See my first post in this thread. I have read about this effect before (glad to know it has a name), and my teaching guru, Paul Hewitt, has explained this phenomenon in his taped lectures.
I do believe it can happen, however in a controlled experiment I did using equal amounts of water(220g) in identical containers (cylindrical, open top).
For 6 different initial temperatures ranging from 10 degrees to 80 degrees C, we got a nearly perfect linear relationship: the time to freeze is proportional to inital temerature.
I think it would be useful to try different shapes of containers. I believe if this effect works, you will need a larger open surface area per volume ratio.
I remain sceptical even now that this will work in ice trays.