I was wondering the same thing when I recently saw this blog post: https://news.ycombinator.com/item?id=6956690
I thought it would be easy to find out how karma is distributed among users using the HNsearch API.
Unfortunately, it's not that easy because this API is not optimized for fetching massive amounts of data. What I could do was looking at the 1000 users with the most karma. I separated those into ten chunks and looked at the total sum of Karma per chunk of 100 users. This is the result:
http://www.chartgo.com/share.do?id=6cca3d09aa
(Data:
[ (0, 2941707),
(1, 1314760),
(2, 936226),
(3, 745944),
(4, 636533),
(5, 557441),
(6, 493996),
(7, 449999),
(8, 413963),
(9, 384491)]
)
So as you can see the 1st chunk is pretty dominant but then it kind of flattens out. So are HN submissions dominated by a small elite? It obviously depends on your definitions, but I guess not.
It would be interesting to have better data available. Did I miss a better way to query the API? Here's the Python script which I used to query the API: https://gist.github.com/johannes-gehrs/77a92284e3509f2ae960
If you're talking about comment karma, clearly yes. I could write a comment consisting of randomly chosen names from the Chicago phone book and someone would upvote it. Name recognition has that effect, as does a small group of people who seem to follow my comments deliberately (as I do for the list of people in my profile).