This is an excellent article, a careful analysis indicating that optiscans were less prone to residuals (the proportion of paper or electronic ballots for which no vote for president was counted) than DREs in the 08 elections in NC.
We can only hope that eventually, lawmakers will come to realize that paper is just more secure and less vulnerable to discrepancies of one kind or another, at least when compared to the DREs, which are made to steal elections IMO rather than count votes. Of course, the best system of all would just be HCPBs. But given a limited choice between DRE and optiscan, optiscan should be the choice every time. It's not just common sense; as the Lindeman paper shows, it's also better in terms of technical outcome.
Link:
http://www.ncvoter.net/downloads/Lindeman_Analysis_NC08_Tech_Effect_on_Undervotes.pdf