published on
tags: Computer Conference

The number of European papers in SOSP and OSDI

I attended The European Conference on Computer Systems (EuroSys), held at ICL in London.

Presentations themselves were of course very interesting, but another thing I was interested in was the General Assembly to discuss how to improve the conference.

Many points were discussed actively, but the most impressive one for me was like this (the quotes are not as-is, rather they are the output after filtered by my brain neural network):

“The percentage of papers from european universites is decreasing in EuroSys. What’s happening? The levels of european system research communities decreasing?

(Some discussions follow)

“They’ve actually been increasing, 10 years ago the numbers of european papers in the top-level system conferences were almost zero.”

This was impressive as the situation is totally the same for us now: The number of asian papers in the top-level system conferences are almost zero. There are few universities doing very well (Tsinghua, Shanghai Jiao Tong), but I think many can agree asian universities (and research institutes like us) should do better than now.

I am too less experienced to claim some reasons for these situations, so rather I’m gonna just survey how european universites have emerged their presence in the systems conferences.

Here I counted the number of papers from european universites and companies in OSDI and SOSP, starting from 1994 when the first OSDI was held.

Year Conf # Accept # Euro # North America # Asia # Other
1994 OSDI 21 2 18 1 0
1995 SOSP 21 2 19 0 0
1996 OSDI 19 0 19 0 0
1997 SOSP 23 1 22 0 0
1999 OSDI 20 2 18 0 0
1999 SOSP 19 0 19 0 0
2000 OSDI 24 2 21 1 0
2001 SOSP 17 1 16 0 0
2002 OSDI 27 1 26 0 0
2003 SOSP 21 2 19 0 0
2004 OSDI 27 3 24 0 0
2005 SOSP 21 2 19 0 0
2006 OSDI 27 0 27 0 0
2007 SOSP 22 3 19 0 0
2008 OSDI 26 2 23 1 0
2009 SOSP 23 2 19 0 2
2010 OSDI 31 3 27 0 1
2011 SOSP 25 0 23 2 0
2012 OSDI 31 1 29 1 0
2013 SOSP 29 3 24 1 1
2014 OSDI 42 4 38 0 0
2015 SOSP 30 4 23 2 1

Here are some notes:

  • Only the first author considered
  • If s/he has two or more affilications, consider all (but not counting twice)
  • Based on the branch locations, not on the headquarters’ (MSR Cambridge, NEC Germany, etc. are counted as EU), but this is error-prone as they may just write “MSR” in the programs
  • Canada included in North America

So the conclusion is that the presence of european universities are indeed growing, even though not as big as my impression. I guess if I look around some other conferences like ATC or NSDI the emergence gets more clear, but I just don’t have the power to do that…