Doing double executions every single day is, imo, the best way to narrow down the suspect list a lot. So far, that hasn't happened: because it's quite hard to pull a double off.

Doubles should be difficult because they require greater consensus and more total votes, not because someone can zip in at the last minute and screw everything up. Conceptually, it makes no sense. We're a group of survivors, we've decided A and B are going to die, at the end of the day B agrees A should die, and then we're all just, "Okay"? It doesn't make sense. If the majority of voters even allowed that vote to count, they could still have the time to alter their own votes. We can't do that in the real world because people have different schedules and aren't always available at the end of the game day. So it's metagaming, because it makes no sense given the fantasy of the situation we're in. It only makes sense from an arbitrary rules perspective, which a dwindling group of survivors wouldn't really be concerned with.

There are a few ways to address this:

1. Conditional voting---you place a vote with specific instructions. "I vote for A insofar as it keeps the vote count between B and A equal." Or, "I vote for B if B votes for A." Or whatever. I don't really like this system. It's too clumsy and can be a pain as a host to siphon through.

2. Voting Posses---players are allowed to form their own exclusive voting circles. "I am starting a voting posse that aims to target suspect A and B. I am excluding players A and B from this posse." If A or B disagrees they are then free to form their own voting group. The creator of the group can place additional conditions on who is allowed to cast a vote in that group, stipulating things like maintaining an equal number to carry out a double. Whichever voting group is larger would carry out the execution. You can also still vote without forming a group for the majority of single executions.