Re:To Know If the Election Was Hacked, Look at the Ballots
by mjevans on Wednesday November 23, 2016 @02:33PM
(#13025144)
Mail in voting is much harder to attack at scale.
The government mails a form to the address you registered when signing up as a voter. You complete a legal statement with a fairly high penalty for being caught lying. Then you return this legal statement.
There are security issues with this process, but for the most part it's no worse than how humans vote otherwise, and careful message encapsulation designs can make the /cost/ of manipulating individual votes in transit high while making the actual voting form 'anonymous'.
Online voting would only be more secure than this if the message were signed, and audit-able by both the signer and the recipient. This is possible, but it also complicates external review since they would also need to validate the identity of the signer.
At a very minimum a record of /who/ voted (and presumably when) would need to be known as a result of validating voter registration.
Vote tabulation could be separated if the registration validation stage fed a stripped (and trusted by the org) packet of cast votes.
However this presumes that you believe in the security of the /entire/ computing stack from base firmware up through the OS and counting applications. Auditing that is a LOT harder than humans tracking paper; we've had centuries to figure out how to do that in fair ways.
The government mails a form to the address you registered when signing up as a voter. You complete a legal statement with a fairly high penalty for being caught lying. Then you return this legal statement.
There are security issues with this process, but for the most part it's no worse than how humans vote otherwise, and careful message encapsulation designs can make the /cost/ of manipulating individual votes in transit high while making the actual voting form 'anonymous'.
Online voting would only be more secure than this if the message were signed, and audit-able by both the signer and the recipient. This is possible, but it also complicates external review since they would also need to validate the identity of the signer.
At a very minimum a record of /who/ voted (and presumably when) would need to be known as a result of validating voter registration.
Vote tabulation could be separated if the registration validation stage fed a stripped (and trusted by the org) packet of cast votes.
However this presumes that you believe in the security of the /entire/ computing stack from base firmware up through the OS and counting applications. Auditing that is a LOT harder than humans tracking paper; we've had centuries to figure out how to do that in fair ways.