Re: Hand count or open source

[Ralph Shnelvar <ralphs@xxxxxxxxx>]

Dear Kellen:

On Tue, 9 Nov 2004 16:53:48 -0800 (PST), you wrote:

>Please correct me if I'm wrong or my logic is in error:
> 
>1) There is good reason to believe that all current voting software is, in theory, subject to fraud and error.

This is correct.

> 
>2) There is good reason to believe that all current voting software is, in theory subject to undetectable fraud and error.

This, too, is correct depending on the definition of "software."

> 
>3) Even if there were a means to detect fraud and error, every time and without any doubt (should the question arise), the only people qualified to make such a determination would be a tiny number of computer security professionals, whose word the citizens would be required to accept.

This is a loaded statement.  It is loaded just like saying "When someone
gets sick the only people we can believe are the tiny proportion of the
population who are doctors."

Truth has a way of getting out.

> 
>Does open source really change things?  Can't self-deleting bugs be planted, leaving no trace of themselves?

Hard but not impossible.

Much easier, though, is to hack the hardware.  It is easier to change the
locks on the door than to understand the schematics of the lock.

And who will be able to validate that the schematics match the lock?

> 
>If the above are true, then it seems that hand counting is the only current means of vote tabulation that offers a very high level of transparency, verifiability, and accuracy.

Accuracy is in doubt.  That is why a secondary machine count would be
useful.  Plus it would not matter if the software and hardware were
proprietary because you would have independent verification.

It's like counting the cash after it comes out of the ATM.

> 
>Neal's hybrid (opscan simultaneously checked by statistically significant hand count) might be viable, though I'd guess 1% hand count check might not be enough.

It depends on how the 1% were selected.

> 
>What's wrong with handcounting?

It takes a long time.  It's expensive.

Both are worth it.

> 
>What's wrong with Neal's position?

A sample of 1% may not be enough in small elections.  A sample of 1% may be
far too much in a large election.

The devil is in the details of a 1% "random" sample.  If the 1% sample were
taken at the Progressive Club in Boulder the results would show that Kerry
won by a landslide.

Kerry might truly have done that, anyway.  But I hope you see my point.

> 
>Shouldn't an average Joe or Jane be able to understand the fundamentals of vote tabulating, rather than rely on experts?  

Absolutely. 

Trying to teach the American public about small sample bias, self-selection
bias, margins of error, systematic errors, etc., is just too much to ask of
an average Jane or Joe.

It would just be easier to say "We counted this by hand and the machines
said we did it right."

> 
>kell

Ralph Shnelvar

> 
>3)    
>
>Some Guy <someguy@xxxxxxxxxxxxx> wrote:
>Classic. The question was chocolate or vanilla, and you chose strawberry
>with caveats.
>
>-----Original Message-----
>From: Neal McBurnett [mailto:neal@xxxxxxxxxxxxxxxxx]
>Sent: Tuesday, November 09, 2004 8:47 AM
>To: cvv-discuss@xxxxxxxxxxxxxxxxx
>Subject: Re: Hand count or open source
>
>I choose option #3, which was the original citizens request, starting
>over a year ago: BOTH fully-disclosed software/hardware with
>procedural protections, and a hand-counted audit of 1% of the votes.
>
>-Neal
>[|>]
>
>
>			
>---------------------------------
>Do you Yahoo!?
> Check out the new Yahoo! Front Page. www.yahoo.com