Re: Computer unmasks Anonymous writer...

At 09:20 AM 2/16/96 -0800, Alexander Chislenko wrote:
>  I ran my essays through Word grammar checker a while ago,
>and was surprised how stable the grammar statistics were.
>Complexity of the text (grade level) was the same to the decimal point,
>average length of sentences was consistent, etc.
>People also use the same styles of smileys or *highlights*, make
>consistent spelling errors, have their habits of indentation, etc.

What's the next step?  Writing a program which "fakes" somebody's style, right?

-----BEGIN PGP SIGNED MESSAGE-----

On Fri, 16 Feb 1996, jim bell wrote:

> At 09:20 AM 2/16/96 -0800, Alexander Chislenko wrote:
> >  I ran my essays through Word grammar checker a while ago,
> >and was surprised how stable the grammar statistics were.
> >Complexity of the text (grade level) was the same to the decimal point,
> >average length of sentences was consistent, etc.
> >People also use the same styles of smileys or *highlights*, make
> >consistent spelling errors, have their habits of indentation, etc.
> 
> What's the next step?  Writing a program which "fakes" somebody's style,
> right?

It's been done already.  A program was posted here a while ago that could
actually fake a person's writing style by analyzing word patterns.  It does
a pretty good job, too.  I can e-mail it to anyone who is interested.

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
markm@voicenet.com              | finger -l for PGP key 0xf9b22ba5
http://www.voicenet.com/~markm/ | bd24d08e3cbb53472054fa56002258d5


-----BEGIN PGP SIGNATURE-----
Version: 2.6.3
Charset: noconv

iQCVAwUBMSVSmbZc+sv5siulAQEbTwQAh8+e51WN10DN6P4aLYoWrmLKLN3NWAJa
TW6hvckc/zq8SnFvsRg9TbwLgVon3uxi+OByYyBR98tLEOSKjNVwv/5XZ5M216Zs
UFk/a4fwWq/eJ2qnYkA9ultP7YQjPovJQCI+r+orJ6uLgoMiw4oLCbggpkpqET+X
lADi5zJncaY=
=+wAT
-----END PGP SIGNATURE-----

I wish I could give more information on this matter than I have, but here 
goes.  In the early 80s, when I was at York Universiy, a friend of mine 
in the Mathematics Department took part in a statistical analysis of the 
Letters of Junius, which appeared anonyously in the Morning Advertiser in 
London in the early 1760s.  They were ascribed to almost every writer of 
the age, from Birke to Gibbon.  Macaulay was convinced on external 
evidence that they were by Philip Francis.  Byu analysing the Letters and 
other writings acknowledged by Francis, my friend assured me that 
Macaulay was right.

I believe that similar tests have been run on works ascribed to 
Shakespeare, though I don't know what conclusions were reached.  It would 
be interesting to be able to write software that could mimic a style.  I 
hope that won't happen, since it would strip me of what small advantage I 
have when writing.

Sean Gabb,
Editor
Free Life.

On Fri, 16 Feb 1996, jim bell wrote:

> At 09:20 AM 2/16/96 -0800, Alexander Chislenko wrote:
> >  I ran my essays through Word grammar checker a while ago,
> >and was surprised how stable the grammar statistics were.
> >Complexity of the text (grade level) was the same to the decimal point,
> >average length of sentences was consistent, etc.
> >People also use the same styles of smileys or *highlights*, make
> >consistent spelling errors, have their habits of indentation, etc.
> 
> What's the next step?  Writing a program which "fakes" somebody's style, right?
> 
> 
>