[umvi]

There's some truth here, but it's important not to use anecdotes too much to guide intuition because the sample size is often too small and too biased.

For example, I have a group of 10 friends. All 10 of them (including myself) got covid, and we all agree it is no worse than the flu. I mean, yeah, we couldn't smell or taste for a few days, but we all recovered within a week. Therefore... I should reflexively question covid studies saying how dangerous it is because in my own experience it's not even worth making a fuss over?

Maybe to some degree, but the fallacy I committed by relying on my anecdote to guide my intuition is that me and all my friends are young, fit, healthy people. My group of friends didn't include the elderly or immunocompromised or morbidly obese, etc.

[aidenn0]

To put it a different way, the people who grew up to be violent individuals are probably not posting on HN, nor working at companies with people who post on HN, so the fact that everybody on HN isn't violent, despite playing violent video games is only weak evidence against there being a correlation.

In fact, I would be surprised for there to be zero correlation (even assuming there is no causation), as it is a reasonable assumption that naturally violent people are more likely to play violent video games.

[ajuc]

How about the fact that gaming grew by every metric by orders of magnitude between 80s and now and violent crime dropped by half of more everywhere in the world.

If the effect was even 0.01 with growth by 10000% we would have noticed.

[troupe]

I'm not sure that provides the statistical proof it appears to. Assume there were 1,000,000 violent incidents in 1980 and they were all caused by lead in the water. Violent video games caused no violence at a level that would even show up. If today we had 500,000 violent incidents and none of them were caused by lead but all 500k were caused by the prevalence of violent video games.

(I'm not trying to say that violent video games cause real life violence, but the numbers you describe don't offer any type of compelling evidence without other data being considered as well. )

[Mary-Jane]

So you're saying if there were multiple causes and one of them was corrected, his numbers wouldn't catch it, right?

[troupe]

Yes. Total violence going down doesn't prove that violence from video games isn't going up.

[ajuc]

Lets say the effect is 1 additional violent crime per 1000 gamers per year.

With 2.4 billion gamers we should have seen increase by 2.4 million violent crimes per year in last few decades.

It's very unlikely that data hides such an increase, especially given the fact that in different countries game adoption and crime follows slightly different curves so it should be possible to separate changes caused by other factors.

If there's any link it must be negligible for the data to fit.

[zadokshi]

> How about the fact that gaming grew by every metric by orders of magnitude between 80s and now and violent crime dropped by half of more everywhere in the world. If the effect was even 0.01 with growth by 10000% we would have noticed.

Have shootings not gone up? Has no one noticed it.

It’s counter intuitive to think that games that simulate practicing shooting people would have zero influence on people this inclined.

[ajuc]

Shootings have gone down significantly in almost every country in the world.

[aidenn0]

That's actually great evidence! Much better than "I and many of my friends who are currently successful played violent games and we are just fine." The latter is about as convincing as "My grandfather lived to be 90 and he smoked like a chimney."

[buran77]

A good friend of mine has conducted such a study some time ago (unfortunately I don't have a link at hand) and the team's conclusion was that the games themselves had no impact on violence, it was the multiplayer aspect that made the biggest difference. As in the frustration caused by the human factor in the game lead to far more negative sentiments and violence than the game itself with the same mechanics (playing against the game) ever could. This was exacerbated in games that allowed "humiliating" behavior from the opposing players (think fatalities and the like).

I tend to agree and I can confirm with anecdotal evidence from all around real life. Take traffic for example where road rage is always directed at other people no matter how small their trespass, while far more damaging incidents where no other person was involved are met with some swear words (getting 2 flat tires at once, seeing a tree fall on your car, etc.).

It's other people and their perceived trespasses that cause the buildup of frustration and violence.

[paulryanrogers]

This was my personal experience too. And people will find ways to humiliate others in games without such measures like Halo's teabagging.

[buran77]

In real sports such a gesture would be met with widespread condemnation and actual in game punishment, bans, etc. Which is why games following real sports do it the same way. No sports game will allow you to simulate a celebration, or score your points in a manner lacking fair play, trying to replicate real life.

Allowing things like "teabagging" without instant ban for the user, even building dedicated humiliating moves in a game, or emphasizing some animations like slow motion kill cams when a certain score style was achieved are the reason I have a hard time respecting the gaming industry or taking e-sports seriously (purely because of games' design decisions, not the player).

[Mary-Jane]

Given the political bent of this forum, I'm curious how people would react to the fact that the same argument could be applied to the prevalence of guns in America, e.g. gun ownership has been steadily increasing since the 50's, but violent crime peaked in the 90's.

[aidenn0]

Public debate on gun policy in the US is amazingly bad. On one side all regulations and restrictions are bad, on the other side proposed regulations and restrictions are extremely poorly targeted.

The former is probably obvious so here's some points for the latter:

Public debate centers around "assault weapons" high powered rifles and mass shootings.

