So the M is one pixel away from symmetry, and one assumes that the symmetrical version was tested and then discarded. One only glances at these things, one doesn’t particularly pore over them, and maybe at a glance this notch reads right, and were that notch filled in it would read as a blotch. You test a notch and you test a blotch, and the notch wins. Want to be a designer, kid? Get used to A/B testing your notches and blotches.