Re: Random choice (was Re: exit polls vs. election results)

From: Edward Cherlin <cherlin_at_pacbell_dot_net>
Date: Sun May 22 2005 - 03:45:01 CDT

On Tuesday 17 May 2005 21:27, Ron Crane wrote:
> On May 17, 2005, at 9:17 PM, Edward Cherlin wrote:
> > On Tuesday 17 May 2005 15:40, Ron Crane wrote:
> >> I prefer publicly-witnessed random choice, since
> >> it's easy for the public to understand and to verify,
> >
> > To understand in a basic way, yes. To verify, no. It is
> > difficult for statistics professionals to verify randomness.
> > I know of many ways to fudge a seemingly-random process, and
> > the experts know many more.
> >
> > This is not an argument against having public witnesses. By
> > all means, let us have yet another open government process.
> > But let us not rely very much on the witnesses for
> > verification.
> The idea is carefully to specify the process for generating
> the random result, not leaving anything to the imagination,
> and keeping everything open to observation. Something like
> this is a start:

Even if you knew how to specify a random process, which you
evidently haven't the first idea of, the witnesses will at best
be able to follow the procedure without being able to verify for
themselves that it is random.

> 1. Each witness volunteers a coin.

Forget it. We use tested devices or there is no point in

> 2. Each witness inspects the others' coins. Only coins that
> pass all volunteers' scrutiny are used for the next steps.

How many witnesses will know how to test for a rigged coin? How
many can do it without equipment by visual inspection?

> 3. The volunteers elect a coin tosser.

Oh, come now. If you're going to have a random process, you can
at least pick the coin tosser's name out of a hat. However,
having a single coin tosser is a security hole in itself.

> 4. Using each coin in round-robin sequence,

A predetermined, non-random sequence is another weakness.

> the tosser
> conducts a sequence of tosses long enough to produce a binary
> number of sufficient magnitude for the purpose in question.
> Endianess is, of course, specified in advance.

I can drive a truck through the holes in that process. Experts
could drive a barge through. You would evidently be astonished
at the large consequences of seemingly small deviations from
randomness that someone can in any way control.

> No doubt there are better processes. But even this one is
> better than simply leaving the selection of challenge
> precincts to the candidates, or to the whims of the officials.

An actually random process, using (for example) the quantum noise
in a diode, is possible. But no amateur knows either what that
means, or how to verify it.

> -R

You know enough to be dangerous, and not enough to be
helpful. :-)

Edward Cherlin
Generalist & activist--Linux, languages, literacy and more
"A knot! Oh, do let me help to undo it!"
--Alice in Wonderland
OVC discuss mailing lists
Send requests to subscribe or unsubscribe to
= The content of this message, with the exception of any external 
= quotations under fair use, are released to the Public Domain    
Received on Tue May 31 23:17:44 2005

This archive was generated by hypermail 2.1.8 : Tue May 31 2005 - 23:17:52 CDT