Travis Brooks of SLAC’s SPIRES database has a blog posting today announcing the availability of various lists of the high energy physics papers most heavily cited during 2009. A full matrix of links to this data is here, data broken out by arXiv subfield is here.
It’s hard to over-emphasize how much the particle theory parts of these lists are dominated by classic papers on AdS/CFT, in particular Maldacena’s original 1997 paper. It now has over 6600 citations and during the next year or so should pass Weinberg’s 1967 paper as the most heavily cited particle physics paper of all time. One remarkable thing about this paper is that in recent years the number of citations of it has increased to new highs, reaching 731/year in 2008. Even at the height of theoretical activity surrounding the Standard Model back during the late 1970s, none of the classic papers of that subject (such as Weinberg’s) reached even half the citation rate of the Maldacena paper. Similarly, during the explosion of interest in string theory after 1984, none of the papers from the first superstring revolution reached half the Maldacena rate.
Among the top 25 entries in the 2009 overall top-cite list, the leading theory papers are 97-98 AdS/CFT classics at positions 3, 8 and 9, as well as Randall-Sundrum extra dimension papers from 1999 at 14 and 20. Among the top 50 entries, there are only two hep-th papers that are not from the last millennium: at number 33 one of the papers on superconformal Chern-Simons/supergravity duality, and Horava’s Lorentz-breaking gravity proposal at number 38 (there’s a very recent article about this at FQXI).
Looking just at the articles cited in hep-th during 2009, gauge-gravity duality is again completely dominant. The top 3 are AdS(5)/CFT(4) classics , the rest of the top 9 are about the lower dimensional AdS(4)/CFT(3) case (except for an AdS/CFT review article). To find something not about gauge-gravity duality, one has to go down to number 10, the KKLT paper that set off the landscape craze.
Taking a look at recent hep-th lists of postings, there seems to be no let-up in the AdS/CFT dominance. The only recent paper on another topic that seems likely to make the top ten of the 2010 listings is Erik Verlinde’s January paper on entropic gravity, which two months later already has 40 citations.
Very many of the top 50 in 2009 are astrophysics papers. (WMAP figures promimently ~ good for WMAP!) I suppose that’s an indicator where the new physics is. As for Maldacena v. Weinberg, I suppose this is an indication of publication inflation? (Substitute favorite deragatory term if so desired…) There was no arXiv in the good old days, quite likely many of the citations are to arXiv postings, also there are many more conferences/symposia/whatever these days (and easily available online), so there are overall many more publications these days. Not a uniquely string theory phenomenon.
I am surprised that Witten only has 2 papers in the top 50. I thought that his papers were very widely cited.
Roger,
Witten has written many highly influential papers, but there is no single one that stands out, and they tend to be on the mathematical end of the subject, where fewer papers are published. In these days of all AdS/CFT, all the time, very little that is not AdS/CFT is getting heavily cited in hep-th. It’s not surprising that the Witten papers that show up are his AdS/CFT-related one and one other.
It is not hard to understand why Maldacena’s paper is so widely cited. It is a striking and tantalizing conjecture that attracts string theorists and particle theorists alike.
It is not hard to understand why Weinberg’s paper is so widely cited. It is a striking and tantalizing THEORY OF NATURE that attracts string theorists and particle theorists alike.
It’s a bit silly to compare citations to Weinberg versus citations to Maldacena. In any field there is a body of accepted knowledge that one doesn’t bother to cite. No one cites Weinberg for the Standard Model any more than they would cite Feynman, Schwinger, Tomonaga or Dyson to justify doing quantum field theory, or Einstein to explain relativity. It’s fully integrated into the infrastructure of the field. This is far more important than any number of citations.
(I imagine that at some point people will stop explicitly citing the original AdS/CFT papers and just assume that it’s known, but — maybe just because they’re already there in the TeX code everyone recycles, maybe because it isn’t in a standard textbook yet — so far it looks like everyone will keep doing it.)
anon,
I was comparing the citation rates for Maldacena and Weinberg during comparable periods, about a decade after they were written, at a time when research in HEP was dominated by the topic that they started. Weinberg’s 1967 paper (after getting no citations in the first few years, but that’s another story…) reached a high of 351 citations/year 13 years later, in 1980. Maldacena’s 1997 paper may not have hit its highest point yet, but was at 731 citations/year 11 years later, in 2008. This increase of more than a factor of two is partly explained by an increase from 1980 to 2008 in the number of papers written in HEP, but I think also reflects the historically very unusual domination of hep-th by research into one particular set of ideas.
