20

In academia, the h-index is a measure of how much of an impact a researcher has. It's defined as follows:

You have an h-index of k if you have published at least k papers that each have k or more citations.

The reason I bring this up is that it might be interesting to adapt this to work for upvotes for questions and answers. For example, we could say that your h-index for questions is k if you've asked k questions that all have at least k upvotes, and similarly for answers. This might be nice because it would help identify users who were consistently doing a good job answering or asking questions, as the metric is robust to outliers. The h-index would let you more precisely dissect reputations to see if the reputation comes from a few really good questions/answers or from a solid body of work. It might be nice to have this in the "top askers" or "top answerers" lists for tags, since right now you can be at the top of a tag if you just ask or answer one really good question, even if it's the only question/answer you've given in that tag.

This request is not to suggest that users who get high upvotes on a single question be displaced from the top user page. Rather, it's a request to display an additional column with consistent contributors rated by the h-index. It could also be simply displayed next to their name, along with total upvotes and total answers provided.

Does this seem reasonable? Unreasonable? Any thoughts?

Improve this question
7
  • 1
    You can probably compute that without too much effort at data.stackexchange.com/stackoverflow/queries – Flexo Aug 24 '11 at 18:39
  • 2
    @awoodland He isn't asking how to see it or where to look for it, but rather requesting that the statistic be shown on the "top users" page. – Lorem Ipsum Aug 24 '11 at 18:44
  • 10
    Here's a query I hacked together. Feel free to play with limiting it to questions/answers only, certain tags, etc. – hammar Aug 24 '11 at 19:04
  • @hammar- Thanks, that's awesome! – templatetypedef Aug 24 '11 at 19:14
  • 3
    And for the curious: Jon Skeet has a combined question-answer h-index of 63, while Eric Lippert has 41. – hammar Aug 24 '11 at 19:26
  • Re: "equally proficient" -- I am certainly no C++ expert; heck, I haven't even written a C++ compiler. (Though I have written compilers of other languages in C++.) However, the reason I concentrate on the C# tag is because I have a professional interest in doing so. – Eric Lippert Aug 26 '11 at 14:58
  • I like "solid body of work". :) However, reputation scores are about popularity; other things are measured by badges etc. I see you answer a lot of old/unanaswered questions; that's what I proposed a new badge for: meta.stackexchange.com/a/254753/248268 – Nemo May 8 '15 at 6:35
10

First off, thanks for the kind words.

Does this seem reasonable? Unreasonable? Any thoughts?

I think you've plausibly made the case that the metric you propose does accurately identify impactfulnesses. (*)

The question then is whether the ability to algorithmically identify highly impactful users is actually of any utility to the users of Stack Exchange sites. Is it? It's not a feature I would use, but perhaps someone else has a use for it. Who wants this information, and what are they going to do with it?

(*) If that's not a legal Scrabble word, it should be. If PACT and NESS are already on the board then you could get a triple-triple-triple word score and a fifty point bonus.

Improve this answer
1
  • 1
    You're a Scrabble king. I would have settled for "fulmines." – nitsua60 Apr 24 '20 at 17:13
1

Propose a different name.

Citation metrics like impact factor and h-index have a lot of connotation and sometimes emotion already loaded onto them. Let's call it something other than h-index.

How about "Square Post Index"

The value is the size of the side of the box where the post count crosses count of posts == score of post. On a plot of count of posts by score of posts the n that satisfies the constraint makes a neat box in the lower left of the graph as x = y.

This Square Post Index query tabulates it for the top 100 users.

Round out usefulness of the SPI with some questions?

  • Is it significantly different than the median post score per user?
  • Is there an easier transform of reputation and posts count that can help answer the same questions?
Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .