Voting Rights Forum Participant
Post Number: 2
Best of Black Box?
|Posted on Wednesday, June 8, 2005 - 6:10 pm: |
Another thing we found, but I want to confirm this with the source code when we do the technical report: It appears that there is a function written into the code to correct "rounding errors." (Which can only occur when votes that have fractions are added up -- and the only way that can occur is vote-shaving, as far as I know.)
Can you think of any legitimate reason to have a function like that in the software?
Bev, sorry for the delay in replying. I don't have a whole lot to add to what other posters have said. The number of votes is an integer, which can be represented exactly, subject to fitting in the number of bits available. So rounding errors don't apply. If the machine is tabulating vote percentages and the like, then you are dealing with floating-point numbers (as we geeks call numbers with decimal points) and rounding can be an issue. So at this point I can't say this is definitely suspicious. I'm still floored by the idea of the base software transferring control to code that resides on a memory card that has zero security. The only time I've seen this technique used is in space software, where you need to change an unmanned spacecraft's programming while it is in space, so you would send up new instructions by radio, which would be stored in RAM, and the ROM-resident code would have been programmed to periodically check for new code and execute it if it is properly formatted, has the correct checksum etc. Even then there is more security than Diebold seems to have implemented.