The SLAC SPIRES yearly list of most frequently cited papers in 2005 is now available. I commented recently on what this was likely to show, quantifying the intellectual collapse of string theory since 1999.
There are exactly three post-1999 particle theory papers among the top 50 in the list. Two of these are about flux compactifications and have moved up significantly since last year reflecting the increasing popularity of landscape studies. At number 18 (up from 29) is the KKLT paper from early 2003, and at number 34 (up from 54) is an earlier paper from 2001 by Giddings, Kachru and Polchinski. The only non-landscape post-1999 paper to crack the top 50 is the 2002 Berenstein, Maldacena and Nastase paper on PP waves (which is part of the AdS/CFT story). It just barely makes it at number 49 (down from 32 last year).
The highest ranked post-2003 paper is the Arkani-Hamed and Dimopoulos 2004 paper on split supersymmetry. It’s at number 106, with a total of 103 citations.
There’s also a new 2005 All-time topcited list. Maldacena’s AdS/CFT paper from 1997 remains very near the top, with 3881 citations. There is nothing post-1999 on this list, which includes the top 186 papers. If recent trends continue indefinitely, it seems entirely possible that no post-1999 particle theory paper will ever make this all-time top-cited list, allowing historians of science to conclusively pinpoint the death of particle theory as having coincided fairly precisely with the end of the 20th century. This is optimistically assuming people lose interest in the landscape. It is also possible that landscape studies will come to dominate the field, with landscape papers then climbing up into the all-time topcited list. This doesn’t really change the conclusion about the death of particle theory.
This doesn’t really change the conclusion about the death of particle theory.
(agree of course)
A bunch of us made predictions about the topcites 2005 list.
Here are the forecast poll results:
I see that your (notevenwrong’s) guess proved correct. Congratulations.
Philosophy. The last refuge of the hypothecator.
No critique is as devastating as a quantitative one ! Kudo’s to Peter for calling it as it is. One wonders, how many hundreds of string papers are still pending review at Nucl.Phys.B, PhysRev D, etc ? How many grad students will join the `String Army’ this year ?
Like the `war’ in Iraq: Massive inertia*snail’s pace = business as usual.
Perhaps one could find another six year period where very few papers made the topcite list (say, 1987-1993 where the top cited papers are pretty far down the list, not to mention few and far between). Looking selectively at the data really doesn’t prove much.
I strongly disagree that I’m looking selectively at the data. I think that if you ask virtually any particle theorist who has been in this business for many years what they think about how many new ideas have come along over the last six years compared to other periods in their lifetime, they will tell you that the number of such ideas in the past six years has been unusually low. I’m putting forward the most honest and objective way I know of quantifying this. If you have a better one, let’s hear it.
Maybe there have been good ideas, but if you’re not a big name….nobody pays attention.
The big names decide what is good physics and what is not.
Maybe we are close to the end of physics.
In science in principle you should not worry what the big names think, but what the experimental measurements say.
I feel that one must take other things into account when making claims like “the death of particle physics is nigh”–the set of string theorists who understand, say, twistor strings is a very small subset of all string theorists, so naturally there will be fewer citations of that work. Papers like RS1 or AdS/CFT represent huge steps in understanding (or misunderstanding, if you like) of physics that don’t just happen everyday. Some of the papers listed on this list (just the theory papers apply) represent really revolutionary ideas (inflation, Beckenstein/Hawking radiation, asymptotic freedom…).
One should look at the theoretical papers (not things like WMAP data) which represent ground-breaking ideas (judged by the number of citations), and look at the years in which they were published. If you can show that the current six year lapse in theory is longer than any other before, then I will accept your claim that particle physics is in a dark ages.
I recall the years from the mid 70’s to 1983 (so that’s more than 6 years) when basically nothing was going on in theoretical HEP (except for Witten’s papers on Chern Simons theory, non abelian bosonization, global anomalies, etc, which were too mathematical to be called high energy physics proper). The standard model was fully established and people started playing mundane games with Kaluza Klein, SUSY, etc.
