## 03 August 2007

### a bunch of links, and a joke

Ranking is a trap, from Statistical Modeling, Causal Inference, and Social Science, originally from junk charts.

There's a plot that illustrates how common various first names are, and have been in the past, but instead of plotting the frequency with which the names occur against time, they plot the rank of the name in a list of all names. At least there are enough different names that this shouldn't obscure trends too much; if I remember correctly frequencies of first names follow a power law, with the nth most common name having a frequency proportional to n for some constant α > 1. The problem is worse when considering ranks in smaller populations, because then the deviation from some "ideal" distribution is likely to be a lot worse.

(Incidentally, do last names behave differently than first names? I would suspect they do, because people don't choose their children's last names -- or, if they do, they choose them from a pool of two -- while they choose their children's first names from a larger pool. So "undesirable" last names should stick around longer.)

Sum Divergent Series, III, from The Everything Seminar; this is the sequel to the post I mentioned when I was talking about generating functions a few days ago, and shows how to do some more strange sums (for example, 1 + 2 + 3 + 4 + ... = -1/12). It's followed by an interesting post about Mellin transforms. The idea on which the Mellin transform is based is stated as follows:
Generating functions are useful when you can break down Objects into constituent elements whose Size adds to give you the original Size and zeta functions are useful when you can break down Objects into constituent elements whose Size multiplies to give you the original Size.<

Since the integers have both additive and multiplicative structure, I can see how having a tool for going back and forth between those structures -- the Mellin transform, as it turns out -- would be quite useful.

Bivariate baseball score plots is the blog of bivariate baseball score plots, which features plots of the final scores in baseball games; unfortunately it's difficult to get any great insights just from staring at the plots. (This may in itself be a great insight -- namely that the good teams and the bad teams aren't that different.) In other baseball-related news, Strange Maps features The United Countries of Baseball, a map in which various parts of the country are painted with the colors of various baseball teams. There are other such maps -- common census's map is based on actually asking people; Geographer Dan at Baseball Think Factory has a map of which teams are blacked out on MLB.TV in various locations (although these territories overlap) and somewhere (I can't find it) there's another map of his which just shows which is the closest team to each point.

Finally, a joke I made by accident yesterday:
my friend: I have baby bok choi that I really need to use soon. What should I do with them?
me: Wait for it to grow up. Then do whatever you do with regular bok choi.
my friend's girlfriend: That is such a mathematician answer.

of course, I have no idea what one does with regular bok choi. I am not a fan of leaf vegetables.