuffed Cthulhu immediately when he visited Georgia Tech, so there are clearly a lot of compelling reasons to respect him (beyond his science).
I've collaborated with Mark a couple of times now, and he's both brilliant, hard-working and a generally nice guy. My one complaint is that he's too busy for his own good (I imagine that he would agree with that accessment), but I guess that comes with the territory. In my mind, he's right up there with Jon Kleinberg.
ReplyDeleteBut, on a more important note, where oh where can I get a stuffed Cthulu doll?!?
It's true: Mark writes papers like a rabbit on amphetamines.
ReplyDeleteSaying he's too busy for his own good is an understatement. I tend to spread myself really thin, but it's nothing compared to what he does. (For the stuff on which we've collaborated, he always mentions the queue he has when I send him a draft. Then when he gets a chance to go through it, he is able to very quickly see where the gaps still are. He did once thank me for a useful conversation in the acknowledgements in one of his papers, and to this day I have yet to figure out what I said that turned out to be helpful.) He is indeed up there with Kleinberg and he'll get his due recognition eventually. (Kleinberg got a bit of a younger start when it comes to getting a tenure track job.) Of course, I don't know if there is an equivalent of the Nevanlinna prize for physicists and I think that many of Mark's fellow physicists don't consider his work to be "physics," so they may not be so inclined to give him awards that his work would merit. (Ah, politics...)
I also especially like, for example, comparing what Mark writes in his paper about people like Zell Miller to what he says in person. It's great. (Check out the footnote in his new paper on community detetction via eigenvectors of the modularity matrix, which is an absolute tour de force. I've gotten several ideas from reading paper, and we've already implemented a version of it and have gotten really nice results on the Congressional committee networks and legislation cosponsorship networks.)
I bought my Summer Fun Cthulhu at Games of Berkeley. (I was advised that I needed to feed it at least one soul a week, which wasn't a problem when I was teaching classes.)
You can order stuffed Cthulhus online here.
I think physics is experiencing a difficult time in its history - on the one hand, you have things that are clearly physics, like a lot of empirical and theoretical condensed matter research, but then you have things like string theory, of which a lot is probably better termed meta-physics at this point. But, more to the point, it's all this "interdisciplinary" stuff that I think poses a bigger problem for the field. Being not-really-a-physicists-anymore, I can say that what Mark does is definitely physics, although it's also a bit of computer science, statistics and probability theory. But, to the condensed matter people, and maybe even more so to the particle physics people, interdisciplinary stuff like complex networks probably seems more like sociology, applied mathematics, and computer science. Yet somehow, Phys Rev E is one of the main places to send networks stuff!
ReplyDeleteI was at the Network Science conference at Indiana earlier in the summer, and they did a whimsical little game where all the participants were given two strands of ribbon. We were to hold onto one end of each of our strands, and then give the other ends to the two people who's papers we had most recently read. Barabasi, Vespignani and some others were there. But, it was Mark who ended up being the highest degree vertex! His fellow physicists may not appreciate his work, but it seems that the networks community views him as one of the (if not the) most important people in the field.
Although the "standard" topics in condensed matter physics are basically nothing more than dotting i's and crossing t's right now. [[looks around the building --- this building has our condensed matter group --- to make sure nobody is watch me type this]] Trying to use BECs (and other stuff with cold atoms) as a mechanism for studying some condensed matter phenomena in a particularly controlled manner is extremely interesting, however, and a lot of the condensed matter people (including the ones at Caltech) are going in such directions.
ReplyDeleteActually, I think the best term for string theory is "math." There is a lot of incredibly good math in that research, but its practitioners are so great at selling that stuff as something more that even most of them believe. (Whenever I tell anybody in that crowd that they're really doing math, they get extremely pissed off.)
Phys Rev E is popular for networks because it was already the place among the APS journals to put things about nonlinear dynamics and "complex systems" (a field which IMO ought to be called "applied statistical mechanics," if you want to use a meaningful monicker). PRE is partly a physics journal and partly an applied math journal.