The situation changed completely in 1984–1985 which were revolutionary years: superstring theory, conformal field theory, quantum groups, knot invariants, and RSOS models in stat mech, and probably more. Probably the most revolutionary years since quantum mechanics and/or quantum electrodynamics, in terms of the number and quality of new ideas floating around.
So yes, I’m hopeful that something will come along.
As I see, these people at SPIRES are he(l)pful. And jointly with arxiv and KEK, it is a resource to be proud of.
Because of some of the comments here, I took a closer look at some of the topcites data, even making a histogram of the time distribution of the all-time topcited papers.
The results for 1975-86 agree with my memory of the period. There were lots of new ideas that people were working on: supergravity, lattice gauge theory, instantons, GUTs, 2d models, etc. (Chern-Simons-Witten was later), although some level of frustration at how difficult it was to either solve QCD or get beyond the standard model. The topcites data show a consistent average of 5-6 papers/year for this period, with a peak of 9 in 1985 corresponding to the advent of strings.
The period from 87-93 does contain many fewer papers, just seven, an average of one/year. This is significantly more than the post 99 zero/year, but more to the point, research during this period was largely driven by new developments in a way that is very different than recent history. If you look at the earliest SPIRES yearly top-cites compilation, covering 92-94, and ask how many of the top-cited papers were from the six years just before this (1987-93), you get 13 out of 40, even though most of these papers are on things like CFT results that were very popular then, but haven’t made the all-time list since people have lost interest in them. This number of 13 out of 40 is very different than the similar number for 2005: 3 out of 50 (or 2 out of the top 40). While people in 1993 felt things were in a lull, the situation was very different back then, there were plenty of quite recent results that large numbers of people were working on.
Also, while 1987-93 was a fallow period for string theory, it was the period when the whole field of topological quantum field theory got going, and Witten did his Fields medal winning work on Chern-Simons. There’s nothing remotely similar to this in the last few years.
Citation counts don’t reliably identify important new ideas, they reflect what the community is working on. Famously, Weinberg’s electroweak paper received no citations for several years, but now is the most cited theory paper of all time. Maybe there is some important idea out there in the recent literature, but its significance is not understood, so hardly anyone else is working on it. Witten’s twistor string paper has received relatively few citations, but that seems to be not because people don’t understand it, but because it has not led to as much as originally hoped, so few people continue to work on it.
Thank you for your detailed analysis. Absent any other indicators, your analysis does seem to indicate that ideas are fewer and further between. It does seem to quantify the dark ages of particle physics you alluded to in the original post. Perhaps we have been “sidetracked”, so to speak, on ideas like RS1 and AdS/CFT, or that the field is more dispersed than at other points in the past.
It takes many years before a paper can gather 1000+ citations (this is the magic number that makes it to the top-cited papers of all time, 990 is no big step down from this). This never happens immediately. Claiming that because no such thing has happened in the last 6 years that theoretical particle physics is dead is a wrong conclusion and does not take into account historical data on the subject.
To make such a statement you have to wait (at the very least) about 15 years and then look back. You also have to wait and see what happens after the LHC is taking data to declare that theoretical particle physics is dead, because if anything unexpected is found, I can guarantee that it will be a very interesting time for theoretical physics.
To judge a field by how many papers from the previous year have gathered 100+ citations is probably more correlated with having fashionable theories/mechanisms. On the other hand, if a subfield is more technical and it requires a lot of hard work to get results, then you would be extremely happy if a paper got 50+ citations in that year. One should always judge by quality, not just quantity.
The fact that experimental physics is dominating the list is actually very healthy: there is very interesting new data (neutrino masses, cosmological precision experiments) and there is a strong effort to fit these observations with our understanding of the standard model and physics beyond the standard model (like inflation).
David B.: I strongly agree with your statements.
