There's also self selection: You're only going to submit to a top conference if you think there is a slight chance of your work being accepted. Thus, one could argue that papers submitted to top conferences are already better than average. This means that the acceptance rate is way lower in fact.