Solving the dk Lag: Spires Keywords vs. Desy-Keywords

[arivero]

As some people here is fond of spires stats, let me note that

As of 7/26/04, the keyword index has changed, and more changes will follow. Currently k or keyword searches the desy-keywords described below, as well as searching the title of the paper and any keywords the author may have written on the paper (only available for papers in Aug 2004 or later). This means that while there is a time lag for the desy-keywords to be applied, due to the hard work taking place at DESY, more recent papers will still be found by the keyword search.
This index, keyword, is now the preferred method to find papers by "subject". If you are searching title=proton because you want to find papers about protons, try searching keyword=proton instead, you'll get more results. The only reason to use title is if you know the title of the one paper you are looking for.

A problem with the new method is that you get some noise, for instance d branes could get some matches on articles about the D particle.

The lag on DESY Keywords can be, it seems to me, as far as 6-10 months.

here you have some fine dk keywords and suggestions (non-keywords) from
http://www-library.desy.de/schlagw2.html
to be used with "find dk ..."

D-brane
string model
string
-superstring ('string model' and 'supersymmetry')
-dual model ('dual resonance model' or 'duality')
M-theory
*anti-de Sitter (e.g. 'space-time, anti-de Sitter')

it seems to me than an appropiate search for string theory across time could be

dk strin# OR dk D-brane OR dk dual resonance model OR dk m-theory OR (dk space-time, anti-de Sitter AND dk super#)


.

 

[arivero]

dk strin# OR dk D-brane OR dk dual resonance model OR dk m-theory OR (dk space-time, anti-de Sitter AND dk super#)

2006 1215
2005 1327
2004 1356
2003 1331
2002 1561
2001 1474
2000 1593
1999 1553
1998 1505
1997 1411
1996 1108
1995 1082
1994 1014
1993 867
1992 968
1991 961
1990 1183
1989 1280
1988 1384
1987 1583
1986 1402
1985 663
1984 290
1983 192
1982 253
1981 192
1980 166
1979 107
1978 106
1977 127
1976 128
1975 86
1974 59
1973 23
1972 16
1971 3
1970 3
1969 2

 

[arivero]

Same thing without the maldacenian anti de sitter keyword.

2006 1157
2005 1282
2004 1321
2003 1298
2002 1501
2001 1429
2000 1527
1999 1456
1998 1444
1997 1405
1996 1107
1995 1081
1994 1011
1993 866
1992 967
1991 960
1990 1182
1989 1280
1988 1384
1987 1583
1986 1402
1985 663
1984 290
1983 192
1982 253
1981 192
1980 166
1979 107
1978 106
1977 127
1976 128
1975 86
1974 59
1973 23
1972 16
1971 3
1970 3
1969 2

 

[arivero]

Same thing without the maldacenian anti de sitter keyword, plus relevant http://www.slac.stanford.edu/spires/find/hep/www?rawcmd=FIND+DK+STRIN%23+OR+DK+D-BRANE+OR+DK+DUAL+RESONANCE+MODEL+OR+DK+M-THEORY+AND+TOPCITE+1000%2B&FORMAT=www&SEQUENCE=ds%28d%29" in the year. Of course, some 500+ in the 2003-2000 period and some 100+ in the recent period could be added to the corpus.

2006 1157
2005 1282
2004 1321
2003 1298  1
2002 1501  1  
2001 1429
2000 1527
1999 1456  2
1998 1444  3
1997 1405  1
1996 1107  2
1995 1081  4
1994 1011  1
1993 866
1992 967
1991 960
1990 1182  1
1989 1280  
1988 1384  1
1987 1583
1986 1402  1
1985 663    7
1984 290    2
1983 192    1
1982 253    1
1981 192    2
1980 166
1979 107
1978 106
1977 127
1976 128    1
1975 86     1
1974 59     3
1973 23     2
1972 16
1971  3
1970  3
1969  2

 

[BenTheMan]

I think marcus (or martin, whatever his name is on this forum) is the only one here who really cares to turn physics into a popularity contest.

 

[arivero]

I think marcus (or martin, whatever his name is on this forum) is the only one here who really cares to turn physics into a popularity contest.

Actually, SPIRES stats are useful. For instance, above you can locate the so-called "string revolutions" straight from data, instead of relying in word-of-mouth descriptions. And by selecting the desy keywords, you also get some perspective: in the inter-revolution periods, the stream of string publications was actually smaller than nowadays.

