So this is kind of the opposite of #24, but sometimes we have a spider we want to test on 4-5 sites but we don't really want 4-5x the number of recorded responses of single-site spiders.
Some way to continue recording with a different site to get a random sample across multiple sites would be handy.