A machine-readable ballot, yes, I would recommend that.
We have that in New Brunswick. We had that for municipal elections. I moved to New Brunswick in 2005, so I don't know what kind of ballot they voted with in 2004. But in 2008 and 2012, they had machine-readable ballots. In the 2006 provincial election, there was a manual ballot. In 2010 they had the design for a machine-readable ballot, but it was still counted manually. In 2014 they had machine-readable ballots for the provincial election as well, even though I heard there were some issues with the software. I used to be a programmer, I know what it is.
It makes sense to adopt the machine-readable ballot. Unlike, let's say, text voting or online voting, a machine-readable ballot can be scrutinized. It's paper evidence. It's always there. If there is a glitch, it can be recounted. If, let's say, we were to have a situation like we had in 2014, where candidates were just, I think, nine votes apart, again, that could be recounted. That could be verified manually.
With the system I propose, when we have probably a dozen different preferences on a single ballot, a machine-readable ballot is a great help. That's why my submission also includes a sample of a machine-readable ballot.