There have been previous questions (e.g. here and here) on Physics.SE about the mechanism that makes ice skating possible. Reviewing these, as well external references, it seems pretty clear that the popular "pressure melting" explanation is bunk: the pressure on a pair of typical skate blades is just way too little to depress the melting point of water by any meaningful amount. The currently accepted explanation is that H2O ice at terrestrially-realistic temperatures has a persistent liquid water film that has a lubricating effect. Heat dissipated by friction might amplify this effect a bit.
So, under this understanding of ice's slipperiness, why do ice skates need to be regularly sharpened? Any hockey player will tell you that attempting to get around the rink on a pair of dull blades is a major chore. This would make sense if the pressure melting explanation were true, but it's not, and so it doesn't to me. The simple model of friction (which clearly must not apply perfectly to this scenario) is that the frictional force is independent of contact area. If that were true in this case, then since the source of slipperiness is just the pre-existing water film on the ice, it ought not to matter whether one's skates were sharp or dull: it's just changing the contact area.
I can suggest one possible unsatisfactory explanation: the sharp blades have to push less ice out of their way on the leading edge when they dig into the ice. I say unsatisfactory because "digging in," were it responsible, could be avoided just by spreading one's weight over an even larger area. Would skating on polished steel blocks work just as well as skating on blades? It seems a bit far-fetched.
Presumably, the sharpness of the blades must correlate with the coefficient of kinetic friction between the steel and the ice/water film. What mechanism could be responsible for that?