>Computer touch-screen voting systems, like all other voting, have flaws, but one thing they don't have is any ambiguity how you count each ballot.
As someone who has designed questionnaires and computer interfaces, I think you miss a huge point here – electronic systems make vote counting easy because they constrain choice... but there's a lot less ability to verify that an electronic system actually captured the intention of the voter – just as this story shows.
So, how do you do that? 3 parts:
1) Use paper ballots, as I've argued for, above.
2) Count/scan each ballot immediately, before the voter leaves. Reject ones that do not process properly (i.e. the one you described above would be rejected if there were conflicting indicators). I thought I remembered MN doing that (I lived/voted there ~a decade ago), for example, and I'm pretty sure my poling place here in Illinois did that last time as well (haven't been there yet today). It's not an impossible task to enforce the same constraints on a paper ballot, doing so with the voter present and able to clarify/fix their ballot.
3)You try really, really hard to design easy to understand and use paper ballots (for all the reasons I said above, you need paper for an audit trail). Good communication design (i.e.: how you design/layout the ballot forms) matters a lot and most of them are terrible.
That, however, is no excuse for accepting an electronic system which gives up any ability to audit the count in a reliable way (and, unless the voter verifies a physical printout, no electronic system can be reliable, as discussed above).