"Planning to E-Vote? Read This First"

"...E-voting systems have to be completely redesigned with security in mind, McDaniel says. In the short term, this means adding more thorough vote-auditing capabilities so that discrepancies can be investigated. "The elections systems should have the same quality, the same reliability, the same testing and the same certification requirements as financial systems," he says. "If the systems used by banks, which have to report to the SEC [Securities and Exchange Commission], had this level of quality, no one would put their money in the bank."..."

If it's electronic, it's no election.

There are ways of designing an electronic voting machine to be more fraud-resistant than conventional paper ballots. Unfortunately, I've yet to hear of any designs which meet what I would consider to be critical requirements; among them:

There must be a hardware means of ensuring the exact contents of all code storage and persistent memory; code and ballot parameters must be write-protected between the time they are checked prior to polls opening and the time they are checked again after polls close.
Code execution from anything other than the write-protected code storage area must be impossible.
The hardware should be simple enough to permit visual and x-ray examination.
Machines should have attachment points for multiple padlocks; if any padlock is installed, the machine should be locked. Each party would supply its own locks for the machines.
Both parties should be able to read the ballots after the election (the read mechanism must be hardware-interlocked against mischievous writes); party representatives should confirm that they both read the same data, and should exchange digitally-signed copies.

Anything involving any version of Windows requires the ability to run code from RAM, and is thus unsuitable from the get-go. Doing a decent user interface without Windows isn't overly difficult, though; even an 8032 or Z80 should be fine.

Supercat you’re making excellent points.

Too many of these systems are based on the
Windows operating system, and they’re built
in secrecy, often in communist China.

The systems are unreliable and expensive to
store, as well as not secure.

The whole thing is a huge taxpayer ripoff.

Too many of these systems are based on the Windows operating system, and they’re built in secrecy, often in communist China.

Using Windows in a voting machine is not just insane, it's also silly.

The machines don't even need any real text display abilities except for write-in candidates; all other text displays could be generated via bitmaps stored on the software cartridge (which would be verified before and after the election). As for write-in candidates, I would suggest that the proper approach would be to require that all write-in candidates be registered, but that candidates could register at any time (from, say, a year before the election until the close of polls) and get a unique identifier which they would then have to instruct voters to enter. Candidates who registered early by mail or in person at any time at a central location would get a short (say, 5 digit) identifier. Candidates who register at a polling place would get a longer identifier which would identify the polling place and would be unique for each candidate registered there. Such numeric assignment would not only improve the efficiency of tabulation, but it would also prevent ambiguity that could result if ballots are cast for "John Smith", "John F. Smith", "John Francis Smith", "John Franklin Smith", "J. Franklin Smith", etc. not to mention the question of what should happen if "John Smith" wins the race and there's no obvious consensus regarding which John Smith people were voting for.

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.