Wednesday, February 13, 2013

You don’t cite me anymore - Scientific publications and the ravages of time

One of the most specific things about scientific literature is that scientific papers and books contain references to other papers. The number of citations has become an indicator for the impact of a result, the more other papers cite an article, the higher is its considered impact.

The number of citations a scientific paper gets is a result of interesting effects and interactions:

First, there is the Matthew effect, also known by the proverb "the rich get richer and the poor get poorer" can be observed, where a preferential attachment to larger nodes causing a power-law distribution of node degrees rather than a normal distribution which would be expected for any repeated random experiment with statistically independent trials. Due to this effect the average paper does not get the mean value of all citations, no, it gets close to zero. Most papers do not get more than 5 citations. But a few papers get cited a thousand times or even more often. Models assume that highly cited papers have a better chance of being cited in new papers can explain this behavior and predict a smooth power law distribution for paper citations.

However, to make the model accurate, there is another factor: time.

While the total number of citations for a given paper naturally can only increase over the years, the actual ability of papers to attract further citations dimishes over time - the paper "ages" (see Citation averages, 2000-2010). This applies even to classic papers, for example from Einstein or Hawking, which are no longer cited as they once were.

Matúš Medo and his colleagues from the University of Fribourg in Switzerland developed a model taking this aging factor into account. They found that a paper’s relevance decreases dramatically a few years after its publication.

Especially in our time of instant communication of results, it thus becomes very unlikely that a scientific paper gains in popularity after some time has passed. Sorry to crush your hopes, but if you have a meagerly cited paper now, it most likely won't become more popular in the future ;-)

  1. Matthew Effect. Wikipedia
  2. Matúš Medo, Giulio Cimini, and Stanislao Gualdi.Temporal Effects in the Growth of Networks. Phys. Rev. Lett. 107, 2011
  3. W. Elmenreich. Why is it important to get cited?. Self-Organizing Networked Systems Blog. October 2012
  4. Citation averages, 2000-2010, by fields and years. Times Higher Eduction 2011.


  1. I am wondering whether the absolute publication time, in contrast to the discussed relative publication time, also has an effect on citations. That is, whether a paper published in 1993 has more citations after ten years than a paper published in 2003.
    As the number of publications is increasing, does the probability of a paper to get cited increase or decrease? I would not be surprised if the Matthew effect applied here too, meaning that with more overall publications in 2003, the richer get even richer. But that's just a guess.

    1. The citation averages statistics from THE indicates that contemporary papers receive most of their citations between 2 and 5 years after their publication. I would guess that a strong publication year then mainly feed the papers 2 to 5 years before. So a strong 2003 will have mostly improved the citations for papers from 1998 to 2001. And, due to Matthew, mostly those from these years which had been already well-cited :-)