Sorry, Olivér, but I'm not convinced by your arguments.
For the M moves, it should stay how it is and I agree that M is 2 moves, not 1. It's not WCA's business what method the competitors use. Everyone uses what he wants and the rules should be made regardless what the solving method is.
Imagine what if we had +2s for every first 2 or 3 off moves. Like, 2 off-moves, being +4s. Would it be unfair? Personally I think not, because everyone sees how to solve a puzzle within 2 moves (the same argument goes also for the current regulations), but it would be very impractical to have such a rule. The current regulations with +2s for 1 off move, is a decent solution I think and not exaggerated.
I think you brought the best arguments with Breandan's example, for keeping this rule. It's obvious that +2 doesn't worth the rush, so every competitor will try to finish the solve properly. So, I don't see anyone having an advantage from it. Changing it to DNF will be just a disadvantage for the good cubers, who might miss an average and not get in to the next round, just because of a rush or unluck. If I remember right, it happened to Yumu Tabuchi at WC2009. He got a DNF in the first 2 solves and unfortunately a POP in the last solve, so he missed the final. He might have won at that time the 1st place. Changing +2s to DNF will just increase the number of cases like this and I see no benefit from it.
What about the cases when a move is made when the cube drops on the surface? I think we all passed through this with an Eastsheen 2x2.
I totally agree that for feet, BLD and maybe even bigger cubes, it should be reconsidered.
As I am not a very good cuber, I have no personal interest in this. I am no fanatic of this rule at all and I am open to change it as soon as I see a more convincing argument.
PS - I would agree to change a U2 (a double move) to DNF though - or more precise an over 90 degrees move. I'm ok with that.

