http://qs1969.pair.com?node_id=35185

in reply to RE: (jcwren) RE: Signal-to-Noise Ratio

I said upfront that it wasn't my model *g*. Nevertheless, the following axioms make it, if not a neccessarily exact measure of S/N, at least well-ordered:

• You can't lose XP for having one of your articles moderated upward, and you can't gain from having one moderated downward.
• You can't gain (or lose) more than one XP for each point of moderation applied.

Thus, we can at the very least use the sum total of all the moderations of all of a user's articles as an upper bound for the amplitude of "signal", and still compute an S/N ratio based on the value of that bound -- think of it as a "Peak S/N Ratio" within the space of all the possible values "signal" could take on for a given user.

Spud Zeppelin * spud@spudzeppelin.com

• Comment on RE: RE: (jcwren) RE: Signal-to-Noise Ratio

Replies are listed 'Best First'.
RE: RE: RE: (jcwren) RE: Signal-to-Noise Ratio
by BlaisePascal (Monk) on Oct 04, 2000 at 08:19 UTC
You are gonna have to give me an example of how this works...

I have 325 XP, the sum of my reps is 458, over 77 posts. The node with the highest rep has 22 rep (and there only one that high), and the lowest is -1 (also a singleton). Although the mean rep is 5.95 (less than \$NORM=6.17), the mode is much smaller, at 3 (13 writeups), which is much less than \$NORM.

How would you compute my SNR?

Short answer: you can't! You, my friend, are ALL signal *g*.

More specifically, since all we have is an upper bound on how many of your XP come from writeups, and that upper bound is actually higher than your total XP, you lie on the other side of the boundary condition where the two are equal.

I conclude by this that you don't vote much? Either that or you are the victim of some VERY bad luck with Tim's random number generator :)

Spud Zeppelin * spud@spudzeppelin.com