I would argue that the signal-to-noise ratio outcome-based reason is the principle: it's off-topic. You could also argue another principle: you're censoring a bot, not a human.