The other comparison one can make is just the integrated number, where Weinberg is about to lose out, despite a 30 year head start. The future is hard to predict, but it seems just about certain that the Maldacena paper will be the most heavily cited HEP paper written during the 20th century, and may very well end up being the one most heavily cited during the 21st.
re Maldacena – Is there any experimental evidence or empirical evidence that would suggest this level of citation is bad for HEP?
dan,
In and of itself, I think the high citation rate is historically unusual, but from that you can’t conclude it’s “good” or “bad”. Personally, I think the way the field is dominated by one very specific research topic is a reflection of an unhealthy situation. Many would argue though I think that this is the way it is just because AdS/CFT is the only good idea out there that has not been fully investigated.
it would be interesting to renormalize to the total number of citations in the field per year
That requires an index of inflation, to be able to state a baseline “the comparisons (of Maldacena and Weinberg?) are in constant 1990 citations”. What of the citations of the papers of Albert Einstein (the statistics must be available somewhere) … how to compare? Pre-WW2 physics was “small science” and a comparison of citations based on inflation-adjusted numbers would still be impossible to formulate meaningfully. I do not doubt that in the decades to come things will get much worse.
M and N,
It would be interesting to normalize these statistics, if anyone has good data on the yearly size of the HEP literature I’d be interested to see it. The comparison between the peak Weinberg and current Maldacena numbers is from 1980 vs 2008. There was a huge historical expansion in the size of physics research employment during the 60s, at least in the US. I’d suspect that there has been an increase from 1980 to 2008, but not as large, and concentrated outside the US and Western Europe.
More relevant may be that the Weinberg and Maldacena papers are not comparable because the Weinberg one was of much wider relevance, getting cited by phenomenologists and experimentalists as well as whatever the 1980 equivalent was of hep-th.
I think it would be interesting also to normalise by the number of citations per paper. When I read older papers I never see the long{recent work on the Buggins model includes [1-55]} reference lists that often occur in the current literature. Maybe that is because with the arxiv people are more aware of the literature, with electronic texts it is easier to cut and paste references, and there are also more people to be offended if you don’t cite them.
There is also an unhealthy herd effect on citations when it comes to fashionable topics that seems particularly prevalent in hep-ph but is also present in hep-th.
As mentioned by others the comparison of absolute citation numbers between Maldacena and Weinberg doesn’t mean very much, any more really than the relative price of a Hershey bar in 1998 compared to 1967.
In the period I have been in the field, which is not so long, I’ve noticed that the size of the topcite percentage bins in the SPIRES playground has grown by around 50%. Acording to SPIRES close to 1 paper in 5 is a topcite now. Glad to see we’re all doing high impact work then 🙂
hi Piscator. Dividing by
N(t) = (average number of citations per paper) = (average number of references per paper)
allows to remove the effect of longer citation lists, which is mostly a consequence of informatics (easier to ask for citations, easier to say yes, easier to access papers).
But also the number of papers per author increased, in part because computers allow to write more useless papers, in part because the field expanded.
To remove both effects of informatics one should therefore divide by the average number of references per author per year.
Citation count (at this level) is usually not a good indicator of what a paper means to our understanding of nature. There are plenty of stringy papers that are far less cited, and ultimately more important.
Instead citation count here is a sort of combination of luck factors and sociological ones. For instance, if your paper is the final word on the subject, and is part of the way nature work, you will get a lot of citations, but not nearly as much as if you only discover a portion and leave questions unanswered.
Likewise, sometimes it turns out that a research direction is too technically challenging, and you hit a lot of brick walls. Consequently, less people will be inclined to take the plunge.
AdS/CFT hits that perfect middle ground, where its not too hard to work in and spawns fruitful research, yet is by no means perfectly well understood.
Perhaps unheralded is the tendency to truly bury contributions in experimental particle physics. Talented young people first get their contributions buried under the collaboration, where first-authorship for the prime movers has been largely abandoned. Second, the papers get rolled up into the Review of Particle Physics.
So for me, the true source of who does what in experiment is the record of doctoral dissertations and internal collaboration notes, where the latter is generally not public. Perhaps those notes should become public, say, 1 year after they are written. The public pays for them.
There has been some discussion by theorists that it is hard to get young LHC researchers interested in new ideas. There are at least two ways to view this… on the one hand, the system discussed above gives little recognition to the people who do the work. On the another hand, the theory community has lost touch with real experiment in their 30 year or so journey through string theory, and suffers cognitive dissonance now when they look at how experimental collaborations make their sausages.