I've mentioned before that when I got into physics, I didn't have some grandiose idea about solving the problems of the universe. I got into physics because I was curious about some of the apparently simplest observations around me. This is why I continue to be fascinated by this Mpemba effect.
I reported this a while back regarding both the theoretical and experimental observation of this effect. It appears that this effect is still being studied and remains a curiosity for others as well. A preprint appeared today on ArXiv detailing a "search" for this effect.
Abstract: An explanation for why hot water will sometime freeze more rapidly than cold water is offered. Two specimens of water from the same source will often have different spontaneous freezing temperatures; that is, the temperature at which freezing begins. When both specimens supercool and the spontaneous freezing temperature of the hot water is higher than that of the cold water, then the hot water will usually freeze first, if all other conditions are equal and remain so during cooling. The probability that the hot water will freeze first if it has the higher spontaneous freezing temperature will be larger for a larger difference in spontaneous freezing temperature. Heating the water may lower, raise or not change the spontaneous freezing temperature. The keys to observing hot water freezing before cold water are supercooling the water and having a significant difference in the spontaneous freezing temperature of the two water specimens. We observed hot water freezing before cold water 28 times in 28 attempts under the conditions described here.
There ya go! This effect certainly has a lot of legs!
Edit 12/20/2010: This article has been published in AJP. The exact reference is:
J.D. Brownridge Am. J. Phys. v.79, p.78 (2011).