Ask HN: I'd like to offer a new HN karma metric but don't know how to build it

2 points by supernova87a ↗ HN
Maybe others are familiar with the "H-index" in academia and other uses, but in case not --

One of the challenges of a karma / rating system is that a user's high rating can come from a very few blockbuster comments / stories, or alternatively posting tons of 1-point crap. You run into this problem when trying to tell whether a potential professor candidate (or maybe even coder) is a productive and respected researcher by his/her published paper count.

So the H-index works like this. Your H-index = how many N times you have posted a comment that has at least N points. Someone whose history has 20 comments that exceed 20 points each has an H-index of 20. It provides a more solid rating of someone's sustained good contributions and a high H-index means someone's comments are both high quality and many.

(You might know this in sports too, for example biking. How many N days have you ridden longer than N miles? That's a H-index score. And it's hard to change with little blips in behavior.)

But unfortunately I'm not an adept coder myself or right now able to use the HN API to figure out how to do this. I see in the documentation that the things you would need to calculate / query this are not available (or maybe I just don't know enough about it):

-- the complete list of HN users

-- linked table from user to all the user's comments

-- score of each comment

(from here: https://github.com/HackerNews/API)

I wonder if anyone else might have an idea about how to do this, or maybe even the HN team. I'm not proposing replacing the karma system which drives the usefulness of the site and people's behavior. I'm just interested in offering a function that can show this to people!

7 comments

[ 3.7 ms ] story [ 30.6 ms ] thread
You can't get scores for other people's comments, only your own. You could crudely infer popularity from position in a thread, but I'm not sure you could actually get something meaningful there

Although can't you just, after accounting for stories posted, divide karma by number of posts and get an average? Doesn't that accomplish what you're after, ie higher average points per post implies higher quality?

> You can't get scores for other people's comments, only your own.

Ah yes, that is a key barrier to doing it. Or maybe this tool would only be to know your own H-index. But that's just a curiosity then.

I guess the dividing by post count is something like the H-index. Although not quite the same information.

Thanks!

But this isn't academia or sports where ppl are focused. Some one here talks about flies one day and the circus the next. And they get points for being novel. So what? The Internet is full of novelty. Over flowing and overloading everyone with it. What is its value?

And tech ppl rank everything under the sun cuz it's easy for them to build. But Good Rankings shrink the info sphere. Random rankings of randomness explode the info sphere.

If you ask me, I would even remove the whole karma system on the account level, and keep it only on the individual comment/submission level. I consider it a feature that there are only a few famous posters, and most of us are generic anons and the overall karma count doesn't really influence much (unless your total is way in the negatives...)
Most are generic anons but in general it seems like there's a culture of keeping your account and not just creating throwaways for single posts. There are some advantages of that for the user, like being able to flag and downvote, not being filtered (posts from new accounts that sign up and post immediately have a higher chance of getting killed), and in some minds the credibility that comes from a long standing high karma account. And having people want to keep their accounts encourages better behavior because you don't want the account banned. I think all this does have an effect that makes the discussions better than they would be in forum with more ephemeral accounts.
The current arrangement probably keeps just one running total on each user. I suspect your solution would be more complex. Since what you are fixing is something that has no particular significance, I wonder if it is worth it.
Sure, I agree. It's just an idea for letting people see their own, and perhaps the top contributors to HN.