Reply #42: Right before the election! [View All]

You check the configuration on the server just before it's loaded onto the machines. You also do a proper logic and accuracy test, using the actual ballots, or touch screen voting, right before the polls open on at least one of every batch of identically configured machines or scanners.

Doug Jones knows whereof he speaks!

As I said, this applies to DREs as well as scanners. They are "programmed" exactly the same way: A configuration is loaded into each machine from a server running GEMS or a similar application used to define the whole election. Typically, the same server is used to aggregate the precinct totals afterward, but that horse has already been beaten to death with various demonstrations and ways to mitigate the risk of tampering with post-election totals on the tabulators. It's a lower risk because the precinct totals will never match the tabulator totals, and precinct totals can be obtained independently or posted on election night (at least in theory). See:
<http://a9.g.akamai.net/7/9/8082/v001/www.democrats.org/pdfs/ohvrireport/section08.pdf>

But if the ballot definition settings are wrong, the totals will be wrong AT THE PRECINCTS and they will always match the tabulator totals in the end -- a much more efficient and dangerous way to steal votes.

Now, when you say Doug Jones said this is "easily detected", I assume he meant only if the ballot definition settings are checked and/or a really good logic and accuracy test is performed on a bunch of machines (which is another good idea but is rarely done in practice, esp. on DREs). Is this right?

In the case of paperless DREs, where there can be no independent auditing of vote totals, this is even more important, because it's a way of determining if the machines have been rigged before the election and especially if errors were made. Once the votes are "in the can" with paperless DREs it's too late so anything you can do beforehand is invaluable.