Log in

No account? Create an account
07 June 2005 @ 12:31 pm
If there was any doubt before, there shouldn't be now  
Investigation of Diebold machines by Blackboxvoting.org and Bev Harris has revealed that it wasn't the touch-screen machines we should have been worried about - it was the optical scanning machines. Their investigation has determined that not only are they incredibly easy to hack, they seem to have been designed for this purpose:

"It's probably not an accident," Harris said, "because you can look back through the source code to see that [Diebold] went through some programming contortions to keep this thing there. It had to have been expensive for them, frankly."
"When we saw the way they designed it [the ‘built-in']," Harris explained, "Harri [Hursti, computer expert] said 'We have the Holy Grail.' The Elections people are very concerned," Harris said.

The most disturbing discovery in recent memory is the latest report from Blackboxvoting.org:

In another test, Congresswoman Corrine Brown (FL-Dem) was shocked to see the impact of a trojan implanted by Dr. Herbert Thompson. She asked if the program could be manipulated in such a way as to flip every fifth vote.

"No problem," Dr. Thompson replied.

"It IS a problem. It's a PROBLEM!" exclaimed Brown, whose district includes the troubled Volusia County, along with Duval County -- both currently using the Diebold opti-scan system.

I would recommend you read both articles carefully - there is a lot of technical detail.
Annielordindra on June 7th, 2005 06:00 pm (UTC)
Honestly, I think it was more incompetent programmers than anything else.

The main reason I think this is the often quoted figure of things going wrong at 32,000 votes. 32,000 is the approximate limit of a 16 bit signed integer, one of the common datatypes in modern computers. The advantage of it is that being 16 bits, it takes up less space than a standard 32 bit(which can go to over 2 billion for the signed variant). The disadvantage is 32000 is a very small number. When a 16 bit integer exceeds 32000, the results can be unpredictable. Normally, it rolls over to -32000, but under most standards documents, the behavior is undefined, therefore anything can theoretically happen. IT isnt' even implementation defined- you overflow an integer, and you could have it roll over, you could have a total system crash, you could have it just stay at the limit, it could reset to zero... and the system could set it up to choose how to react to this based on the temperature of the CPU, and it would be perfectly standards compliant.

So, I think the root cause was programmers tryign to shave off a little storage space, being too clever for their own good. The 32,000 issues are obviously the case of an incompetent programmer. And thats such a boneheaded mistake that it would take a lot to convince me that there was actual intent to swing the vote. Someone who makes that mistake trying to be clever may well do other things that appear suspicious.

That said, the problems are unnaceptable in any event(though I somewhat prefer incompetence, while unaccpetable in this context, at least it isn't evil). The systems need some heavy duty auditing and testing before they are used again.
Hoc Est Qui Sumusdiscoflamingo on June 7th, 2005 08:17 pm (UTC)
I understand the problems of underflow/overflow, and the needs for rigorous testing. I work on safety critical applications for aircraft - the verification process for our current process is still going strong after a year. Most of these voting machines (specifically, the optical scan machines) run on 16-bit processors, so they are dealing with a well-known ceiling of operation that their tests should have dealt with. That is a sloppy design, audit, testing methodology, or all three.

That the machines do not require flash-loading is suspiciously problematic - most avionics machines require a dataloader to reprogram the machine's EEPROM, specifically to prevent rogue code from entering the machine. That the machine will use the code from the ballot box card instead of the code from its internal EEPROM is especially suspicious. I don't believe that fraud necessarily happened in any of the recent elections, but:

1. For some states/precincts we have no way to audit the process, and can therefore never know.
2. It's definitely possible - specifically, it takes only three insiders to rig an election (with the optical scanning machines).
3. There is no requirement to use certified equipment in an election if hardship can be shown.

If nothing else, this should tell us that their testing and audit process is broken - there are three US governmental agencies that certify embedded devices, and they have no shortage of staff to do audits right now.
Happylittledevilhappy_l_devil on June 7th, 2005 09:24 pm (UTC)
Ug. I hate the world. Seriously, putting the executable on the memory card? And making verification illegal? Fucking meat puppets.
Hoc Est Qui Sumus: Totally Ferretdiscoflamingo on June 8th, 2005 07:00 pm (UTC)
Caves. Meat caves.

I love that phrase.
jarnikles on June 8th, 2005 11:22 am (UTC)
but really, this is just more of a tool in an already corrupt system. im all for restricted tools whose purpose is limited to only corrupt methods ( bombs, especially of the nuclear variety come to mind ), but it is much more important to look at problems at the root of our electoral system. of course its important to bring attention to these tools, but only if it leads to a better understand of how to fix the overall structure of our elections. i mean, really, each time someone talks about whether or not a candidate can sell himself and or some policy idea to the people, an innocent fairie dies. im hoping to see some positive argument for democracy because im starting to, uh, lose the faith. one thing democracy has going for it: its better than fascism.
Hoc Est Qui Sumusdiscoflamingo on June 8th, 2005 06:59 pm (UTC)
Re: eh
I genuinely lol'ed at the fairy dying line.