As for who is turning physics into a popularity contest, my oppinion is that management of resources (in mid-range universities, mostly?) is to be blamed, more than Marcus/Martin. I think the buzzword is "indicators"

 

[arivero]

let me to put names to the 1000+

[FONT="Courier New"]2006 1157
2005 1282
2004 1321
2003 1298  1   De Sitter vacua in string theory.
2002 1501  1   Strings in flat space and pp waves from N=4 superYang-Mills.  
2001 1429
2000 1527   
1999 1456  2   String theory and noncommutative geometry
               Large N field theories, string theory and gravity.
1998 1444  3   Phenomenology, astrophysics and cosmology of theories with submillimeter dimensions and TeV scale quantum gravity.
               New dimensions at a millimeter to a Fermi and superstrings at a TeV.
               Gauge theory correlators from noncritical string theory.
1997 1405  1  The Large N limit of superconformal field theories and supergravity. 
1996 1107  2  M theory as a matrix model: A Conjecture.
               Microscopic origin of the Bekenstein-Hawking entropy.
1995 1081  4   Dirichlet Branes and Ramond-Ramond charges.
               Bound states of strings and p-branes.
               Heterotic and type I string dynamics from eleven-dimensions.
               String theory dynamics in various dimensions.
               ===> this year, 1995, the top quark is found.
1994 1011  1   Unity of superstring dualities.
1993 866
1992 967
               ===> hep-th launched 91-08 (it saturated above 3200 papers, at the start of the XXIth century
1991 960
1990 1182  1   A Possible new dimension at a few TeV.
1989 1280  
1988 1384  1   The Cosmological Constant Problem.
1987 1583
1986 1402  1   Strings on Orbifolds. 2.
1985 663    7  Conformal Invariance, Supersymmetry and String Theory.
               The Lund Monte Carlo for Jet Fragmentation and e+ e- Physics: Jetset Version 6.2.
               Noncommutative Geometry and String Field Theory.
               Strings on Orbifolds
               Strings in Background Fields.
               Vacuum Configurations for Superstrings.
               Heterotic String Theory. 1. The Free Heterotic String.
1984 290    2  The Heterotic String.
               Anomaly Cancellation in Supersymmetric D=10 Gauge Theory and Superstring Theory.
1983 192    1  Parton Fragmentation and String Dynamics.
               ===> the long search for the W bosons finishes, meeting predictions, in 1983.
1982 253    1  Superstring Theory.
1981 192    2  Quantum Geometry of Bosonic Strings. 
                Topology of the Gauge Condition and New Confinement Phases in Nonabelian Gauge Theories.
1980 166
1979 107
1978 106
1977 127
1976 128    1   Topology of Cosmic Domains and Strings.
1975 86     1    Stability of Classical Solutions.
1974 59     3     Hamiltonian Formulation of Wilson's Lattice Gauge Theories.
                A Two-Dimensional Model for Mesons.
                Confinement of Quarks.
1973 23     2   A Planar Diagram Theory for Strong Interactions.
                Vortex Line Models for Dual Strings.
1972 16       
1971  3
1970  3
1969  2
[/FONT]

Some recent 500 and 100 topcites should be accounted, perhaps. Here Marcus detailed stats can be used. Let me note the Rolling tachion in 2002, and Hierarchies from fluxes in string compactifications in 2001.

There is also external influence: discovery of W, Z, top, oscillations, cosmological constant... And the launching of hep-th.

It could be useful to do a bidimensional table, usin the "institutions" index, and then showing how many papers come from the main research institutions.

 

[arivero]

It could be useful to do a bidimensional table, using the "institutions" index, and then showing how many papers come from the main research institutions.

Actually the 1000+ come from only a few institutions.

Princeton+U. *10
Princeton,+Inst.+Advanced+Study *6
Stanford+U.,+Phys.+Dept. *5
Harvard+U. *5
Caltech *3
CERN *3
Rutgers+U.,+Piscataway *3
Texas+U. *3
Chicago+U.,+EFI *2
Cornell+U.,+LNS *2
Ecole+Polytechnique *2
ICTP,+Trieste *2
Landau+Inst. *2
Lund+U.,+Dept.+Theor.+Phys. *2
Queen+Mary,+U.+of+London *2
SLAC *2
UC,+Santa+Barbara *2
Bohr+Inst.
Cambridge+U.
Imperial+Coll.,+London
Santa+Barbara,+KITP
Tata+Inst.
UC,+Berkeley
Yeshiva+U.

 

[arivero]

The basic pattern for all the top institution is the same: a huge surge of publications in 1985, some oscillation and a current rate, in 2006, between 2/3 and 1/2 of the peak rate. A small issue happens with Harvard, which in 2002 splits its affiliation between Math and Phys departments, it seems. The total sum, harvard#, keeps strong.

