This week’s Economist includes an unusually poor article on security. It explains that the upcoming Swiss election will be using quantum cryptography to transmit the results from polling stations to central tabulation centres. It alleges that this makes the whole electoral process more secure. This is wrong.
What this is essentially saying is that there would otherwise be a risk of manipulation of this data in transit. The chief polling officer at one station might send a set of figures that get altered by a malicious agent en route to the tabulation centre. Having an encrypted link prevents this man-in-the-middle attack. It does not prevent the polling officer from lying, or the person at the tabulation centre from manipulating the results they input into the counting machines. It doesn’t prevent ballot-stuffing, vote buying, or the compromise of computer systems used to collect or tally votes. In short, it provides no security for the parts of the electoral process that are actually vulnerable to attack. In the absence of good security at the more vulnerable points in the electoral process, using quantum cryptography is like putting a padlock on a paper bag.
Hopefully, they will print my brief letter taking them to task for allowing themselves to be seduced by technology, rather than think sensibly about security.
[Update: 29 October 2007] Bruce Schneier has written about this. Unsurprisingly, he agrees that using quantum cryptography does not increase the security of the Swiss election.