:: makes obligatory comment about the statistics kabbes compiled on posting rates and longevity of posters ::
(Try searching for “kabbes’ hypothesis”.)
Since I don’t believe anybody has pointed it out, I’ll take this opportunity to mention that there’s an odd feature to the kabbes hypothesis.
Formally, the hypothesis is that the probability of a poster still posting after time t is:
p(t) = 0.009134[sup]t2.3[sup]-t[/sup][/sup]*.
But p(t) -> 1 as t -> infinity. More specifically, the probability falls rapidly, then after a year or so it starts to rise again.
Three possibilities suggest themselves.
- The hypothesis is accurate. New members tend to drop out initially, yet are drawn inexorably back to the board after several years. Everybody who has ever posted will eventually become a regular. Since people will continue to register, the number of permanent regulars will grow, lagging in step with the number of newbies. In the next few decades or so this influx of permanent regulars will still fall short of the rise in the global human population. Still, eventually the latter will eventually level off and the board membership will start to catch up. The mathematical limit is that everybody on the planet will become a regular poster. Cecil is worvshipped in every home. Total Enlightenment, at least about factual questions, reigns throughout the species. The lamb lies down with the lion. Fighting ignorance takes longer that we all thought, but the Good Fight is won.
[Quite what happens to the BANNED I leave as the interesting theological exercise.]
-
As above, except that the hamsters eventually explode.
-
The kabbes hypothesis breaks down for large t.
In support of the latter, there’s always the Keynsian objection: p(t) can’t tend to 1, because in the long run we’re all dead. Assuming that the hypothesis works well for small t, it’s presumably because it has a rapid exponential drop, which is then tempered. Most newbies drop out quickly, but some become regulars, who have a much longer half-life. I suspect that a sum of two exponentials would be a better empirical match. At least that makes more sense eventually.
Actually no, although at first I fell into exactly the same thinking. It has to do with figuring the priority of the multiplication and the final -t exponent. (I forget if you multiply first then exponentiate, or exponentiate first then multiply, but there’s one way that works and one that doesn’t.)
The hypothesis needs parentheses to make it clear which way it is supposed to be calculated.
No wait, you’re right. Kabbes’ hypothesis, as written, does increase after t exceeds a certain threshhold. Would you believe it was my misunderstanding of the equation that gave me the values approaching zero. :eek: I misread it as:
p(t) = 0.009134[sup]t * 2.3 - t[/sup]
Which gives me the following (expressed here as percentages):
Yesterday 97.09412
One week ago 87.82784
One month ago 60.80711
Six months ago 4.714165
One year ago 0.2176956
Two years ago 4.88099E-04
Five years ago 5.410361E-12
(Thank Og for the Visual Basic immediate window.)
Perhaps this is what kabbes intended in the first place?
You guys have WAAAAAAAAAAAAAAAAAAY too much time on your hands!
I’ve been to Babelfish, but [inothing* will translate the last few posts. :eek:
I thought it was frowned on to use foreign languages on this board!