[EM] March 29 Newsweek article on verifiable voting

[Adam Tarr]

Dave Ketchum wrote:

>I do this last pass to warn others - Adam seems not to hear me.

I think I hear you fine, you may dislike the system but it does at least 
address your concerns.  I'm not sure what I've said or done to draw your 
hostilty...

>PROVIDED the programs that did the pretty printing were designed, 
>honorably, for that purpose, they might be useful.
>
>HOWEVER, I ask for and do not hear of verification that the programs are such.
>
>LEAVING a concern that a program could be written and used which appeared 
>to conform, but destroyed secrecy and, perhaps, found a way to falsify 
>ballot counts.

Quoting directly from the article, which I am gathering you just skimmed:

"Even if all the election computers were compromised and running colluding 
malicious software (even having access to unlimited computing power), there 
are only three ways that a system could change a voter’s correctly posted 
ballot without direct detection:

• It could print an incorrect layer, gambling that the voter will choose 
the other layer.

• It could use the same serial number for two different receipts, hoping 
the two voters choose the same layer.

• It could perform a tally process step incorrectly, taking the chance that 
the step will escape selection during audit.

For each ballot and with any of the three approaches, the chance that it 
would go undetected is one half. Thus, the chance that two ballots could be 
changed without detection of at least one is only a quarter, three ballots 
without a single detection an eighth, and so on. Changes in just 10 ballots 
will avoid any detection fewer than one in 1,000 times, and changes in 20 
ballots will avoid detection fewer than one in 1,000,000 times."

So you see, the issue you bring up is addressed there.  Open source 
software is a good idea, but given the security built into this system, 
open source offers FAR less protection to the voters than the nature of the 
system does.