Interesting post about improving web typography. I got totally sidetracked when the author mentioned frequency distribution of letters in the English language:

aaaaaaaabbcccddddeeeeeeeeeeeeeffgghhhhhhiiiiiiijkllll

mmnnnnnnnooooooooppqrrrrrrsssssstttttttttuuuvwxyyz

That got me thinking about Scrabble, and how the ‘x' always seems overvalued, and the ‘v' undervalued. Compare the breakdown: 

          a b c d e  f g h i j k l m n o p q r s t u v w x y z

English : 8 2 3 4 13 2 2 6 7 1 1 4 2 7 8 2 1 6 6 9 3 1 1 1 2 1

Scrabble: 9 2 2 4 12 2 3 2 9 1 1 4 2 6 8 2 1 6 4 6 4 2 2 1 2 1

Which means the letter distribution is pretty right - amazingly it has never been changed nor needed to be adjusted since invented by Alfred Butts in 1938, who famously did things like hand counting the letter distribution on the front page of the New York Times to work out the numbers. ‘H' is about the only really wonky one.

Next look at the points per letter:

         a b c d e  f g h i j k l m n o p q  r s t u v w x y z

Tiles  : 9 2 2 4 12 2 3 2 9 1 1 4 2 6 8 2 1  6 4 6 4 2 2 1 2 1

Points : 1 3 3 2 1  2 2 4 1 8 5 1 3 1 1 3 10 1 1 1 1 4 4 8 4 10

Again, it’s all pretty spot on - more points for the scarcer letters and those hard to use. But given their similar scarcity, ‘x' being worth 8 when ‘v' is only worth 4 seems slightly wrong. Especially when “ox” or “ex” or “xi” on a two way triple letter can be win you the game, whereas using a ‘v' is tricky and not rewarded accordingly.

(via Tim Bray)