I'm also convinced convection is the reason why it can happen under completely equal conditions (which are arguably missing from most experiments).
1. The larger the temperature differential, the faster the heat transfer.
2. The outer layer of hotter water will lose temperature very quickly and generate a convection flow inside the container, which would replace outer layers of now colder water with hot water relatively quickly, accelerating the heat transfer.
3. The outer layer of the colder water will lose temperature much more slowly, thus the convection will be less pronounced, not significantly accelerating an already slow heat transfer.
Of course all this will depend, non linearly, on:
1. The temperatures involved (1C water will certainly freeze faster than 99C water when put in a -25C freezer).
2. Shape of the containers (as it affects the convective flows and hence the heat transfer).
3. Total mass of water: I suspect for small (in range of up to decilitres) and large (beyond decalitres), cold water tends to freeze faster. The first, because convection effects are very limited for both waters, and second, because convection effects on cold water become significant enough to improve heat transfer.
4. Environmental conditions: relative humidity of air in the freezer, temperature of the freezer, is freezer empty, are the walls close or far, the actual freezer performance, etc. All of that affects the heat transfer coefficient and make it potentially non constant and not equal between the two waters.
1. The larger the temperature differential, the faster the heat transfer.
2. The outer layer of hotter water will lose temperature very quickly and generate a convection flow inside the container, which would replace outer layers of now colder water with hot water relatively quickly, accelerating the heat transfer.
3. The outer layer of the colder water will lose temperature much more slowly, thus the convection will be less pronounced, not significantly accelerating an already slow heat transfer.
Of course all this will depend, non linearly, on:
1. The temperatures involved (1C water will certainly freeze faster than 99C water when put in a -25C freezer).
2. Shape of the containers (as it affects the convective flows and hence the heat transfer).
3. Total mass of water: I suspect for small (in range of up to decilitres) and large (beyond decalitres), cold water tends to freeze faster. The first, because convection effects are very limited for both waters, and second, because convection effects on cold water become significant enough to improve heat transfer.
4. Environmental conditions: relative humidity of air in the freezer, temperature of the freezer, is freezer empty, are the walls close or far, the actual freezer performance, etc. All of that affects the heat transfer coefficient and make it potentially non constant and not equal between the two waters.