If by "clues" you mean something along the order of "now were switching" I would say that ABX actually controls for that with the "X". Please give a practical example of a "double blind" test that is "ears only, no other clues whatsoever". That would probably help clarify.
It is OK to know when there is switching. Indeed it is required for testing.
What has to happen is that you don't know what it is being switched to. Harman speaker listening tests for example puts the speakers behind an acoustically transparent black screen. You know speakers are being switched because there is a 4 to 5 second pause while it shuffles speakers. But you don't know which speaker is what. You hear the sound of the speaker and that is all that you have to judge. Here, you give a preference score of say, 1 to 10 as to whether you like or don't like the sound.
Tests are repeated to assure that random guesses don't get in the way. So if you vote speaker A is a 9 in one instance, and then a 2 in another, then your vote may get thrown out completely due to inconsistency.
The tests are double blind because the sequence is randomized by a computer so the person running the experiment can't control what comes next.