David B. and A. Torok,
Obviously I was making a provocative statement about the “death of particle theory”. We don’t know yet how it will turn out, maybe the LHC will change everything. However it’s not going to be telling us much for at least another three years or so, at which point the period of zero progress that I am identifying will have stretched out to about a decade.
As for your comments about the all-time list, you’re just refusing to look at the data. If someone comes up with a new idea that is recognized as important in this field, it tends to get a lot of attention quickly. There are a lot of smart people out there everyday looking at new papers on hep-th, ready to jump on anything new they see. Historically, it is not at all unusual for important papers to accumulate nearly 1000 citations within 2-3 years (the most heavily cited papers generally get 3-500 citations/year). I picked one recent year at random (2002), and looked at the SLAC all-time list which that year included the top 104 papers. Eight of these were from within the previous 5 years. It doesn’t take 15 years for most important papers to show up on the all-time list. Please look at this data, and don’t just assume you know what it says. And no, the reason for this data is not that the list is being overwhelmed by a huge number of important new experimental results. Besides WMAP, which dominates the list, this has not been a period of unusually many or important HEP experimental discoveries, although HEP experiment is in better shaper than HEP theory.
As far as I can tell you’re both students and lack historical perspective on this issue. If you ask anyone who has been working in this field for twenty years or more, paying close attention to new developments as they arose, I think you’ll find that virtually all will tell you that the dearth of recognizable new ideas during the last six years compared to earlier periods has been dramatic. If you don’t want to believe me or other people who have this experience, look for yourself at the SLAC data.
I think it is extremely unhelpful for people to try and deny the facts of the matter here. Particle theory is in a bad way, and the community needs to figure out why this is true and whether something can be done about it. Denying that this is true won’t make the problem go away.
I am not a student and I have been in the field for more than 10 years, so I have some historical perspective. Furthermore, I am not in “denial”. I was just pointing out that it is too early to make concluding statements like that. You have an undeniable right to have a negative opinion of the current state of theoretical particle physics and nobody is trying to rob you of that fact.
Looking at the papers on the topcited list that have between 1000+ and 1200 citations you will quickly realize that the grand majority of them have been around for more than ten years, so my estimate is broadly correct.
To make a point: the paper of Faddeev and Popov that introduced
ghosts and gauge fixing in 1967 took until after the year 2000 to get to the top. Are you telling me that their work is unimportant because it didn’t make it to the top in three years (as it obviously should have)?
It is not true that you generically get to that number within three years. That is truly exceptional, even today.
You also failed to mention that many of the topcited papers that accumulated 500+ citations per year for three years in a row around 1997 were very closely related to each other, and not completely independent ideas either. The year 2002 is not random.
Why don’t you make the experiment with the year 1995? Not a single theory paper from the previous five years is in the topcited list.
The theoretical high energy community is waiting for new data. The extreme precision of current data places enormous constraints on new ideas. The big breakthroughs you are probably referring to (circa 1997) are related to new paradigms to avoid these constraints based on ideas of extra dimensions, etc.
New ideas that people are waiting to “pounce on” require two ingredients: avoiding current phenomenology constraints in a clever way and most importantly: a simple calculation attached to it. Without the second fact you can not hope that there will be 500+ citations within a year, no matter how important your idea is.
My apologies for mistaking you for a student, I was going by the e-mail address you gave and very quick results of a google search.
As for the points you make:
1. Sure, at all times most top-cited papers are older ones, but that’s irrelevant. What I’m trying to get at is the number of recent ideas recognized by the field as significant. Often these will appear on this list within a few years of when they come out.
