On DREs vs. Opscan...

There seems to be a sentiment in the e-voting activist community that

"DREs involve secret recording and counting of votes and optical scanning technologies (Opscan) only involve secret counting."

(that's my paraphrase... I'm not quoting anyone in particular)

Someone has asked what I think about this:

Let's look at this from a scientific point of view. Both DRE and Opscan voting technology involve translating voter intent into digital records. With a DRE, the voter touching a touchscreen (or other form of input) is translated into digital selections recorded in digital storage (and possibly also recorded on a contemporaneously-produced paper record). For Opscan, the voter's marks on a piece of paper are translated into a digital record. So, in both of these cases, there is an individual digital record produced for each ballot from how the machine interprets the voter's intent; these are later (counted) tabulated at either the machine-level or after uploading them to a central computer (a tabulation server).

In both cases, unofficial results typically come from counting (tabulating) the digital records. So, it is the digital records that are counted. In places that require manual audits, some subset of machines or precincts may be manually tallied to serve as a check on the results from digital tabulation. Rarely, save in the cases of 100% recounts or jurisdictions that still use hand counting methods, are all physical ballots recounted.

As to the issue of how a voter's intent is recorded: in both cases, voter intent is translated into digital records. Is this recording secret? Well, that's a loaded term but I would agree that these digital records are not directly perceptible by voters; that is, you cannot see the process that translates the voter's intent into a digital record and you cannot see the result. The tabulation process is equally not perceptible.

The sticking point is that with Opscan technologies (and ballot marking devices, etc.) the direct record of voter intent -- the ballot that the voter marked -- is preserved and available for later analysis. With a DRE, no machine records and preserves the details of how the voter touched the screen (for example) at different times to indicate their preferences.

So, I believe it is incorrect to say that "DREs record and count in secret; Opscan only counts in secret." The correct thing to say, in my opinion, is that both DREs and Opscan technologies count and record digital records translated from voter intent using methods that are not perceptible to voters. Of course, if we care to have any of the added values of DRE, etc. technologies, we have to find ways (like the VVPAT with audits) that will short-circuit this imperceptibility of translation, recording and counting. And some manifestation of voter intent must be preserved -- either directly in the case of Opscan, etc. or indirectly in the case of DRE+VVPAT, etc.