|
|
|
|
|
|
by dredmorbius
526 days ago
|
|
|
A ranking need not have absolute certainty of be fixed over time to be useful. We could well rank into tiers of contributors or participants. I'd generally suggest that such tiers would likely be roughly exponential, with tier n+1 having m^1 more members than tier n, but also a lower net value. (n and m are arbitrary, I'm not insisting on log base 10, and the natural log e might well be a better fit.) This is typical of almost all large network functions which exhibit power laws, Zipf functions, or the like. Measurement itself is difficult and subject to both cost and error, as well as variability over time. |
|
|