As I understand it, reputation is a complex subject of my personal evaluation,
not something that anyone else should decide for me. And in particular, the
belief that reputation can be represented by a number and that there exist a
magic reputation-computing formula for that is plainly outrageous.
This means, any such formula is completly arbitrary and playing and tweaking
with that may be fun, but not serious. As such, it can only be a source of
disagreement. Moreover, as I do not see any reason why such a figure is
needed, I find this whole activity as a waste of energy.
Hence my proposal: what is better and simplest is just to publish objective
information. Namely, in primes 1 terms, for each user the numbers of ordinary
posts, blogs, etc, with links to their lists. Let each one make his own
evaluation, in case that he needs it.
And I think that a voting system, if that is wanted, should be article
centered instead of author centered. This means, what is important is to
identify good quality contributions (I think that pagan has already stressed
this point). For instance, in a toy scenario, author A posted two articles,
one very good and one very bad, and author B posted two mediocre articles. It
may well occur that A and B get the same reputation figure. Does it help me to
arrive to the good article by A?
And good or bad articles may be located elsewhere, in what are now called
posts, answers or comments. But the latter are not eligible for voting. I find
this distinction also arbitrary. Moreover when applied to legacy articles.
Actually, I have doubts about a pure voting based system, as it gives room to
manipulations of diverse kinds. Will has promised to keep an eye on that, but
I have doubts that he could cope with all possible maneuvers. May be that a
refereeing complementary system could help. But who would volunteer as referee?