What evidence do you have that the """bad guy""" did anything wrong?

I think John Oliver did a deep dive and found in the history of shootings, a small handful were stopped by "a good guy with a gun" while a significant number beyond that were stopped by people without guns.

He went further: the good guy with a gun often were used for disciplinary measures (read: send black students to the school resource officer) instead of guarding the place. Not only that, but in school shootings were more violent where there was good guy with a gun.

That's actually incorrect, the lowest estimate for defensive gun use (by Harvard Phillip Cook using NCVS data, whoops), only counting verifiable police reports and completely discounting defensive display, the most common form of DGU) is 100,000/yr, while gun deaths including accidents, homicides, and suicides, are ~60,000/ yr. That's still a difference of 40,000.

Also good to keep in mind, only 45% of Americans even own a gun, and only somewhere around 20% carry it ever, and even less carry everywhere every day. Combine that with many mass shootings taking place in schools which federally ban firearms or businesses which often have "no weapons" signs on the door (which depending on the state can actually be legally binding), with all of that the chances of there being a "good guy with a gun" in the first place are still pretty slim. If he's a "good guy" even if he brought it he likely left it in the car in accordance with the law and posted signage on the mall (or whatever), and it can't do any good there.

And importantly, a guy running away.

Not just once, it's happened multiple times. There was also the one at the mall where the good guy was then mistakenly shot by the police, that one time 4 store patrons all pulled guns on an armed robber, those are just the ones that I can recall off the top of my head.

Actually Defensive Gun Use (hereafter DGU) estimates vary wildly, from the CDC (self) reported numbers by John Lott and Gary Kleck (500,000-3,000,000), to the more realistic estimate (based on verifiable police reports, and completely discounting the concept of defensive display) by Harvard Phillip Cook using the National Crime Victimization Survey data of 100,000 DGU/yr.

But in any case, by the lowest estimate we have, DGU still happens at least 100,000 "once"s per year, and more if we count defensive display (though that's hard to get numbers on so I leave it out of the "official" count, it does still happen, and likely more often than bullets actually get fired.)