Keeping tabs on stringy research output

[marcus]

Several people (I think Arivero was one) have shown an interest in data on recent string publication and the question of how properly to interpret it----which basically comes down to whether you can predict.

In sampling the string publication rate, we've never used this exact set of keywords. I have added "compactification" to the list so we will tap into a larger sample. This will be similar to some things we've done in the past but with different numbers. The challenge is to predict stringy output during the FIRST FOUR MONTHS of this year, as compared with the same months in 2002, 2006, 2007...
and the keywords used to get a core sample of research papers are
keywords = superstring, M-theory, brane, compactification, heterotic, AdS/CFT

Here's what the Harvard abstract service comes up with for the first four months of the following years.
2002: http://adsabs.harvard.edu/cgi-bin/n...txt_wgt=YES&ttl_sco=YES&txt_sco=YES&version=1

2006: http://adsabs.harvard.edu/cgi-bin/n...txt_wgt=YES&ttl_sco=YES&txt_sco=YES&version=1

2007: http://adsabs.harvard.edu/cgi-bin/n...txt_wgt=YES&ttl_sco=YES&txt_sco=YES&version=1

1596, 1799, 1776,...
The figure for the first four months of 2008 has not stabilized yet, but will over the next month or so as they finish entering stuff in the database. What will the figure ultimately be?

Will the published output figure be closest to 1200, 1300, 1400, 1500, 1600, 1700, or 1800?
Here 1200 or less, is understood, likewise 1800 or more, in case the real number is out of range. The winning guess is whatever is closest to right.

Ultimately, in a month or so, we will be able to check whose guess is closest by performing the same keyword search:
http://adsabs.harvard.edu/cgi-bin/n...txt_wgt=YES&ttl_sco=YES&txt_sco=YES&version=1

 

[arivero]

Another hint comes from the "top-ten". One gets the SPIRES list of top papers for each year and selects the theoretical ones. It can be seen that they are some long period regularities, and that the number of citations to the top papers keeps more or less constant.

On this optics, in can be said that the label of "string theory" is applied liberally. The top gun is Maldacena conjecture, with the recent raising of some papers related to Ads/QCD. In some way, it was a kind of "silver bridge" for theoretists to scape from strings to field theory, but people has installed their houses in the bridge.

The previous rising, and now starting fall, star was large "at millimeter" extra dimensions, again not exactly the core of string theory but just string-inspired.

An index taking into account only hardcore strings (models, landscapes, m-theory, branes,...) could show a more pessimistic picture. But we should disscuss paper-by-paper

 

[marcus]

Another hint comes from the "top-ten". One gets the SPIRES list of top papers for each year and selects the theoretical ones. It can be seen that they are some long period regularities, and that the number of citations to the top papers keeps more or less constant.
...

Arivero, thanks for pointing out that "top-ten" method of getting stats. I find that I used it in a past thread in response to comments by you and by Demystifier! The thread was back in February
https://www.physicsforums.com/showthread.php?t=215004
and one post is here

I can see from the responses here that there is intense interest in Spires numbers among hep PhD students and others. This is not surprising since many High Energy Physics scientists, including science administrators, members of hiring and funding committees, department chairs and so forth, pay close attention to these numbers. The physics community funds Spires and DESY librarians to generate the data, presumably because they find them useful.

Here is a brief summary of what I've extracted from the Spires numbers so far in this thread.

Number of cites garnered in a given year by the top four recent hep-th string papers

Year    Total Cites
2002    1304
2003    1230
2004     712
2005     649
2006     624
2007     550

By recent I mean published in the past five years. In 2007, those with publication dates 2003-2007. The same emphasis on recent applies throughout.

Number of recent string papers making the Spires Top Fifty, by year

Year  Number of papers   Total cites during that year
2002        10                   2955
2007         1                    225

As shown in the previous table, the greatest drop in citations occurred after 2003, a year associated with the Landscape shock (the 10⁵⁰⁰ different string groundstates) and the promotion of Anthropics as a way out. One can imagine that the drop may have had something to do with this.

Top 10 most highly cited Quantum Cosmology papers. Loop/String citations ratio

Publication    LQC     string     other   Loop/String ratio
1998-2002     366      2168       168      0.17 (approx. 1/6)
2001-2005     802       790         0      1.0
2004-2008     685       188         0      3.6

I've used three different time periods and determined the ten most highly cited Quantum Cosmology papers published during each period. The Loop contingent from the early period totaled only about 1/6 as many cites as the Stringy bunch.
Loop and String papers from the middle period totaled about the same number of cites.
If one looks at the most recent period (2004-2008) the picture has changed and the Loop papers total somewhat over 3 times as many cites.

This probably indicates something about what is going on in these fields, or in the general area of Quantum Cosmology. Different people will want to offer different explanations. It might not mean anything, of course----but then again it could have some significance.

have to go. I will get back and edit out the essentials here

 

[marcus]

Ooops! I had to go to supper and the 30 minutes ran out! I was trying to edit the above down to barebones quotes. Mainly wanted to show the tables. Here is comment on the last table:

Arivero, this last mini-table was actually produced using the "top ten" sample approach that you just now mentioned in the previous post. I took the top ten (most highly cited) QC papers which happened to be published in the three time periods and divided the ten up into LQC, string, and other. Then I added up the cites. What it says, for example, is that if you were a QC researcher and you cited a 1998-2002 QC paper in your article, then it was SIX TIMES MORE LIKELY TO BE STRING than Loop. But if you cited a 2004-2008 Quantum Cosmology paper then it was 3.6 times more likely to be Loop. It is interesting that the ratio has changed and different people would have different ideas as to why and how to explain it.

The last line of the last table can be updated since we now have more information:

Publication    LQC     string     other   Loop/String ratio
2004-2008     750     195         0        3.8

In case anyone is reading who is new to this discussion, the citation count is an indicator of how important a paper is considered to be by other researchers active in the field. As a general rule, the more useful researchers find a paper to be, the more it gets cited in subsequent papers.

I think Arivero is suggesting we look at the top ten most highly cited HEP-th papers of any type, and see how the favored topics shift over the course of time. This is extremely informative and in fact the renowned Michael Peskin used to do this every year for the Spires Topcite lists. Too bad the institution of Peskin's review has lapsed :-(. But we can probably supply a home-baked version.

 

[marcus]

Time to declare the winner of this poll. The final total was 1319. (string publication index for the first 4 months of 2008)

Three people registered predictions. Mine, which was 1200, was the closest.

This thread seems now to have run its course.

Meanwhile, as it turns out, the Social Science subforum seems to be a good environment for prediction polls. I set up a prediction poll for the same index, this time for the first SIX months of the year. We already have 5 people who recorded their estimates. Most recently a physics student at the University of Dublin.

More are welcome. Here is the link
https://www.physicsforums.com/showthread.php?t=233106

Reminder: the point of prediction polls is that you can test how well you understand the present situation by how well you can forecast next month or several months from now.
The aim is not to express what you would LIKE to see happen but what you think actually will happen. In this case----a string publication index down roughly 30 percent from last year---it is something of a mystery what factors are involved. Some could be scientific others environmental of one sort or another. This is not the place to discuss it, so if you have any ideas please go to that link and post them on the Social Science thread! I'd be very interested to hear what people think might be the cause.

Let's not post further on this thread but continue the discussion (anyone who wishes) here:
https://www.physicsforums.com/showthread.php?t=233106