|
I don't agree with Mr. Fleisher's conclusions, but I do agree with many of his points.
Put simply; we already accept machines as part of the election process. Converting some of these machine to computerized alternatives is a natural extension of our current process. Safeguards can be put into place to make these new systems less tamperable then our current systems.
Up through the nineteenth century, voting was essentially a completely unautomated process. Ballots were hand marked, hand counted, hand tabulated, and manually reported. Errors, both accidental and purposeful, can and did happen. Ballots could be "lost", misplaced, miscounted, remarked, etc.
Through the 20th century, a certain amount of automation entered the electoral process. Voting machines and punch-card ballots became common. But these systems were still prone to manipulation. How does anybody really know that a Chicago voting machine was counting every lever the voter switched? How does anybody really know that every California punch-card ballot truly had the proper hole aligned with the name the voter punched? Why do we trust that the machine counting the punch-card ballots was counting them accurately?
What both of these processes above lacked was transparency and repeatability. There has effectively always been some point in the election process where we didn't really know what was happening to our ballots. And with any of the above processes, the exact repeatability of recounts was low. Humans make counting errors, and punch cards are fragile and deteriorate.
We should seize the opportunity of computerized voting to address these problems, and to improve the transparency and repeatability of the entire process.
Here are some things I would propose.
1: All software involved in the election process *must* be open source. All software actually in use must be from digitally signed distributions provided by a trusted source. The software must provide a challenge-response method of verifying the distribution, based upon something like RSA, which would allow representatives of any interested party to randomly audit compliance.
2: All hardware integral to the election process *must* be off-the-shelf commodity hardware available commercially. It needs to be possible for any interested party to assemble for them self a "trusted" hardware/software combination.
3: The voter interacts with a machine that helps them prepare their ballot. After all voter choices are made, the machine confirms them on screen, prints a paper ballot, and instructs the voter to review their ballot.
A note here: this is where I diverge most strongly from Mr. Fleisher's views. I believe that we must put the responsibility on the voter to look at their ballot and verify that it matches their intent. Mr. Fleisher holds the view that we cannot rely on voters to do this. But my proposal puts an additional "check" in the hands of the voters that we don't have now. How does a voter today know that the pattern of holes in their punch-card matches what they want? How does a voter today know that the lever-and-wheel system they used even registered their vote correctly? Think of my proposal so far as essentially being a hand-marked ballot, just that a computer has helped the voter to make sure that their votes are marked in a consistent and legible manner.
4: A key part of my proposal: the machine that is used to CAST the vote has no role whatsoever in COUNTING the votes. All it is used for is to simplify the voting process and produce a very consistently marked ballot.
5: The voter, supervised by an election worker, inserts their ballot into a counting machine. The counting machine is built to the same standards as outlined above (open-source signed software, commodity hardware). The ballots are stacked and retained by the machine.
6: Because ballots were produced and retained, we now have the ability to reproduce the count whenever we want.
7: Provide, by law, that any challenger to the election has the right to recount the original ballots on hardware and software that they provide. This is a *key*. If you don't think the people making the official count did it fairly, you have the right to use your own "known good" tabulator to count the votes. If there is a discrepancy, then there is a problem.
All in all, I don't think we as a nation can retreat from the idea of computerized voting, and if we as a party retreat from supporting it at all, we lose the ability to influence the system we will end up with. When we should be doing (IMHO) is driving toward a system that we can all trust, and all verify.
|