Hacker News Reader

In the algorithm depicted in the paper, if no elements manage to be eliminated from the set, why not just retry rather than return ⊥?

Retrying would cause errors in the math. Personally I think a better modification would be to get into next round right away.