E-index

Okay. I didn’t write any “best of” lists for 2010, or describe my resolutions for 2011. The New Year came and went with nary a whisper. I hope you didn’t miss me too much.

Anyway, the last few days, I’ve been thinking about a new index proposed by my friend and colleague Rich Gott, what he calls the “E-Index” (the E is for Einstein). This is another in a long series of metrics to try to figure out whether scientist “A” is better than scientist “B.” Other important ones include the self-explanatory “total number of citations” and the famous “h-index.”

You can look up the details of the h-index in the link above, but like most good indicators, the basic idea is that a high number is better. Rich’s objection against the h-index is that, Einstein has an h of 27, while Ed Witten has an h of 125, and Rich himself has an h of 46. As Rich puts it:

Any index on which I score higher than Einstein is not optimal!

Mine, for what it’s worth, is a more modest 13, but that still makes me half as good as Einstein by this standard.

There are other flaws with the h-index method. To name a few:

  • Everybody on a 100-author paper gets full credit for the impact of that paper. Einstein wrote most of his papers alone or in small collaborations.
  • Any one of Einstein’s 1905 papers (photoelectric effect, Brownian motion, special relativity), would have been of enormous impact, yet would only count 1 toward his h-index.
  • Rich’s point: Einstein is so famous that we don’t even cite his original papers anymore when we talk about, say, special relativity or Bose-Einstein condensation. It’s just understood, which means that h-index-wise, he doesn’t get any credit.

So Rich’s proposal is for a new E-index is:

  1. For each paper that you’re on, take the total number of citations times a contribution factor. The factor is:
      f=delta_{me,first author}frac{1}{2}+frac{1}{2N}
    So, for a paper in which you’re first of 2 authors, you get 0.75 times the number of citations. If you’re 2nd of 3, you get 0.166 times the number of citations.

  2. Add to that the number of times your last name is cited in a paper abstract.
  3. Add to that the number of times your last name is cited in a paper title.

Einstein’s total, by this standard, is 71,444. Rich suggests measuring people in terms of milli-Einsteins (74 normalized citations each). Alas, by this standard, I measure a measly 5 mE.

This is not a bad metric, I suppose, but in some sense it misses the point. The importance of the h-index (or others) isn’t to compare people at the very top. Feynman has 313mE, while Hubble has a weightier 815mE. Was Hubble more than twice as important as Feynman, or does he benefit from the existence of the Hubble diagram and the Hubble constant? True, Feynman has his own diagram, but even when they’re used in a paper, I bet they don’t appear in the title. And at any rate, nobody would argue that either of these guys aren’t tremendously important.

These metrics really should exist much more to deal with people who aren’t household names. When trying to evaluate people for tenure (by far the most important situation for doing these rankings) it is unlikely that Rich’s abstract or title considerations will normally come into play. What’s more, this particular metric is much more favorable to theorists (who have small papers and things named after them) then experimentalists. Unless you are first author on a major particle physics experiment (which can have hundreds of authors), you’d get virtually no credit by the E-index standards.

There won’t ever be a single standard that will satisfy everybody, but I’m not too concerned about Einstein not getting the credit he deserves.

What’s your favorite index?

-Dave

This entry was posted in Uncategorized and tagged , , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *