Speaking to Industry Gamers, id Software's John Carmack touched briefly on the violent games "debate" by saying: "People just play games now and I never took seriously the violence in video games debate. It was basically talking points for people to get on CNN and espouse their stuff on there."
He continues by mentioning how there was an E3 where this was a hot topic, during which time he wasn't supposed to talk about it to reporters. Now that we are about 10 years past the worst of it, as the debate was sparked by the Columbine massacre and the hugely influential if flawed Anderson & Dill (2000) and Anderson & Bushman (2001) studies, Carmack is more willing to publicly share his thoughts on the matter.
Carmack says: "And I really think, if anything, there is more evidence to show that the violent games reduce aggression and violence. There have actually been some studies about that, that it’s cathartic. If you go to QuakeCon and you walk by and you see the people there [and compare that to] a random cross section of a college campus, you’re probably going to find a more peaceful crowd of people at the gaming convention. I think it’s at worst neutral and potentially positive."
It's worth it to look at what he is actually saying here. First of all, there is indeed some evidence to suggest that the process of catharsis can help offline aggression through video game play. Because violent games tend to be more aggressive in nature, these games are usually used for those kinds of studies. And you've probably felt annoyed about something before, and felt better when letting off some steam (catharsis in a nutshell) by shooting some virtual people in the face. Especially if they explode in an over-the-top fountain of blood and gibs to make you smile.
The problem with that is that it's hard to measure if that actually has any long-term effect. You usually have to test this in a laboratory experiment for the same reason most studies are done like that; because it's cheap and it costs a lot of money to run a 10-year longitudinal study. While you might measure a reduction in aggression after playing a violent game, the whole problem with violent game studies is that proponents of the "violent games lead to violent behavior" causal link do the exact same thing, and find different results.
These type of experiments are usually not indicative of actual real-life behavior (i.e., they lack validity outside the lab), so to use the same type of experiment for a positive effect means you are susceptible to the same kind of validity concerns. So yes, catharsis could have a mediating effect on aggression. But it's not the reason why crime rates are dropping while video game play is rising, as Cunningham, Engelstätter & Ward (2011) recently showed in a study that also touches on positive effects of violent game play.
Whether games actually directly lead to a reduction of crime singlehandedly is hard to say, because crime rates are higher where poverty is rampant and it's pretty hard to have a console and a fancy TV if you are born into poverty and live in the ghetto. On average, people who play games the most tend to have a higher income, a better education, and a better family structure or social safety net -- factors that reduce crime. And no, throwing people in jail for longer periods of time does not reduce crime. However, the evidence that violent games do not create disturbed individuals that go on rampages is solid.
The main causal factor for violent behavior is the aggressive personality, which in turn is affected by family factors, genetics, etc. Once someone intends to act out an overtly violent action, this could be done in a way that mirrors what that person has seen in real life, read in a book, saw in a movie, or played in a game. That doesn't mean any of those things cause him or her to do it, it's just a stylistically influenced way to act out what he/she already planned to act out. Hence wearing trench coats after The Matrix because it looks cool.
Whether they are violent in nature or not, video games do keep people occupied and off the street, and they do bring people together regardless of their race or social status. I'm not sure if competitive fist fights are more likely to happen during Call of Duty XP than Quakecon, or whenever someone gets really mad at losing a fighting game, but if they happen it says more about someone's personality than the game they're playing.
Carmack mentions that people at Quakecon are probably a more peaceful crowd than a random cross section of a college campus, and perhaps that's because we're just more peaceful types of geeks who grew up with Doom, Quake, and building our own PCs rather than being stereotypical Type A personality douches with an aggressive drive for competitive dominance through acts of physical superiority. But both types can go to college and they probably both play the same (violent) games.
There are a plethora of studies that show specific positive effects for video game play, from playing Tetris to overcome traumatic memories to creating better eye-hand coordination through first-person shooter gameplay. Since practically everyone plays games nowadays, I'm hard pressed to believe that there is a general causative effect on long-term overt aggressive behavior either way, be it positive or negative.
As Carmack said, the effects are "at worst neutral and potentially positive" which is a good and balanced way to put it.
Doom Creator: Games Are Cathartic, Reduce Aggression and Violence [Industry Gamers]