Tuesday, April 4, 2017

Most downloaded ≠ Most read

Yesterday at ACS San Francisco conference, a paper (and its first author) was awarded for being the "most read" in 2016 by Organic Letters. Apparently, the metric that was used was "most downloaded" article. I have some objections against it. If an article has been downloaded X times, does it really mean that it was read X times? I don't think so. Moreover, what is the source of the downloads of the article? If a professor assigned the article to his class and they all downloaded it, does it really mean that the article was "read"? Maybe it was assigned for an NMR lecture? Since I don't know the details, the metric that was used may have been "most viewed" too. According to the webpage, it was viewed 90,594 times as of now. But "most viewed" is even less reliable than the "most downloaded". Just post the link on Reddit and a few other news outlets and if you get enough attraction, you get 100K views in a day.

According to Web of Science, the article has been cited 0 times (or 1 according to here, or 3 according to Google Scholar) so far. This is not really surprising since it came out only in 2016. But, there are articles that are not "read" as much as this one and they are cited 10+ times already. In the end, I have my own problems with citations anyway. I think being cited also does not mean being read.

I wrote it before, I am against using metrics such as "most read", "most downloaded" and "most viewed" since they tell absolutely NOTHING important to anyone. And they may become new impact factors for the young people who rely on internet more than anything else.

Note: I have no relationship with any of the authors or the research area of the paper. I also congratulate them for their research and publication. I read it in the science news last year and I understand the importance of the discovery.


No comments:

Post a Comment