how are reputation points calculated?

I suspect the "time variable" overwhelms any measure of a weighted average score (i.e., up/down score * number of posts with that score). It's possible that a high score on a recent post wipes out a very low score on an old post. There's some sort of "delta-T" variable. Much too complicated, IMHO. ( Delete or Edit)

Corrected reply to wgstrand:

The "time variable" is not a possible explanation in the case of _bwg_ as you well know. That user's postings receive heavy negative votes almost immediately, and those negatives continue as others encounter his unwanted blog spam.

When less than 8% of 800+ postings receive a positive score, and 91% of all postings ever made receive and maintain negative score averaging less than -3 , there can be no meaningful objective measure that ranks such a poster's reputation in the 81p very good range. By any objective measure such a posting history should earn a reputation score nearly zero.

The only unknown variable that could explain such an erroneous result would be the "length of post" , which would unfortunately give spammers like _bwg_ who repeatedly post huge tracts of cut-and-paste copyrighted material an unnaturally high score in spite of the near universal condemnation of nearly all of his posts by the nearly unanimous members of the blogs in question.

If that is the case, rewarding unoriginal, often irrelevant, excessively long cut-and-paste spam should NOT carry weight over shorter original on-point content.

Let's hope the ID developers address and correct the reputation score algorithm to produce a meaningful measure of a poster's true reputation relative to the participating online community of readers/bloggers. ( Delete or Edit)