There's a fascinating and disturbing article in Science this week. A sociologist at the University of Chicago has looked at citation patterns before and after the growth of full-text online databases and search engines. You'd think that with the enormous growth in the number of accessible journals and powerful search tools that you'd see an enormous growth in the breadth and depth of the citation lists of articles. In fact, it appears that the reverse is occuring: journals from the electronic era cite references that are overall newer than before, and the total number of articles in a specific journal that get any references at all from other articles just plummets. People can get the articles they want, but it appears that the tools are making them narrow their finds.
I'm not at all surprised at the first- i remember back in grad school when I first started using INSPEC and noting how often I would get articles from 1984-> but not before. Gee, that's when INSPEC's database started- it was so much easier to use it than to dig through paper I ended up only using the articles it gave me, and once in a while digging through a reference chain.
The latter effect is both less obvious and more disturbing. Evans estimates that 5 more years of online availability will cut the overall number of cited articles in that journal from 600 to 200. People may be citing more, but they are all citing the same articles and leaving the rest to gather dust. Why this is is unclear- perhaps losing the browsing inherent in searching a paper journal loses the "Hmm, What's this other article about" bit, or perhaps those 200 articles are just better keyworded or indexed than the others. In any case it implies that scientific consensus will happen quicker- sadly because everyone's reading the same articles, not because they are looking at lots of different ones.
Given that nobody in their right mind is going back to the good old days without search engines and full text databases, how do we deal with this problem?
No comments:
Post a Comment