Talk:Rosetta Code/Tasks sorted by average lines of code: Difference between revisions

Content added Content deleted

Inline

@@ Line 38: / Line 38: @@
 == Alternatives ==
 Since the purpose of this section is to discuss how to gauge the complexity of a task, let's start a discussion on alternatives. Why not just allow a vote or something similar in the discussion area of a task to gauge the complexity on some predefined scale: say 1 to 10? --[[User:Rob.s.brit|Rob Britton]] 14:17 18 November 2010 (UTC)
+: Ok...
+::  Instead averaging just lines of code, we could also average character counts and token counts (with a comment being 0 tokens).  Or we could count the average number of lang blocks.
+::  Since token formation rules are language specific we could ask that each language only measure task implementations for that language, and/or we could focus on a specific language.
+::  Instead of averaging over all tasks, we could average over a sampling of tasks, either randomly selected or preselected.  We might also consider independent sampling groups (multiple averages based on different selections).
+::  Instead of averaging totals, we could use average other aggregate measures.  Perhaps unique token counts or averaging the number of occurrences of each character.
+::  Instead of using arithmetic mean as our averaging mechanism, we could use some other kinds of average (median or mode, or geometric mean, or ...)
+::  Instead of finding average we could find variance.
+::  Since we are talking about tasks here and not implementations -- some implementations are incomplete while other tasks may have multiple implementations for the same language label -- perhaps we should instead compare lang blocks to surrounding text?
+: That's all I can think of at the moment, but I am sure that there's other possibilities. --[[User:Rdm|Rdm]] 13:13, 15 October 2012 (UTC)