When I was growing up it was a known phenomenon in my family that hot water pipes would freeze before cold pipes in winter. And that using hot water to de-ice your car would cause it to freeze up again more than cold water.
It's something I've always been meaning to test out properly to see if it was real.
Hot water pipes have water running through then less often. Hot water might cause your car windows to fog up, but I doubt the Mpemba effect, even if it's real, could be noticed in that situation.
I worked at a hockey rink growing up, and the ice machines lay down a sheet of hot water after the auger scrapes off the top layer of old ice. I was always told the reasoning was that it froze faster... hardly scientific, but that's the standard in the rink-management world.
Maybe this is also why it is recommended by food persona Alton Brown (among others) to thaw a turkey in the sink with running cold water, not warm, to get the fastest results.
I'm pretty sure that this Mpemba Effect is folklore. The hot water must necessarily pass through the temperature that the cooler water starts at. In fact if we are to believe this than at some point the initially hot water and the initially cold water must, at some point, be the same (average) temperature. Why then would it start to cool faster?
Options that I can think of
1) initially hot water cools very asymmetrically so some hot portion gives of heat (to the air!) faster
2) more rapid temperature change sets up strong convective currents that, because of momentum, stay around, causing the initially hot water to give up heat faster.
These both seem very unlikely. Fluid water is a very good heat conductor (and air a good insulator) making it unlikely that there would be significant temperature gradients and intuition tells us that freezing water takes much longer than it takes any sort of current to die down. (also note that these two effects would act to cancel each other out)
As for the reasons mentioned in the wikipedia article:
1)Definition of frozen: I suppose that this could account for the effect if you measured the time for cold water to completely freeze and hot water to reach 0 degrees C! But in general as long as you use the same metric this won't matter
2) Evaporation (reducing the volume to be frozen): So an initially hot fluid will evaporate more and I suppose result in a slightly smaller volume to freeze. Since the thermal mass is proportional volume and the heat transfer is proportional to area this is a real effect. However the % change in volume of any reasonable shape of water is so insignificant compared to the difference in energy of hot water and cold water that this effect is almost certainly negligible. (perhaps if the claim was hot water freezes faster than cold water if it is laid out in very very very thin sheets I would buy it because of this.)
3) Convection: see above.
4) Frost: Since the water must pass through the same temperature (0 C) for frost to start to build the frost must grow differently on initially hot water than initially cold. I really see no reason to believe this is the case
5)Super cooling: again, if the convective forces persist for a mush longer time than seems reasonable...
6) Solutes and dissolved gases: I mean if one fluid is a different chemical composition than the other then I suppose all bets are off, but I feel the claim of "warmer water freezes faster than colder water" implies they are both the same substance
However the main reason I believe this to be an Urban Myth is that it makes such a good story. It fits our belief as a culture that there are weird things about science that we don't understand and it challenges our initial intuition in a memorable way. In addition it requires more effort to test than most of us are willing to put forth (including me: I would be very interested if someone would actually do the test recommended in the wikipedia article.) Lastly; hot water freezing faster is a much better story than the other way around. Telling people that hot water freezes faster makes you seem to have some kind of secret knowledge; a spell you can cast over ice cube trays to make them into ice faster.
I bet a reasonable rule for judging truth is that, all else being equal, the better story is less likely to be the true one, because if the truth was also the better story then no one would tell the boring one.
However the % change in volume of any reasonable shape of water is so insignificant compared to the difference in energy of hot water and cold water that this effect is almost certainly negligible.
You underestimate the energy the evaporating water takes away. The evaporation energy is much higher than the specific energy. The amount of water that evaporates is not only significant: it also causes the temperature to drop more sharply than you would expect based on only the larger temperature difference. When they reach the same temperature, there is less of the water that was hotter initially. As such, that body of water will be frozen earlier.
One possible cause not mentioned is the expansion of water with increased temperature. One litre of water at 4 degrees weights 3% more than 1 litre at 90 degrees, i.e. there is 3% more water. So this might explain the effect especially if the length of time needed to convert liquid water at 0 degrees to ice, i.e. latent heat, is significant.
With no other substance involved besides the water, I don't see how this could happen. Say you put a vessel of hot water and a vessel of colder water in the freezer at the same time. If the hot water does indeed reach 32F before the cold water does, then there has to be an earlier point where their temperature versus time slopes cross, at which point their temperatures are equal. Once that point is reached, there's no longer any difference between the "hot" and "cold" water.
Since the point of temperature equality can't be reached in negative time, the hot water can't possibly freeze sooner than the cold water that was placed in the freezer at the same time.
So: given that the effect does, in fact, happen, there's some factor at work other than the water temperature. Some property of the container, perhaps? If the hot container loses its heat at a faster rate than the cold container, perhaps it ends up cooler than the surrounding water, and can continue to accelerate the freezing process beyond the point where the temperature slopes cross each other. That'd be my guess.
The requirement for reproducing the result is that the bucket containing the water be a good insulator. A wooden bucket works well, a metal bucket does not.
If you do that then the cooling mostly happens through evaporation. If you start with full buckets of water at different temperatures, at the point where the temperatures cross the bucket that started with hot water has less water in it and therefore is able to freeze faster after that point.
I remember in my first year of physics some people had taken this experiment has class project. No one was able to reproduce that "effect".