response 1 of 11:

Oct 20 05:19 UTC 2003

Here's a link to my op-ed piece which was printed in Sunday's Ann Arbor
News:

http://www.mlive.com/columns/aanews/index.ssf?/base/news-0/106655859347062.
xml

response 3 of 11:

Oct 20 17:17 UTC 2003

Even better, localized counting and cetralized counting used as an 
intentionally redundant system will highlight irregularities at either 
level.

response 5 of 11:

Oct 20 17:53 UTC 2003

I thought Larry pointed out a lot of things that most people don't 
think of.  I liked how he emphasized that many of the problems for 
elections aren't the technology of the machinery but human error.  I 
thought he has a great point (which I've seen elsewhere as well) that 
vote-counting software needs to be open and peer-reviewed.

Larry denounced touch-screen voting because it's tabulated with secret 
software, and isn't reviewable.  I'd add that touch screens wear out 
quickly.  (Visit any hands-on children's museum and just *try* to use 
the touch screen computers.)  Voting machinery is rarely used, and so 
not always well maintained.

In addition to that, I've read (here on Grex, wasn't it?) that touch 
screen voting is known to be more error-prone than even punch card 
machines.  I don't know why that was.

response 7 of 11:

Oct 20 20:54 UTC 2003

Great article Larry - thanks.  I can't imagine any possible argument against
making the voting software open-source, except maybe that if anyone can
compile it, then someone could modify it and substitute a modified version
into machines right before the election.  Can anyone think of a way to
prevent that?  If the software was in some kind of ROM, that would help a
lot, but that's still beatable.

response 9 of 11:

Oct 21 14:56 UTC 2003

Re #7: Physical inspection, done on a random basis before, after, or
even during polling.  It might be worth looking into how the Nevada
Gaming Board keeps casinos from tampering with slot machines to lower
the payouts below legal minimums.  Also, if you have a human-readable
receipt scheme in place for voter verification, you could check a few
sample machine counts against the receipt count.

response 11 of 11:

Nov 27 08:14 UTC 2003

whore.