[E-voting] Compulsory voting

Fergal Daly fergal at esatclear.ie
Sun Sep 11 13:08:39 IST 2005


I'm not very concerned with bit flipping at the count stage, counts can be 
verified and redone using different software and hardware and the input 
exists for further study.

It's at the vote recording that all the danger is located. You mention 
relying on computers but there is almost no situation when we currenlty rely 
on them the way we would rely on them for vote recording. With DRE the 
entire record of the vote exists only as a microscopic pattern of electrons 
and worse still, this pattern can be changed without leaving any evidence 
that it ever said anything else.

Contrast this to banking where each transaction takes the system from a 
known state to a new state (that can be predicted) and both the original 
state, the new state and the transaction stored in multiple places under the 
control of different parties, allowing error detection and correction long 
after the fact.

When recording a vote, the initial state, and the new state must be kept 
secret and the transaction should only be known to the voter. This is 
totally different from any other use of computers.

As for PhD theses, candidates will read and reread paper copies looking for 
errors, they will back them up etc and sadly there are still occasions where 
hard work goes down the drain due to software or user error.

I think in every field where people rely on computers you'll find example of 
the computers failing them. I don't see why DRE should be any different and 
given how important the outcome is, it seems like a bad idea to hope that 
DRE will be the first computer application that the human race is able to 
perfect,

F


On 9/11/05, Craig Burton <caburt at alphalink.com.au> wrote:
> 
> I still assert that bit flipping (if we constrain our argument to this 
> random risk) is a problem which affects all computers everywhere. If it was 
> a serious problem (perhaps due to solar flares) we'd have to rely a lot less 
> on computers for everything, PhD theses, computer games etc as you can't 
> flip a lot of bits in a computer without crashing something (as you assert : 
> different bits have more importance than others bits).
> The risk has necessitated redundant hardware and error correcting hardware 
> in critical systems. Most remote voting systems have lots of redundant 
> hardware at the server end where all the votes go. The tallying should and 
> is done more than once on more than one machine, this being pretty quick and 
> easy to execute.
> In comparison, I can't accept paper vote processing is entirely linear and 
> all forms of error affect one vote. This would imply that no single vote is 
> processed in any way with any other single vote anywhere. Paper votes are 
> boxes and packaged, sorted and piled at lots of places. In a PR count they 
> are counted and heaped together several times. Presumably single errors 
> involving fire or water also affect many votes.
> 
> Craig.
> 
> Fergal Daly wrote: 
> 
> 
> 
> On 9/11/05, Ciaran Quinn <election at polarbears.com> wrote: 
> > 
> > Fergal Daly wrote:
> 
> >
> > > 1 Errors in counting paper votes are almost always on the order of
> > > single votes and even when multiple errors occur they are very
> > > unlikely to accumulate in one particular direction. The "distance" 
> > > between humand counted votes and actually cast votes is at most equal
> > > to the number of errors but is likely to be smaller.
> > 
> > There was an error discovered in one of the Fermanagh-South Tyrone
> > elections in the 80's in which 1000 Nationalist votes had been counted 
> > for the Unionist candidate
> 
> 
> I don't think that contradicts what I'm saying. To achieve that 1000 vote 
> error required 1000 errors. Of course, if it happened because they were 
> dealing with bundles of 1000 and 1 of the bundles was misassigned then it 
> was just 1 error. However at the bundles of 1000 stage, you can afford to 
> recheck many times.
> 
> The point is that for the 999 out of every 1000 steps in the manual 
> process the maximum error that can be introduced is 1, whereas on a 
> computer, an arbirarily large error can be introduced at every single step.
>  
>  F
> 
> ------------------------------
> 
> _______________________________________________
> E-voting mailing listE-voting at lists.stdlib.nethttp://lists.stdlib.net/mailman/listinfo/e-votinghttp://evoting.cs.may.ie/
> 
> ------------------------------
> 
> No virus found in this incoming message.
> Checked by AVG Anti-Virus.
> Version: 7.0.344 / Virus Database: 267.10.21/96 - Release Date: 10/09/2005
>   
> 
>  
> _______________________________________________
> E-voting mailing list
> E-voting at lists.stdlib.net
> http://lists.stdlib.net/mailman/listinfo/e-voting
> http://evoting.cs.may.ie/
> 
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.stdlib.net/pipermail/e-voting/attachments/20050911/dc52b79c/attachment.htm


More information about the E-voting mailing list