Various ideas exist about determining the role difficulty in Nethack. The approach applied here results in the ranking depicted below. It is based on player data taken from nethack.alt.org at 2014-10-24, at that time running Nethack release 3.4.3.
The sample set is taken from players with at least 13 ascensions and an ascension ratio of at least 2%, and whose games have a random distribution of played classes with a standard deviation of at most 0.20 (i.e. excluding players who focus on few roles only). The players are grouped in four classes w.r.t. ascension ratio (A: 50-100%, B: 25-50%, C: 10-25%, and D: 2-10%), and in four classes w.r.t. randomness of played classes (<0.10, <0.12, <0.16, and <0.20, biased towards the more random distributed games).
I've grouped the rank results by the ascension class A-D (see above), and built an overall ranking from those. (An unbiased result for all players, the wiki data, and my own ranking can be found for comparison in the last three columns.) The overall result's coloring spans from the easy (green) to the hard (red) roles, and these colors have been used in the other columns to illustrate the specific differences.
Note: braces around pairs of entries denote equal metric values of those entries.
© 2014 – 2018 by Janis | Page content: Creative Commons Attribution 3.0 License