Over on Meta Stack Exchange, someone's reported having received the same post as a review audit in the First questions review queue five times overall. Per this post from a then-employee, when the system generates a review audit, it does not check to see if a given user has previously received the same post as an audit, and just randomly selects an eligible post to generate an audit from.
That the same post was randomly selected as a known-bad audit for the same user five times leads me to believe that the pool of eligible known-bad audit questions is extremely small: per the probability rules, the chances of the same post being selected are one in the number of eligible questions to the fifth power. (Unfortunately, as the post is deleted, I as a <10k user cannot see its timeline, and so don't know how many times it was used as an audit in total, for all users.)
Why is the selectable pool of known-bad first question audits so small, and what can be done to increase it? Are there some classes of questions being excluded from selection (e.g. those asked more than one year ago)?
I'm asking this because even if the feature change requested in the linked report, to check to see if the same post was used to audit the same user, is implemented, this could lead to a dearth of available questions to select once a user has audited all the eligible questions, which means they're no longer being checked for robo-reviewing.