Its not that Mark's fellow physicists don't appreciate his work per se, it's that his fellow physicists don't appreciate fields like nonlinear dynamics, complex networks, etc. (which are all inherantly interdisciplinary) as part of physics. Most of them don't appreciate the work that anybody studying networks does. I've even heard experimental nonlinear dynamics people lament the fact that they have to convince traditional-minded physicists that they are doing physics and it's usually only the theorists who are forced to do such things! :) It's not
I think Mark is more of an applied mathematician than a physicist in terms of the problems he studies, but the labels are arbitrary and things like "physics" and "applied mathematics" are far more useful terminology for states of mind than they are for any specific set of subjects. I approach a problem in a certain way and my degrees are in applied mathematics, so that's what I consider myself. Mark is trained as a physicist, so while I believe he thinks of himself as more of an applied mathematician at this point, he can certainly fairly label himself as a physicist when it's convenient to use that label.
It would be nice to dump many of these labels entirely, but of course that's not realistic. I am in between what the US physics and math departments want and there are very few applied math departments, and I feel the effects of this very acutely on the job market. (I get annoyed when, say, a number theorist thinks I shouldn't be hired in a math department because he's never heard of the journals in which I publish -- and that includes PRL! I have been told by multiple people during interviews that they've never heard of Physical Review Letters and these people have a big say in whether I get a job offer -- and because my style isn't theorem/proof. I'm ok with the nonlinear crowd who are qualified to judge my work (and these people do know about these journals) but when it comes to some person studying some random other subject in math, you get ridiculous things like this. (There can be analogous ones in physics departments, but the thing I get asked there is that they wonder what physics courses I can teach because I don't have a Ph.D. in physics.)
There are some very messy things that interdisciplinary people hace to deal with because of the insistence on labels.
Oh, and check out who the most important node was in the network of networks. :)
ReplyDeleteI'm not surprised at Mark being the highest-degree vertex there, although one may wish to have some scaling by number of papers. (You can't just get rid of it completely, but I think there needs to be some correction for volume of work because of the particular choice of the metric in this case.) In particular, I am wondering if (hypothetically) Mark writing twice as many papers as, say, Barabasi (I very much doubt this is the actual ratio) would be the deciding factor in a small difference in degree in the example above.
Maybe there is no controlled way to do this and one should just be aware that there is a bias in this direction?
Applied statistical mechanics is definitely an important part what I would call "complex systems", but doesn't it seem ironic to you that most statistical mechanicists don't know the first thing about doing real statistics? Cosma Shalizi is the notable exception to this trend, of course, but I think that's why he's faculty in a statistics department, rather than in a physics one.
ReplyDeleteAnother thing that sits uncomfortably with me about statistical physics and complex systems is that I'm not sure that a lot of the "complex systems" stuff that gets published in physics journals really belongs there. Shouldn't a study of terrorism go to a political science journal? (Which, incidentally, is where my study on that topic has gone.) And, a study on social networks go to a sociology journal? The answer doesn't need to always be 'yes', but a part of me wishes that it was that way a little more often than it seems it is. There's a lot of really excellent work that does show up in PRL, PRE, etc., but it just seems a little problematic to me for physicists to be ostensibly doing interdisciplinary work, and then have it be reviewed by physicists, published in a physics journal, and read basically only by other physicists. I'm not sure what the solution is. Places like Nature, Science and PNAS seem like more interdisciplinary places, but I think they're really just glorified biology journals at this point.
Computer science is, I think, only marginally better than math and physics in tolerating interdisciplinary people. Which, were I less cynical about this sort of thing, should seem surprising to me, since it is arguably one of the handful of disciplines that are revolutionizing biology these days. Just like physicists look at Mark's work and wonder what he's doing in a physics department, I'm pretty sure that had I not been careful to pick sympathizers for my dissertation committee, I might have gotten some awkward "How is this computer science?" questions. As a guy from the CRA (a trade group for academic CS people) once put it to me like so, everyone loves interdisciplinary research, so long as some other department is funding it.
There may be hope though. Kleinberg getting the Nevanlinna prize is a big coup for people who ask interesting interdisciplinary questions about the real-world, and want to get support for doing so.