- 2/3rds of gun deaths are suicides (usually with handguns)

- The vast majority of gun homicides are with handguns

- The broadest definition of "mass shooting" I could find was 4 or more victims (excluding the shooter). That puts mass shooting deaths on par with accidental gun deaths; about 1% of total gun deaths and single-digit percentage of gun homicides.

Now, I'm not saying ending mass shootings is a bad thing, but it's as if everyone worried about traffic fatalities in the US was informed by watching "The Fast and the Furious" and wanted to ban forced-induction sports cars and was super worried about deaths from underground road races.

[aidenn0]

As another aside, California (with the harshest gun laws) is one of the only states to show declining gun deaths since 1981. I'd like to see the homicide and suicide numbers brought out separately, since (in most states) any decrease in homicides is completely shadowed by the huge increase in suicides. With coastal California having a much lower gun-ownership rate, I assume people committing suicide choose other means (nobody I know who killed themselves in California used a gun, while 2/3 of those I know who killed themselves in the midwest did, the third being a woman, who is much less statistically likely to choose a gun).

[mulmen]

Correlation does not imply causation.

[aidenn0]

A lack of correlation is pretty strong evidence for a lack of causation though. (My personal inability to find correlation is much weaker evidence, since who knows how many confounding variables there could be)

[adamsea]

I mean it’s either good science or it isn’t.

[tomp]

> as it is a reasonable assumption that naturally violent people are more likely to play violent video games.

It's also a reasonable assumption that people who naturally want to avoid actual violence play violent video games instead.

This paranoia against violent video games was obviously fake from the start, and such a typically American thing - in line with the American puritan view that blowing up people in films or broadcasting police car chases is a-OK but (female) nipples or saying "fuck" is a big NO NO.

[aidenn0]

- The paranoia was not uniquely American; as a minor, I could walk into a store in the US and buy e.g. Duke Nukem 3D, while my friend in Germany could not do so.

- I'm not making the claim that video games lead to violence, I'm just saying that we should apply the same level of analysis to studies supporting views we already have that we do to supporting views we do not already have. If this study was well done, great! If it was poorly done, then it is appropriate to call it out. If the study truly wasn't able to find any correlation (positive, per my hypothesis, negative per yours), then either there really isn't any correlation (meaning all of the potential forces that would cause a correlation happen to be perfectly balanced (which would be surprising to me), or there is something wrong with the study.

- To be clear, I don't think there is a causal relationship between violent video games and violent behaviors.

- There are likely correlations between video-games and ADHD, as well as single-player video games (actually any solitary activity) and depression. There are links between both of those and violent behavior, particularly self-harm.

[skrebbel]

> a typically American thing

Not only. Eg in Germany there's long been an enormous stigma against violent video games. I haven't talked with German gamedevs for some years so maybe this loosened up since. But basically, if you worked in gamedev you preferred to tell your neighbours and acquaintances that you work "in IT". If they would find out you work on video games they'd pretty much fully cut you out of their lives.

(Note, to my anecdotal experience this is a purely German thing. I don't know other European countries where the same stigma exists)

[jmchuster]

FPSes are cool in the US because of how we glorify our victories in WW2 and the Persian Gulf. I'd imagine it might be opposite in Germany for the same reason.

[verst]

Most German PC games seem focused on trade / mercantile or historical themes. There are also some popular simulation and role playing genres.

- Anno Series (I just picked up Anno 1800 yesterday)

- Fugger / The Guild 1-3 / Patrician etc

- The Settlers / Die Siedler

- Games set in the Dark Eye (role playing framework similar to D&D but more popular in Germany) universe.

[saluki]

This is true in board games. German board games rarely involve combat or taking over territory. Where there are lots of classic American board games with these themes.

[Aerroon]

>In fact, I would be surprised for there to be zero correlation (even assuming there is no causation), as it is a reasonable assumption that naturally violent people are more likely to play violent video games.

There's a lot of violence in society. There's even more violent media available. From news broadcasts to internet forum posts to books. Perhaps violent video games aren't particularly worse at that than all of the other media that portray violence? The news tends to have stories that are more gruesome than the things that happen in video games. The violence in video games is usually toned down or limited.

[gregmac]

> There's some truth here, but it's important not to use anecdotes too much to guide intuition because the sample size is often too small and too biased.

In a community like HN there's also kind of a "meta-anecdote" that emerges. If there was a link, you'd expect to see at least someone posting an ancedote about how their friend got into violent games and then became violent themselves; instead it's mainly the opposite.

I'm still not arguing this observation is scientifically rigorous in any way whatsoever, merely that multiple ancedotes can be more useful than a single one. There could be other reasons such as those that are violent or hang out with those that are don't post on HN - but that also hints there is something more to it then just or even the violent games themselves.

[throwaway0a5e]

>In a community like HN there's also kind of a "meta-anecdote" that emerges

>multiple ancedotes can be more useful than a single one

Sure. But that meta-anecdote based narrative can be highly biased based on the community that forms it.