2. As I noted earlier, historically there are examples of papers whose importance was not recognized for quite a few years. Weinberg’s “A theory of leptons paper” was a dramatic example of this. The Faddev-Popov paper of 1967 was another. When it came out, very few people cared about non-abelian gauge theories (they were too busy working on the predecessor of string theory…). It was only after ‘t Hooft showed renormalizability in 1971 and asymptotic freedom was discovered in 1973 that gauge theories became popular. If you’re going to compare citation counts of papers from the sixties and ones from nearly forty years later , you need to take into account two factors. The number of papers being written was many fewer forty years ago, and, unless you are looking at a different data source than me, the SLAC all-time lists from years ago are much smaller and nowhere near as deep as the current ones. The latest all-time list (2005) that has zero papers from the last six years on it is 186 deep. The lists from 1995 and 2000 that you refer to have only 47 and 87 entries respectively.
3. The year 2002 was randomly chosen as a recent year that would pick up the pre-2000 period before the point at which I claim brain-death set in. Sure, among the 8 papers on that list, there are correlations. I wasn’t making much of that absolute number, just that it is very different than zero. I’ll also point out that I wasn’t comparing numbers, just responding to your incorrect argument that recent papers don’t appear on the all-time list.
4. Let’s discuss 1995. As others have remarked, 1987-93 was considered by most people the previous low point of new ideas in particle theory, so 1995 should be the worst previous year in terms of numbers of new ideas showing up on these lists. The 1995 all-time list doesn’t include any recent papers, but it is only 47 long. If you can get ahold of a list of the top 186 papers, which is what you need to directly compare to 2005, I’m willing to bet it will have a significant number of recent papers in it. The 1995 list of top papers recently cited (in this case 93-95, there is no data for just 1995) includes in the top 40 eight from the period 1990-95. Compare this to 2005, where the top 40 includes just the two recent landscape ones. Also note that 1995 was the year of the second superstring revolution, and the year after Seiberg-Witten. Anyone looking at these numbers at the end of 1995 would also be well aware that there were quite a few papers from late 1994 and 1995 that were accumulating citations at a very fast rate. The 1995 situation was just radically different than that of today. If you know of any papers too recent to make the 2005 top 40, but that are accumulating citations very quickly like Seiberg-Witten, let me know.
5. The new ideas that dominate the late 1990s list are not just extra dimensions. Many of them have to do with AdS/CFT and other aspects of string theory, which have nothing to do with the precision of current data.
6. Making the top 40, or the top 186 all-time list doesn’t require 500+ citations in a year. To make the top 40 just requires about 150 citations/year, to make the top 186 all-time list over five years would require 200 citations/year. Historically, in all previous periods, there have been many papers getting cited at these rates (although as you go far back in time, you have to rescale for the significantly smaller number of papers being produced). My point is that the only recent papers from the last 6 years in this league are landscape papers: the KKLT paper (223 citations last year), and the Giddings et al paper (169 citations).
7. People don’t just pounce on phenomenology papers, they also pounce on string theory papers. I don’t think of KKLT as a phenomenology paper. If someone had a decent new idea about string theory, even if it had zero phenomenological implications, it would get pounced on.
Finally, what is the point of trying to make these tendentious arguments that you are using? Do you really believe that the last six years has been a healthy period for particle theory? If you believe that, let’s hear some examples of important ideas from the last six years, of a number even remotely comparable to previous six year periods. If you agree that this has not been a healthy period, why try to come up with quantitative arguments for the idea that it’s not so bad?
The last few years have been a healthy period in high energy theory. They are definitely not as intense as during the 1990’s, and I don’t feel like there is nothing to do or work on, quite the contrary. My “counting” is just as tendencious as yours. I picked 1995 exactly because that was another period where particle physics had been declared dead a few years back (I had been told that there was no future in high energy particle physics before entering graduate school).
There have been very interesting recent results (they might not make it to the topcite 1000+), but here are a few of extremely high quality results of the previous five years (apart from the ones you have already mentioned) that indicate that theoretical high energy physics is doing just fine:
Minahan and Zarembo hep-th/0212208 provided new tools to address the AdS/CFT. They showed that one loop computations in the AdS/CFT gave rise to integrable spin chain models. The presence of an integrable structure in the calculation of anomalous dimensions in N=4 SYM was unexpected.