Barabasi's output (number of publications per year) is about the same as Mark's, actually. Mark writes more solo papers, and has advised many fewer students, so perhaps he deserves some additional credit for that. One notable difference between the two is that Mark is much more of a methodologist (so, yeah, applied mathematical physics, with a hint of algorithms), while Barabasi is a lot more of a biologist than a physicists these days.
I certainly wish I knew more statistics (and graph theory) than I do. I have been accruing more such knowledge by working on problems that require such knowledge, but I wish I had had more of it in the classroom back in the day. I think many of the stat phys people working on networks are gradually picking up knowledge of some of the more traditional statistics.
ReplyDeleteI think it would be nice to have a 'complex systems letters' sort of journal. Places like Physica A and J Stat Mech are available, and they include applied stat mech as part of their purview (which is also why we see this stuff in PRE). I think some of the papers covering methodology/theory can go there, but the ones concentrating on the data sets are often better elsewhere. We specifically put our Congress paper in PNAS because we felt the best way to have an impact was to try to get it in there (because a poly sci person won't read PRL). We are now starting to collaborate with political scientists, so our choice in that case happened to be a good one. We submitted the sequel to Social Networks, although they take forever (which I believe is one of the reasons more of this stuff ends up in places like PRE---the social and political science journals take so damned long to referee papers). The dynamical systems journals sometimes have some networks stuff because that community has adopted even the part of this business that isn't technically a dynamical system (pretty much because many people in this community have added this area to their repertoires, and of course that's where I come from). By the way, PNAS is more open to papers in a variety of areas than Science or Nature (where I agree with your criticism, though I would like to get papers in them anyway because of the prestige and [especially!] visibility).
It will indeed be interesting to see the effects of Kleinberg getting that prize. I can certainly envision these areas being taken more seriously by people in the mainstream. We'll see if this comes to pass.
I just prefer Mark's work to Barabasi's. I have found it to be continually insightful. Of course, I'm sure a lot of that is correlated with my learning about this stuff from that crowd from the very beginning, but when I hear Barabasi suggest that a power law is the end of the story (as opposed to the beginning of or a very small part of the story) in a general colloquium, I want to scream because that's just sending the wrong message. (And I know he knows that's not true, but he oversimplies too much when he gives his talks.) I've occasionally gotten ideas from reading Barabasi's work as well, but not as many. And some of those really came from prior authors who aren't the salesman he is. That is what I think his biggest contribution has been. I have not read his more biological stuff because my own work hasn't really gone in that direction.
A Complex Systems Letters journal would be excellent. And yet, doesn't it seem to you that PNAS has sort of already become that? Much of the "complex systems" stuff that appears there is really quite good, I think. (I don't think all the best complex systems stuff ends up there, but maybe it will in time.) Or rather, the fraction of it that is poor is much lower than at places like Science and Nature, who apparently have a hard time finding good reviewers for this kind of stuff. I guess this is a place where the interdisciplinarity of complex systems style research works against it - it would be hard to persuade people from so many fields who are all actually working on complex systems stuff to stop sending their papers to their native discipline's journals and start sending it to this new, untested journal.
ReplyDeleteYour comment about data studies more appropriately going to a non-physics journal are, I think, right-on. But it seems that a lot of the modeling ideas that end up in Physica A, J Stat Mech and PRE should also go into a non-physics journal, too. That is, models matter to the data people, too, but I doubt the various data people read the physics journals where physicists publish models. Perhaps, deep down, I sympathize with the physics traditionalists about the physics-ness of this stuff...!
The arxiv is, I think, having a strong positive effect here, in many ways. For instance, via our paper on terrorism, I've discovered that there are a fair number of technically sophisticated political scientists who watch what gets posted there, and likewise for some sociologists. It seems possible to me that in 10 years, complex systems people will be less united by the set of journals they publish in, than by the way they keep up with each other's work, i.e., online repositories like the arxiv. Maybe it's already like that, but my experience has been that "complex systems" is now a widely spread but highly disconnected community. In a way, the philosophy that SFI has been promoting for 20+ years is now pervasive in the (much of the) common scientific lexicon, even though the core methologies (which, once upon a time, were things like genetic algorithms, cellular automata and artificial life - not so much anymore...) have not.
