Wednesday, December 10, 2008
The publication game
When you are in academia, most of your time is devoted to publishing. The working cycle can roughly be characterized like this:
1. Do some research
2. Write a paper about it
3. Give a talk
4. Go back to 1.
Publications are seen as a sort of currency in the academic world, only that they are not really exchangeable for any goods or services. Still, everybody needs publications: students need to have a minimum number of published papers before they can graduate, professors also need a good publications track to attract more research funding and push ahead in their career and finally the department and the university also need publications to do well in the next ranking.
A downside of the enormous pressure to publish is that there are actually a lot of poor quality papers out there (another reasons for the poor quality of a paper in computer science is that computer scientists are not always the most skilled writers, but that is another issue).
There are a number of places where you can publish your results, but the most important ones are certainly peer-reviewed conferences and journals. Not that all conference or journal would be the same, of course there are also different levels of prestige and every field will have its own top-ranked journals and conferences (yes, these things also have rankings). But even if the paper is published in a good conference or journal does not always guaranty that the stuff is qctually good. Just recently, I read a paper from a top ranked journal in Bioinformatics and was quite shocked to see a lot of flaws in their system. Although none of them was threatening the final conclusion of the paper, it at best shows that the authors were not experts in machine learning algorithms. Still this paper got published. On the other hand, I sometimes feel a bit happy inside when I see a poor paper, because that means we can go and build a better system than theirs! So maybe poor papers eventually have a purpose in this world, too.
1. Do some research
2. Write a paper about it
3. Give a talk
4. Go back to 1.
Publications are seen as a sort of currency in the academic world, only that they are not really exchangeable for any goods or services. Still, everybody needs publications: students need to have a minimum number of published papers before they can graduate, professors also need a good publications track to attract more research funding and push ahead in their career and finally the department and the university also need publications to do well in the next ranking.
A downside of the enormous pressure to publish is that there are actually a lot of poor quality papers out there (another reasons for the poor quality of a paper in computer science is that computer scientists are not always the most skilled writers, but that is another issue).
There are a number of places where you can publish your results, but the most important ones are certainly peer-reviewed conferences and journals. Not that all conference or journal would be the same, of course there are also different levels of prestige and every field will have its own top-ranked journals and conferences (yes, these things also have rankings). But even if the paper is published in a good conference or journal does not always guaranty that the stuff is qctually good. Just recently, I read a paper from a top ranked journal in Bioinformatics and was quite shocked to see a lot of flaws in their system. Although none of them was threatening the final conclusion of the paper, it at best shows that the authors were not experts in machine learning algorithms. Still this paper got published. On the other hand, I sometimes feel a bit happy inside when I see a poor paper, because that means we can go and build a better system than theirs! So maybe poor papers eventually have a purpose in this world, too.
Labels: University
# posted by Daniel @ 5:43 PM 
