# very long post tl;dr (and more) at the bottom
Ok. Let's make a list. This will be a list over words that will be banned in a hypothetical game.
Great. Our list, while bad, is now complete. But wait. What happens when someone then writes '*Pízza*'?
Oh well, we'll just add it to our list
Hmm. But now someone could just 'Pizzza', right? Ok, say that somehow we can manage to know every single variation of the word 'Pizza'. Of course, we'd also have to know every other variation of every other word on our list. So:
1. Pizza, etc..
2. Hamburger, etc..
3. Fries, etc..
Now this list is already unusable. There'd be a thousand different variations of each word.
And what if we branch out. In hypothethical game, we actually don't care about people that say's Pizza. It's OK, Pizza isn't inherently offensive to us.
It's *eating* pizza that is the problem.
So would the list now look like this?
1. I like to eat pizza
2. I like to eat hamburgers
3. I like to eat fries
But people could just phrase it differently i.e. '*I enjoy consuming pizza*'. Ok, so we figure out all the different variations of sentences you can use to describe eating pizza.
1. I like to eat pizza, etc..
2. I like to eat hamburgers, etc..
3. I like to eat fries, etc..
So now the list actually may be unending, because there could theoretically be a unending amount of ways to say these things.
Oh, and you also have to avoid every singe variation of every single word in these sentences.
Ok, lets say that you can magically do this. You've found out a way to list every single combination of phrases, including every variation of every word.
What happens when someone says 'I like to eat pizza' and someone else responds 'me too!'?
You'd need a list of all phrases describing eating pizza, with all possible variations of words used in those sentences, and all variations of all responses that would indicate that you agree with that sentiment of enjoying pizza, including all variations of words to possibly use in those sentences.
Ok, you have that. You have the list. The extreme, infinite list of banned things. (Funny thought really, since the amount of banned things to say is infinite, there are as many banned things, as there are allowed things)
What happens when the chatlog looks like this:
> PizzaHater34: Wow, I really hate pizza
> PizzaLover12: I like to eat pizza
No problem, right? Well, now 'AgreeMan34' chimes in:
> AgreeMan34: Me too!
What is he agreeing to? There isn't actually a way to know. Do you punish for this deplorable act, or not?
Ok, let's just say that you have godly powers (which you have, since you've made an infinite list, know every variation of the phrase '*I like to eat pizza*' and every possible variant of the words in that sentence). You make a list that also have perfectly maps every response to every sentence.
No one else can read it. It's infinitely long.
So you summarize it. The general idea is: Don't advocate eating junk-food.
A couple of weeks later, someone write on your forum: "You should really make a list of what can be punished".
This is the problem developers have been facing since the dawn of times. Censorship in games fucking sucks. You cannot make a list.
You and I know exactly what not advocating eating junk food is like. But computers are deterministic. They don't know.
But we can teach them. You and I are pattern finding geniuses. We are experts in the field. We do it impressively well, literally before we are born.
When we see a phrase saying: "I enjoy consuming pizza", we KNOW what it's about. Even if we had NEVER seen that exact phrase before, we know what it means. That's because we are pattern finding experts. We can from our past experiences determine what that phrase means. And only because our brains work so well, can we know what each word means, and dynamically create an understanding of the sentence.
A computer *can* actually do that aswell. You teach it. The principle can be called 'machine learning'. Riot actually employs a subgenre of machine learning: Neural Networks. They function almost as our brains does, but they are initially empty.
You feed the computer data. You show it an example of someone saying they enjoy pizza. Then one more. Do this for as long as you can, but you don't need to give it an infinite list. Also give it list of data with examples of someone NOT saying that they enjoy pizza.
The computer will now be able to take in an input (i.e. '*I like pizza*'), and determine whether or not that input is actually advocating eating pizza.
Riots algorithm works on this principle (Although, I have MASSIVELY simplified it, there are MANY more variables than the ones I mentioned), but it works with toxicity. It can actually do all the things you can't make a list of.
The data Riot uses for it's algorithm comes from reports (although, a single report wont significantly affect it in ANY way to prevent false reports from clogging it up). That way, if something *becomes* offensive, when it previously wasn't, the algorithm can instantly respond.
But a side effect of this is that, just like you can't possibly make a list, the computer can't either.
The list would end up being infinitely long, you generalize it and someone complains that you haven't made a list.
I mostly copied what I wrote above from another comment I just made, so sorry if it seems weird and out of place.
I know I'm also mostly preaching to the choir, but not everyone thinks of computer and algorithms in the same way. Some people genuinely don't know why making a list is impossible, and it's not because they are dumb. For the same reason a lot of people intuitively know why it's impossible, a lot of people think that it is.
All our previous experiences leads us to determine something. We know that it's impossible because we've had the right experiences to know the logistics. We kind of understand how computers and development works.
But a lot of people also see computers as magical things that can do anything, because they've previously experienced incredible things without ever wondering about the underlying mechanisms.
I see a lot of people fall down the rabbit hole of condescendingly explaining things to players on here (I've done it myself, so no hard feelings). And not just towards those who also responds in a hostile way.
I often see posts of players actually wondering why they got punished, and someone just responds '**nice to see the system is working**' or '**I hope you get permabanned**'
So I'd like to think that my (overly long) explanation of why you can't make a list is an example of a way you can explain something (that you find extremely intuitive) without being a dick about it.
If you're a dick about it, you'll just leave a discussion with no parties having learnt *anything*.
All I'm saying is, that there is a reason why Player Behavior gets called unsympathetic and rude. And of course, you don't need to stretch out an explanation as long as I have (it's too late at this point, I might as well have written a fucking novel).
And while it may be an unpopular opinion, it's one thing the moderators of these Boards actually do quite well, IMO. They are often patient and will explain things while not being pricks about it. (Of course, I HAVE seen cases of moderators being pricks about it aswell, and it always leaves a sour taste.)
I initially made this post on EU Boards, but felt like it belonged here too, especially since this is the most active of the two.