So, in that light, maybe there will never be a Complex Systems Letters, because complex systems itself is not a discipline, but is instead a way of approaching certain problems, and a theme to the questions you ask or the form of answers you create.
PNAS has many of the excellent papers in the field, but it also has many excellent papers in other fields and is not easy to get published in if one is not a member of the Academy. There is also room for short papers that may not be good enough to be in a journal with such a wide scope and yet which may not actually be physics papers (i.e., should ideally be sent to places other than PRL).
ReplyDeleteI think fewer of the non-physicists (even the ones who are mathematically sophisticated, and it is certainly true that there are very sophisticated poly sci and social science people out there) are inclined to read the harder-core modeling papers. I know that some of them do (as you point out with your comment on the arxiv), but I suspect that many of the abstract papers would have a great deal of difficulty getting into those journals. Another issue that arises is a very practical one---the physicist trying to get tenure or a tenure-track position is (all else held equal) going to be better off with an article in PRL than one in, say, Social Networks. (Of course, if they can get their work in PNAS, so much the better.) Nature and Science have some good articles, but they are somewhere in between journals and magazines rather than true journals (whereas PNAS is a true scientific journal), and their goal is to sell magazines rather than to disseminate research. (They pretty much admit as much, though with fluffier language.) It is extremely useful to get an article in there, but the ensemble of articles they have aren't necessarily better work than what's in much less prestigious venues. (We actually did send our Congress paper there, and they annoyingly gave it to a biology editor [which made no sense] and rejected it without review on the grounds that it was an application of network theory rather than something that gave new results on the abstract theory. Of course, that statement is true, but completely beside the point because we never claimed that it did that. And I am annoyed that this was in the hands of a biology editor rather than somebody relevant, like what we requested.)
The arxiv serves the purpose that journals used to---that of disseminating work. The purpose of journals now is just to put the final stamp of approval and also force people to improve their papers for their final version. If you haven't heard this story yet, ask Mark to tell you about what happened when he was asked to be on the editorial board for PRE. (I was amused when he told me.)
It's true the complex systems stuff is much more pervasive now than before. Even in the last 10 years, the complex networks research in communities like mine has gone from practically non-existent to a cottage industry.
That's true enough about PNAS, and, I think, succinctly puts the problem the "complex systems community" faces with regard to where to place excellent work that doesn't quite rise to a PNAS level, but is neither appropriate for PRL, Nature, Science, Ecology Letters, etc. On that note, if you ever get pulled into running a journal that fills this kind of gap, let me know -- I'm not quite old enough to be so cynical about peer review and journals that I'm boycotting the whole system, yet.
ReplyDeleteMy impression of Nature and Science has been that they tend to be relatively good in their core areas, which are currently most things biology, having a number of editors, and a strong base of reviewers, with advanced degrees in those. But that they do a relatively poor job (i.e., lots of arbitrary rejections and an inability to find good reviewers, if things even that far) with things that fall outside of their core - your experience with the Congress paper sounds about par for that course, really. But, as you say, having articles appear there helps a lot with certain other bureaucratic things, e.g., tenure, grants, etc. And, it seems to me that if arxiv really does take over from journals the role of disseminating work, then the fight to get work into "good" places will degenerate even more, because the incentives will be wholly aligned behind getting as many of those places on your c.v. as possible in order to get even a little of the shrinking amount of funding available for basic research. I've heard some suggestions about completely abolishing the current academic journal system, but I haven't yet heard a convincing argument about what should be erected in its place. Peer review, I think, does serve some purpose for people who don't write perfect articles the first time around (which is, perhaps, almost everyone outside of Mark, Cris and a select few others). For at least two of my articles, I've gotten some honestly good comments from the anonymous reviewers, and after addressing their concerns, the articles were genuinely better. I worry that by abolishing peer review, we might lose those kind of candid, and occasionally quite constructive, suggestions from people outside your natural circle of collaborators and colleagues. Plus, as you say, the current system of academic promotion puts a lot of stock in the information supposedly captured by where an article gets published (as opposed to what is in the article itself), and if we didn't have journals, hiring and promotion committees might have to actually spend time evaluating the candidates one at a time. Hmm... I can feel much cynicism here, but I'm not sure it's rightly mine. I must've been hanging out with my elders too much.
