There is a section in the document, "Mythbreakers", by VotersUnite (
http://www.votersunite.org) laying it out.
"Section 301(a)(3) of HAVA requires that each polling place provide at least one voting method that allows disabled individuals to vote in privacy. Accessibility is required; DREs are not."
"Voting systems that record votes electronically (Direct Record Electronic - DRE) are only one of the many available voting systems that provide accessibility for disabled individuals. Alternative voting systems that allow the disabled to vote unassisted are available and cost a fraction of the price of DREs. For example:
Electronic ballot-marking devices, such as the AutoMark by ES&S. (3)
Ballot templates (tactile ballots) like those used in Europe and Rhode Island. (4)
Free ballot-printing software offered by Open Voting Consortium to run on PC systems. (5)" p. 9.
They go on to say that "HAVA does not prohibit punch card and lever systems, etc. "
On page 10: "HAVA also explicitly preserves jurisdictions' rights to use paper ballots. Section 301(c)(2) specifically says that the term "verify" may not be construed to forbid the use of paper ballots. It states:
(2) Protection of paper ballot voting systems.--For purposes of subsection (a)(1)(A)(i), the
term "verify" may not be defined in a manner that makes it impossible for a paper ballot
voting system to meet the requirements of such subsection or to be modified to meet
such requirements."
As far as the Diebold printing technology they've added, it is designed to be extremely difficult to use for counting or recounting. The paper records just scroll into a big roll. That could also be contested in the courts, since the voter's anonymity is not preserved because having all the votes on this big roll preserves the order of the voting.
ES&S is marketing the AutoMark which is accessible to the blind and read with Optical Scan equipment. It can only be used with ES&S, not Diebold, but is a better alternative, if you can get it, as ES&S bought the company that developed AutoMark and now are trying to squelch it's sales.
There are people currently auditing the 2004 election in Georgia and they are finding that the results are a mess. Diebold and electronic voting is touted as being "accurate." That's one of their big selling points. Going over precinct tapes and daily recap sheets and comparing them to the certified results online gives a very different picture, as the poll workers struggle to rectify the differences in their voter certificates (evidence of how many people showed up at the polls) and the count on the DREs.
Having fewer undervotes in Georgia (aka CalTech MIT study) is another point used to sell Diebold's "accuracy." However, a Ga researcher found that the daily recap sheets show that there were zero undervotes in all races in Dekalb County (a county that reported a total of 276,771 votes in the 2004 General Election), but hundreds of "blank votes." What are "blank votes'? It appears the Diebold software is moving hundreds and hundreds of votes into a category called "blank votes," votes not accounted for in the figures given by the Ga SOS to CalTech for their study. That researcher has reported that this is looking like it's the rule across the board, not just for Dekalb county.
I think what we're going to find out after the audit is that the machines are not accurate, and that doesn't even bring into question whether the votes were recorded by the software as the voter intended.