    
Brant Lamb
Frequent Voting Rights Forum Participant
Username: Brantl
Post Number: 751
Registered: 01-2005
Best of Black Box? N/A
Votes: 0 | | Posted on Wednesday, August 9, 2006 - 4:36 am: | |
Fwiffo said: "by fwiffo -- Natural limit -- 65535 is the natural limit for an unsigned, 16-bit integer. Most small processors and microcontrollers for this type of device at 16-bit proccessors (some are even 8-bit). Each voting machine never counts more than a few thousand votes (most states have a maximum precinct size of like 1500). It's an elementary programming error and a very common one, and is not likely to be found when testing under ordinary circumstances, because you'd never count that high." This is an ignorant statement; it isn't an elementary error only, it is a logical error that no moderately competent programmer would make. A signed integer would still give you more than 32,000 votes (more than adequate for a single voting machine, don't you agree? A signed integer uses the most significant bit (the left-most bit, like the leftmost digit in decimal, except this is binary) and warns you (in all code beyond assembler) when the sign changes by the result of math operations, which happens when you set that highest bit. Every programmer in his first class is warned of this.
Admin said: "It is an error, and an oversight, but by itself, is not evidence of malice. An intentional security flaw (i.e. a back door) would look much different. " It's evidence of incompetence, at least.
How does this schmuck explain that zero checking on these machines is done by adding together all the candidates vote totals to see that they add to zero instead of each individual count being zero, as anybody knows it must be to be legitimate? |
