A few quick calculations lead me to believe that all three of the suggested
mechanisms contribute.
1. Water on the surface which is not bound into the crystal structure as
firmly as water within the ice crystals seems to be an acceptable contender.
2. Cliff Swartz calculates 120 atmospheres pressure to lower the freezing
point 1 deg C and then calculates 6 atmospheres for a 4mm skate blade.
Apparently Cliff is not a skater. 4mm is huge! Racing skates use a long thin
blade with no rocker and are flat ground. Figure skates and hockey skates
use a wider blade, but are hollow ground and have some rocker so the entire
blade is not on the ice at any given time. Further more skating is a dynamic
sport and much of the time the load on a blade exceeds the weight of the
skater. Thus, reductions in melting on the order of a degree or so are
certainly within reason.
3. Anyone who has pulled a skater knows that there is friction. Just
estimating the frictional force at 10 N for convenience I get 10 J of work or
2.4 cal for every m the skate slides. This will melt 0.03 cc of water which
will create a film of water 1 mm X 0.03mm X 1 m. Reduce the force, widen my
baldes or whatever and you still get a very substantial film of water. I did
my calculations for ice at 0 deg, but since heat of fusion is 80X heat cap.
there won't be much trouble melting ice that is many degrees below freezing.
Even when coasting, skates leave quite a mark in the ice, much more than can
be accounted for by the loose surface molecules. Some additional melting is
apparent. While pressure certainly helps, it can not do the job alone. My
calculations suggest that frictional melting is the biggest contributer.