One of the things I'm most looking forward to about moving to SFI in January is getting more of an insider's look at what SFI thinks about the current state of complex systems research. Since mini-SFIs are popping up all over the country now (three the come to mind immediately are at Michigan, Northwestern and LANL), the term "complex adaptive systems" is even now appearing frequently in business press releases, and they've otherwise won their battle to get their kind of science accepted, I wonder what the Institute thinks of both its success and where the field is going. (I secretly suspect that they are surprised / confused and unsure, respectively, but that seems like a normal reaction under these circumstances, to me.) Having formerly been very interested in things like genetic algorithms and artificial life, I admit that I'm currently quite happy that complex systems research now includes things like complex networks, which seem both more practical, more amenable to rigorous mathematics, and more relevant to basic science. Had the field remained stuck in the "edge of chaos" paradigm, I suspect the field wouldn't be as broad as it is now.
My response from a few minutes ago got eaten... I'm too lazy to retype the whole thing, so let me give a short version:
ReplyDeleteNature and Science are indeed better for biology. Biologists tell me that PNAS stands for "Post Nature and Science" or "Probably not accepted to Science" in their field. (And my Congress paper followed this path, although I ended up deciding that PNAS was a more appropriate venue anyway. Poly Sci people are more likely to see it there, although I'll try to get a forthcoming one in Nature anyway.)
Boycotting peer-reviewed journals is a luxury that requires tenure and other good stuff like that.
I too have gotten genuinely beneficial comments from referees that have improved my papers immensely (in addition to the annoying stuff). I agree this is necessary for that reason. Mark's papers before peer-review have occasionally had annoying flaws (at least for people who are implementing the algorithm therein) that were corrected in the published version. (The implementation sin question were sometimes done before the published version, and a couple times I didn't realize the published version had a correction.) Also, a couple of his papers are excellent scientifically but poorly organized ('random graphs with arbitrary degree distributions' comes to mind). Anyway, mortals like me still benefit when people like Mark are subject to peer review. :)
For job stuff, getting rid of peer review would be very bad. This would make things even more arbitrary than they already are, and forcing people to immediately look at every application closely will result in more nepotism than we already have. (They'll just see someone they know, hire them, and never look at the other applications if the only way to judge anything is look at everyone closely.) While we're nowhere close to perfect, there's a great benefit in getting shortlisted from things like journal publications and then having people look at your application closely.
I too am curious how things are running at SFI now. I've read some popular accounts about the good old days (especially concerning the group that came from Santa Cruz ... there is some discussion of this by Gleick and I believe additional details have been promulgated).
Another new complex systems center is in UC Davis. Oh how I wish my interview with their math department had led to a job...
(I think I wrote at reasonable enough length on everything I included in the first version of this response.)
I'm cautiously optimistic about the UC Davis complex systems center - they have a couple of really good people, which is probably the most important component for success. Once I get settled-in at SFI, and have an idea of what it's like and where it's going, we'll have to chat again. I know they've been doing some soul searching in the past couple of years (at least partially from shuffling their president once or twice), but I don't know what their "post-win" success model is now that complex systems is getting the recognition it deserves. On that note, what's CalTech like?
ReplyDeleteRe: peer review, I try to view it as a civic duty, even though it can be highly annoying at times. (For civic duties, e.g., jury duty and honking at bad drivers, that seems par for the course.) I know a couple of people whom I hold in high regard that are boycotting peer review, and while I understand their choice, I can't help but wonder if it ultimately leads to more bad papers getting published (read: papers with more mistakes) on account of these good people self-selecting themselves out of the pool of reviewers. But yeah, the process itself definitely generates a lot of good, for all the hassle and energy consumption it takes.
