>>> Of course pairwise comparisons produce a sorting; in fact, it's fairly
>>> difficult to sort in any other way.

>> Radix sort.

> radix sort requires that the individuals not only be directly comparable,
> but can also be given a valuation in a totally-ordered positional system
> (with known radix, etc.)

My comment on radix sort was a purely pedantic response to the
"difficult to sort other than with pairwise comparison" not intended
as a serious proposal on how to get relative rankings on individual
skill.

> however, for the majority of situations where the bottom line is not so
> clear, it's pretty easy to argue that by the time one has weighed together
> all the qualities and faults of an individual to get a number suitable for
> use with a radix sort, all the hard work has actually been done, and in most
> cases done by means of comparisons.

However, my mention of "bucket sort" was actually a more serious
counterexample, and one more suited to how people actually do relative
rankings. You can quickly group into large sized categories of
similarly ranked items, then within each bucket do a more detailed
comparison. This works for restaurants, wines, golf handicaps and so
on.

-- Charles

Reply via email to