The tradeoff here is a classic false-positive vs false-negative issue. If the cost of the bad-bot false positive (which blocks your bot out) is lower than the cost of the bad-bot false negatives (which allow bad bots in), then it is still a good tradeoff, if a suboptimal situation.