Heap,
Great point, and an important one for the guy asking the question to hear. M&M is, in my opinion, a wonderful system -- but I've got a good gaming group and the time to go over their character concepts and make sure that nobody is useless or broken.
Any experienced gamer who spends a day or two reading M&M will be able to figure out how to make one of the classic broken characters, or some new monstrosity ("Yes, I'm incorporeal, and can only be harmed by energy attacks; also, I have Absorption(Energy), which I use to channel enormous blasts of radiant power back at my enemies every time somebody tries to shoot me. And I have this linked attack form, a combination of Snare, Stun, Dazzle, Mental Blast, Paralyze, and Drain, that's triggered whenever anyone tries to use a Mental Attack on me. I'd like to take "Brooding Loner" as a 10-point weakness, too, if that's okay.")
The GM has to be involved in the character creation process to make sure that that doesn't happen.
The GM also has to make sure that he doesn't kill the heroes. If the heroes are all strongmen and energy blasters, a mentalist with good defenses is not a fun challenge. He's a death sentence.
However, if you do have the time to deal with that stuff, I'll maintain that it's great for 4-color heroics.
Right now, I'm using it, as I said, for a swashbuckler game, but my group occasionally takes a break to play a good old-fashioned game of Justice Pals, which is... fairly tongue-in-cheek.
Doctor Myopic: Welcome, Justice Pals! I wanted you to come see my new robot.
Jinx: Wow, that's great!
Jem: It's not going to go crazy and attack us like the last one did, is it?
Doctor Myopic: No, no, nonono... I've discovered a way to bypass the Hawking Threshold* and stop the robot from becoming sentient and trying to kill all human life. It's a circuit that prevents the robot from thinking of any "Kill the humans"-related idea!
Jinx: Wow, good thinking, Doc! There's no way that could fail!
Robot: Must -NOT- kill humans. Must -NOT- kill humans.
Jem: So... it's just a circuit that stops it from becoming sentient and trying to destroy all life?
Doctor Myopic: That's right. As long as the circuit doesn't get wet, we're all perfectly safe. In fact, to make extra sure, I've protected the circuit behind a flimsy plastic panel. There's no way that FrankenHalSkynet can go crazy and try to kill us.
(sproing)
Jem: Hey, Doc, some flimsy plastic panel on the robot's chest just popped open.
Doctor Myopic: Oh, dear. It always does that.
Jem: Hey, is it starting to rain?
Robot: Must... BZZRT! ...kill humans. Must kill humans.
Jinx: Aw, man!
* The Hawking Threshold is the computational power point at which an artificial intelligence becomes self-aware. In comic-book rules, this means that it immediately decides that the smartest and most productive thing to do is to obliterate mankind as quickly as conveniently possible.