The follow-up work is very technical, but it is extremely exciting and of very high quality.
Dijkagraaf and Vafa in a series of papers in 2002 made a conjecture relating matrix models and non-perturbative effects in supersymmetric gauge theories. Although they were inspired by string theory, these results hold regardless of this motivation and permit one to calculate many results exaclty, as was proved later. You might complain that they relate to supersymmetric theories, but then, it is very hard to come by with exact calculations in any context.
Arkani-Hamed, Cohen and Georgi hep-th/0104005 introduce the ideas of deconstruction and show how we might be fooling ourselves into thinking we live in higher dimensions if the high energy theory has the right set of discrete symmetries.
Witten gave a huge push to the computation of helicity amplitudes in QCD in hep-th/0312171. Many of the techiques devised since then on these ideas are extremely efficient at computing gluon scattering. This is the dominant source for uncertainties in the LHC.
I can keep on going. Most of the interesting works of these previous five years involve computations that require a lot of work. I don’t see any reason to believe that because calculations are harder that progress has stopped. It just moves slower and that is reflected in the number of citations per year that a paper receives.
The above are a few of my favorite papers of the past five years, and you might appreciate that none of them are directly related to the landscape. Even though I try to keep up to date in high energy physics, there is not enough time in the day to read everything that comes out, so I can not give you an in-detail presentation of what is going on everywhere, but for me, personally, it feels like a healthy environment to be working on right now.
I think you’ve identified the best work in particle theory during the past few years, but I also think that it is very clear that these are results that are of much less significance than ones in any similar list that people would have generated in the past when asked to produce a list of the most important work during the past 5-6 years.
To have a serious discussion of the significance of those papers, one would have to go into a long discussion of what exactly is in them, what one could in principle hope they might lead to, what exactly they have led to, etc. They are all papers that have received a great deal of attention, and are certainly not things that have been ignored or not understood. The relatively modest number of times they were cited by other papers during 2005 reflects the fact that other theorists find these results of limited interest and have not been able to get very much out of them. The only one that has more than 100 citations is Minahan-Zarembo, because many people are trying to find a useful integrable structure in N=4 SYM, partly because of the relevance to AdS/CFT. It will be interesting if that goes anywhere, but progress seems limited, and this is still a very long way from leading to something that you can do any physics with.
The Dijkgraaf-Vafa results give very limited information about certain supersymmetric models, again, a long way from being useful for physics. My impression is that many people have looked at this, but few people are still working in this area.
The twistor string again is an idea that seems not to have led to as much as people had originally hoped. Again, many people have looked at this (probably most serious particle theorists), but very few people are still working on it.
Those are general impressions from listening to talks, scanning hep-th papers, reading conference talks off the web, etc., but they are borne out by the quite small number of citations those papers have been getting compared to the leading papers from previous periods. Take a look at the total number of citations of these papers so far:
Witten, twistor string: 206 citations in 2.25 years, 92 citations/year
Dijkgraaf-Vafa: 322 citations in 3.6 years, 89 citations/year
Minahan-Zarembo: 283 citations in 3.25 years, 87 citations/year
Arkani-Hamed, et. al: 300 citations in 5 years, 60 citations/year
I think it is extremely unlikely that any of these papers will ever reach 1000 citations or the all-time topcited list. The only one of these where there still seems to be sizable interest is Minahan-Zarembo, and whether people keep citing it depends on whether the search for integrability in N=4 SYM goes anywhere. Maybe interest in this topic will grow significantly, or continue at the present rate for the next seven years necessary to get to 1000 citations, but I find that hard to believ.
All in all, I’m sorry, but I think the evidence that you’ve given that this is a healthy period just doesn’t stand up to scrutiny. It’s always true that there is “something to work on”; one can always calculate something, the question is how interesting the result of the calculation is. The SPIRES data I’ve quoted conclusively show that the theory community sees these results as much less interesting than the results achieved in any other similar period in our lifetimes.