A forum with an large respresentation of hospital doctors and a forum with a large representation of convenience store owners are going to have completely different meta-anecdote based narratives on whether packing heat is good or bad.

[ABCLAW]

I think there's a fetish for this "don't trust your anecdotes" reply, when the reality is more nuanced.

People will trust their anecdotes because in general they are right to. They don't need a double blind study to figure out that placing their hand on a heated element is a bad idea.

The question is whether the Bayesian confidence an individual places on their anecdotal knowledge is appropriate or not. In this case, plenty of people have a ton of personal experience with the subject matter, a ton of friends and colleagues with experience, and little to no evidence to suggest the contra factual is true.

Does that mean they're certainly right? No. But it certainly lowers the bar to evaluating yet further evidence of the position already held; why waste the time? If a study was released that was solid evidence that games did cause a significant uptick in violence, then examination would be warranted.

I know a bunch of people are going to take a contrarian position, but we operate in the real world with the same time complexity restrictions as the machines we work on. You don't need a double blind study to determine how to position your head while tying your shoelaces in the morning.

[SkyBelow]

The issue is there is entire areas of psychology studying when we get things wrong. If it works until it doesn't, is it really something you should be trusted? Let us not forget that confirmation bias would also apply to how we judge the accuracy of our anecdotes. Is it that we don't really have that many cases where relying on personal anecdotes betrays us, or is it that we just diminish those incidents and pay more attention to when it does go well for us?

[ABCLAW]

>If it works until it doesn't, is it really something you should be trusted?

Depends on why you need to trust something. If the failure state is multiple nukes going off, sure throw money and time at it until your systems are bullet proof. If you've purchased new shoes and need to design a study to figure out which foot you should start with when lacing up, maybe you're going too hard. What about something more chemical in nature; say you have a family recipe you and your family love. You might tinker around with it, but do you stop making it because you are upset by the intellectual sloth of not running double blind studies on Grandma's Famous Cumin Chili? No. Unless something slaps your Bayesian priors in the face, like a news report that says Chili recipes using Cumin might cause cancer.

Again, the idea is that we all come into this with Bayesian priors. You don't need (and will never get) 100% certainty to operate in the real world. I don't need to spend a few hours and $60 to go from 99.5% to 99.7% on this issue and I'm betting you don't either.

We will be, and largely are wrong about a ton of things. And in the majority of cases it doesn't matter one tiny bit. Like here. This is a study about a political flogging point that is 15 years past the point it conceivably could have been made into legislation. The study will be dated by the time there's a political push for the issue to be re-examined.

This is shoelace study v.2.0. We have more pertinent, more complicated and more impactful problems to study both on an institution and personal level, so why devote our limited mental processing time to this? Because you want to be "right"?

Have you ever read a recipe you've done before and thought to yourself "I really shouldn't make this, because I haven't presented this foo

[kbenson]

> Therefore... I should reflexively question covid studies saying how dangerous it is because in my own experience it's not even worth making a fuss over?

Should you question them? Yes.

Should you let your own experience and that of a few people around you override what you see to be a rigorous statistical study because you questioned it and looked into it? No. If you look into it and have serious questions about the rigor of the study? Maybe. Is that the case in the Covid studies? Probably not.

Whether this is about Covid or anything else, the important thing is to be able to asses information as it comes in and reassess your position based on the data available to you. Anecdotal data is invaluable and used by everyone every day to make decisions prior to receiving good outside rigorous data, because for many things we don't have good outside rigorous data.

[ehsankia]

To be fair we have well defined equations to decide if a sample size is large enough and to calculate the error in our errors.

In the case your examples, we already know that the CFR is lower than 10%, so a sample of 10 definitely is nowhere near enough to tell us anything useful. Now if the CFR was 50% and you had 0/10, maybe it'd be worth looking into it.

I'm curious though, have any of your friends had longer term effects? longer term smell/taste loss, brain fog, heart issues, memory issues, breathing issues, etc? The early reports on long term impact are at ~10-20%, so you may have 1 or 2 in your sample, but again your sample is small enough that you very easily could've had 0 too. Again, it depends heavily on what you're trying to disprove and what the rate of it is.

[Quarrelsome]

well to put it another way; in the time that I have lived violent video games have exploded in popularity and violent crime has apparently decreased substantially. That's not to say there's causality between the two but it would definitely support the premise that violent video games at least don't encourage violent crime.

[Natanael_L]

I also don't think there's a strong correlation / causation, but this isn't solid proof.

If there was a causation then the reduction in violent crime could be muted by a small factor due to games, while the net decrease is (as before) caused by an even larger independent factor. In other words, the net decrease could have been larger without the presence of games.

What it does instead show is a maximum plausible effect size, and that it likely isn't worse than the activities which it is replacing.

[WD-42]

What the hell? Were you all practicing CPR on each other or something?