Sunday, July 20, 2008

Fascinating article on the effect of online scientific journals on citation. Counter-intuitive result : easier availability leads to fewer citations of more popular articles.

Should this really surprise us? Isn't an emergent power-law distribution something we expect from networks with any kind of preferential attachment?

That raises two further thoughts : why should there be preferential attachment? And why wasn't there before, under the old system?

It could be that academics have got lazy now they sit in front of their screens? Perhaps they just pick the first couple of results off the list that Google provides them. But that strikes me as a highly cynical conclusion to jump to just yet. Are academics *reading* fewer papers? Or merely citing fewer?

Another explanation is that when the system had less liquidity, academics would be forced to accept what was available in their library. Perhaps they'd cite -
sub-optimally - papers which had been expensive to acquire and sort of said what was necessary, but perhaps not as comprehensively or clearly as another paper the academic had failed to acquire. Now with cheaper acquisition, academics are more likely to cite the "right" paper, leading to those papers which best encapsulate a common understanding of an idea being more frequently cited. The positive feedback comes in when, being well cited and read, the "A-list" paper comes to define the community's understanding of the idea.

Even if there is a less jaundiced explanation than academic laziness, that doesn't mean this conformity isn't a problem?

Hat-tip Zby

1 comment:

zby said...

My theory is that there always have been that power law and preferntial attachment - it is just that the parameters of the equation changed. The model is the same - just the coefficients changed.