[7919] in cryptography@c2.net mail archive
Re: human failings question
daemon@ATHENA.MIT.EDU (Ray Dillinger)
Wed Oct 4 18:17:14 2000
Date: Tue, 3 Oct 2000 19:36:10 -0700 (PDT)
From: Ray Dillinger <bear@sonic.net>
To: "Nina H. Fefferman" <feferman@Math.Princeton.EDU>
Cc: coderpunks@toad.com, cryptography@c2.net
In-Reply-To: <Pine.LNX.4.21.0010032103510.28648-100000@Math.Princeton.EDU>
Message-ID: <Pine.LNX.4.21.0010031923500.1207-100000@bolt.sonic.net>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
There've been some observations based on captured soviet
one-time pads (in the sixties, when they still had cipher
clerks they were telling "type randomly" instead of using
computers to generate better quality randomness). They're
probably not what you're looking for because the alphabet
is different, but they're interesting. (this is all from
_The_Codebreakers_ by Kahn)
To start with, in one-time pads that were supposed to use the
digits 0 to 9 as their alphabet, the cipher clerks tended to
alternate digits in the 1-5 group with digits in the 6-0 group
more than random chance would predict -- which corresponds to
alternating left-hand and right-hand keystrokes.
When the pad groups were supposed to be in cipher groups of 5
characters, a markedly too large percentage of the groups
started with digits in the 0-5 group -- corresponding to
someone hitting the spacebar with their right thumb and
then starting a new group with their left hand.
Another pervasive effect, and as far as anyone could tell
independent of the tendency toward left-right alternation,
was a marked tendency to avoid sequences where two or
three digits were the same - when a digit repeated, apparently
it didn't look "random" enough to the humans involved, so
even though they knew that doubles and triples should happen
occasionally, they did it far less than a random distribution
would have.
On a binary alphabet, I don't know any good source of info.
Hope this helps....
Ray
On Tue, 3 Oct 2000, Nina H. Fefferman wrote:
>
>
> Hi all,
>
> Does anyone know where (if at all) I can find statistics for the
>predictable strings humans tend to produce when asked to create a
>"random" sequence of zeros and ones? Maybe cognitive science papers?
> Has anyone seen these?
>
> Thanks,
>
> Nina Fefferman
>
>
