Predicting the future citations of scientific publications

A new paper in Science magazine has just published a very interesting article (DOI: 10.1126/science.1237825) about quantifying and predicting long-term scientific impact of a publication (and a commentary about it for non-experts DOI: 10.1126/science.1245218). Authors developed a model allowing to predict the future citations for a scientific paper.

I am amazed - its a simple formula, a few years of history (actually they got less uncertainty and errors using 10 years of history) and voila, here you have it - the nature of the scientific publishing is predictable! Or at least it is predictable now, basing the model on past history.

Anyway, authors write: "In contrast with the IF (Impact Factor) and short-term citations that lack predictive power, we find that c∞ (their specific parameter, which means total number of citations a paper acquires during its lifetime) offers a journal-independent assessment of a paper’s long term impact, with a meaningful interpretation: It captures the total number of citations a paper will ever acquire or the discovery’s ultimate impact.

However, I believe, it is important to remember that times are changing, publishing habits are changing, the nature of how scientists cite papers is changing all the time. So, this specific model might not be a panacea for the long term future predictions. But, for shot term I believe it will hold well, and that where the whole fun and new horizons come from!

As a final note, authors write: "... an ultimate understanding of long-term impact will benefit from a mechanistic understanding of the factors that govern the research community’s response to a discovery." Well well, I believe those factors are pretty clean, no? In general, the research community has a very simple response to a discovery - either they find it useful for their work or not. I always say: "Science is like business, the more people happy you make, the greater the success!" :)