 

[BenTheMan]

I think it's a bit odd that marcus hasn't commented in this thread to defend his blind posting of SPIRES numbers every year.

 

[arivero]

I think it's a bit odd that marcus hasn't commented in this thread to defend his blind posting of SPIRES numbers every year.

Well, I hope he will make use of these techniques in the future.

Another one:

FIND EPRINT 0704 AND ARX HEP-TH

it substitutes the old format hep-th/xxxxxx

I find it becomes interesting when used with topcite 50+ or with FORMAT=wwwcitesummary

 

[BenTheMan]

Well, I hope he will make use of these techniques in the future.

He tends to ignore evidence that doesn't support his conclusions, so I doubt it.

 

[marcus]

Well, I hope he will make use of these techniques in the future.

Another one:

FIND EPRINT 0704 AND ARX HEP-TH

it substitutes the old format hep-th/xxxxxx

I find it becomes interesting when used with topcite 50+ or with FORMAT=wwwcitesummary

I didn't know about the "find eprint 0704" command, or the "find arx hep-th" command.
Thanks for the new techniques and all this stuff, Arivero!

What happens when you use this with "and topcite 50+"? You say it becomes interesting, so I am curious to know.

 

[marcus]

BTW as far as I can see the numbers here tend to reinforce the stats that I have dug up in past years.

these have mainly been of two kinds:
A. plain string research output (based on Harvard database, which automatically screens by words occurring in the abstract----no "k versus dk" problem and no delay).

B. hand counts of string papers occurring in the Spires "top fifty" list. Again in that case I don't need any DESY keyword search since I just go thru and identify the string papers by hand.

In effect, my methods have been more primitive, but the results AGREE with the what Arivero has gotten using the Spires keyword search. Arivero's methods are smarter, but have the problem that there is a DELAY waiting for the DESY librarians to tag the papers.

So now we can have another PREDICTION POLL this time for 2007!

Because of the 6 month delay that Alejandro mentioned we won't really know the number of string papers (measured his way) published in 2007 until June 2008. Here's the list including the antideSitter hits.

2006 1215
2005 1327
2004 1356
2003 1331
2002 1561

So we can turn this around and try to forecast the 2007 number:

2002 1561
2003 1331
2004 1356
2005 1327
2006 1215
2007 ?

 

[arivero]

What happens when you use this with "and topcite 50+"? You say it becomes interesting, so I am curious to know.

I was not thinking from the point of view of statistics but as a kind of catch-up browsing (as you see, now I am out of game for long periods, due to my job requeriments). You can abstract very fast what is happening in the town if you browse month by month but cite ordered.

 

[arivero]

In effect, my methods have been more primitive, but the results AGREE with the what Arivero has gotten using the Spires keyword search.

Well I supposse that Ben refers to the long time perspective: the current decrease of string papers is still smaller, and smoother, than previous inter-revolution decreases.

(Of course, given my current goal of causing a new revolution in the field I feel biased towards this interpretation)

 

[marcus]

... a current rate, in 2006, between 2/3 and 1/2 of the peak rate. ...

I think this is about right, maybe overstates the drop in output a little. I don't ordinarily INTERPRET, what my past posts have mainly done is point out that there was a peak around 2002 and a decline since then. It looks like your figures bear this out.

I leave interpretation to others. Anyone who wants is welcome to interpret the variation as CYCLIC. Our forecast polls always allow for a bottoming out and a cyclic upturn, with output going back up to previous levels. As usual I am interested to know who is willing to predict this will happen, say, this year or next!

Another thing that is interesting is how one seems to get an outraged reaction simply by pointing out the decline in output and current citations, from say 2002 to 2007.

 

[arivero]

1977 127
1976 128
1975 86
1974 59
1973 23
1972 16
1971 3
1970 3
1969 2

It seems that the search is not working right; also FIND DK MODEL,DUAL RESONANCE should be taken into account.

Let me add, for the record, the existence some extraacademical incidents in the early days of string theory. I know of the early passing away of J. Scherk in 1979, but only recently I have read about the "defenestration" of Ramond theory gang in the fermilab around 1971:
http://bama.ua.edu/~lclavell/Weston/
Also, http://www.weeklyscientist.com/ws/articles/geometers.htm that John Schwartz "nearly had to beg for tenure." at Caltech (and it is true that his position was Research Associate for a long period, but I am not sure how usual or unusual it is in the States)