Back in March, California decided that after years of negative publicity about the security of e-voting machines (and certainly enough evidence to suggest they weren't very secure) that it would allow independent security experts to try to hack into any machine
before it got approval to be used in California elections. Those researchers have gone ahead and found that every machine they tested was hackable
-- often very easily. The researchers were able to hack into Diebold, Sequoia and Hart InterCivic machines. They didn't get a chance to test ES&S machines because, as you may recall, ES&S stalled before handing over their source code
(and included a nasty threatening letter with it). To be fair, these machines were tested in non-normal conditions, where the researchers had access to all sorts of documentation, the full source code and no election going on where people might spot them tampering with a machine. That is, this doesn't mean that it's necessarily easy to hack an election. It just means that all of the machines have some insecurities -- most of which we didn't know about before. The key here is that we can now understand these insecurities and whether or not they're adequately protected by other measures. What still doesn't make sense is why the e-voting firms are so against this process. All it's really doing is helping those companies improve their products to make them more secure. Of course, one key reason is that the researchers found that many of the security problems are because the machines weren't built with security in mind -- but only had it added as an afterthought. In other words, these companies probably should be redesigning their machines from scratch, which they don't want to do. Of course, does it worry anyone else that the machines weren't designed with security in mind in the first place?