[jdrock]

(1/26)^7 .. without getting too fancy with any sort of linguistic probabilities...

[ars]

Well, let's get fancy:

  f	%2.228	0.02228
  u	%2.758	0.02758
  c	%2.782	0.02782
  k	%0.772	0.00772
  y	%1.974	0.01974
  o	%7.507	0.07507
  u	%2.758	0.02758

  =

  1/185,399,389,457

This is global letter frequency, I need a table of first letter frequency.

[ars]

Could not find a good first letter table, so made my own from gutenberg, so:

  f	%3.779	0.0378
  u	%1.487	0.0149
  c	%3.511	0.0351
  k	%0.690	0.0069
  y	%1.620	0.0162
  o	%6.264	0.0626
  u	%1.487	0.0149

  =

  1/487,158,294,227

[gort]

That's the odds of it happening in 7 specific lines. But what are the odds of it happening by chance at any point in time to any person of Schwarzenegger's importance or higher?

[dangoldin]

Just multiply those odds by the percentage of people that qualify as important.

[gort]

And by the number of 7-line sequences produced by each.

[pjvandehaar]

One has to factor in that there are thousands of things he could have printed that would have had roughly the same effect (eg. "Piss off", etc).

[10ren]

= about 8,000,000,000 to 1

But capitalization and spacing was correct too.

[yan]

Capitalization would be correct in almost all cases. You always capitalize first letter and chances are, first word of each line won't be the beginning of a sentence. Ditto for spacing, 3-5 line paragraphs are fairly common.

[jcl]

How do you figure "almost all"? Even if the text could only contain paragraphs of length 3 and 4, the spacing alone would be incorrect in half the cases (assuming that a 3-4 and 4-3 split is equally likely).

(Of course, we are leaving aside the fact that the full body of the message contains an additional one-line paragraph, which could also be considered part of a correctly capitalized and spaced message.)

[rbanffy]

One would have to analyze all previous relevant correspondence out of the AS's office as to measure vocabulary and other structural elements.

It would be a fun project.