This whole journals business really bothers me though. If we were to abolish all journals, what would we put in its place that would fill the (beneficial) roles it currently serves? I think you're probably right that the hiring, etc. process would degenerate quite a bit if people weren't able to claim to rely on "objective" measures of success by looking at the set of journals a candidate publishes in. I guess this debate parallels the one raging in the college admissions community about whether SAT scores actually mean anything, and whether they should be used at all in the admissions process. (Being someone who has scored quite highly on standardized tests, I tend to think they should be abolished posthaste. I feel similarly about course grades, actually. But, I'm not sure what to put in their place that wouldn't be totally onerous.)
My message got eaten again. Today has been another one of those days. I can already tell how things are going.
ReplyDeleteCaltech is wonderful. There isn't much complex systems around here, but there is a ton of other stuff and I've been continuing my work with other people that I already established. When Caltech hired me, they basically said, 'You've done good research. Here is a salary, an office, some academic funds, etc. We just want you to have fun and keep doing good work.' In addition being back on my old stomping grounds (and I have always been a Caltech homer, so that in itself is awesome), this is just a wonderful gig.
Good people not doing peer review is a bad idea. Boycotting submitting to and peer reviewing for a journal (Elsevier journals come to mind) because of its pricing procedures is one thing, but it's not good to do further damage by boycotting the whole process.
The publication record gives relevant information, whereas the SAT (and GRE, etc.) does not. It is designed to test minimum comptetence, but I don't buy that it actually does that. Many math grad schools are now ignoring the GRE and even those that expect it just check if you're above some threshold. It makes no difference how much you are above that threshold, but your application might get culled if you're below it.
I think that grades need to stay for several reasons. They are very flawed, but except for very small colleges (see New College of Florida for an interesting example of an alternative means to do evalulations), the numbers make it impossible. In a larger class, the professor won't be able to know everybody (or anywhere close to everybody). When it comes to graduate school, a lot of very good people who apply never did undergraduate research, so this helps in evaluating those people (and if they come from a larger school, the recommendation letters will often be using their grades in a given class in a fundamental manner). Finally, we need to take ourselves as outliers and remove us from the equation and remember that a majority (perhaps the vast majority) will work more for the chance to get a better grade and wouldn't otherwise do that. (Of course, there are people who also won't work hard for the prospect of grades either.) As you imply, one can't just get rid of it without proposing its replacement. It's a necessary evil.
In terms of both peer review and grades, let me refer to Churchill's comment (it was Churchill, right?) that Democracy is the worst form of government except for all the others.
buy ativan online long does ativan high last - ativan withdrawal long does take
ReplyDeletexanax drug xanax white pill - xanax drug side effects
ReplyDeletexanax online prescription xanax online prescription - xanax pills do they look like
ReplyDeletelorazepam without prescription ativan withdrawal nightmares - ativan xanax klonopin
ReplyDeletebuy xanax bars online xanax generic drug - xanax maoi drug
ReplyDeletegeneric soma the drug soma in brave new world - soma drug schedule
ReplyDeletecheap soma soma medication generic - carisoprodol 250 mg dosage
ReplyDeleteambien generic ambien 2.5 mg - ambien cr generic vs brand
ReplyDeletesoma buy will soma show on a drug test - carisoprodol side effects long term
ReplyDeleteambien online purchase generic ambien cr problems - ambien cr 25
ReplyDeletebuy soma soma medication interactions - carisoprodol oral side effects
ReplyDeletecan you buy valium online no prescription valium 10mg dosage - how to order valium from mexico
ReplyDeletebuy soma generic carisoprodol buy online no prescription - carisoprodol usp 350 mg
ReplyDeletebuy valium online where can i buy valium in hanoi - valium seizure medication
ReplyDeletebuy diazepam generic valium valium 5mg 10mg - klonopin or valium for anxiety
ReplyDeletebuy soma online somanabolic muscle maximizer review yahoo - soma drug description
ReplyDeleteorder zolpidem ambien dosage lethal - ambien cr how fast does it work
ReplyDeletebuy ambien online ambien dosage kids - ambien schedule iii drug
ReplyDeletesoma for sale carisoprodol 350 mg xanax - soma ultram drug interactions
ReplyDeletecan you buy valium online smoking valium pills - valium green round
ReplyDeletebuy valium cheap online 10mg valium equal to xanax - buy valium xanax
ReplyDelete