Co-op games with randoms are just the worst [part 1]

The Setup
[Player 1] “Alright, let the scouts pick up the energy and we’ll build a turret as soon as we hit 10,000”
[Player 3] “Turret?! We should build a sentry! Why would anyone build a turret?”
[Player 1] “You’re joking right? You can hit enemies half-way across the map with a manned turret. A sentry only shoots like, 10 feet in front of it”
[Player 1] “Wait… Where’d our money go? Who’s spending all our money?
[Everyone stares at an oblivious teammate who’s happily building his 5th spike trap]
[Player 3] “STOP! WHAT ARE YOU DOING?! WHY ARE YOU BUILDING THAT?!”
[No response]
[Player 4] “That does it, I’m outta here.”
[Player 4 has left the game]
[Player 5 has left the game]
[Player 1] “Did everyone just leave? We’re only on wave 9?”
[Player 3 has left the game]
[Player 1 has left the game]

God. I hate randoms.
I love playing multiplayer games. But it only take 1 or 2 to ruin a multiplayer game. This is especially true if you’re playing a co-op game where every player has to pull their weight to win. So what makes a “bad player”?

Here’s my list:
Griefers. Those who get enjoyment by intentionally making others suffer. Usually their own teammates. (These are the worst)
Those who subject others to unwanted verbal or audial abuse.
Those who don’t know what they’re doing, AND have no intention of learning.
Those who quit early.
And I’ll just throw in the catch-all bucket of those who make you go, “What… the… heck is he doing?”

But I’ll just boil all that down to:
Those who don’t act the way we want.

Huh. So it sounds pretty easy to get pissed off by some anonymous, backwards schmo. To be fair, if you throw any random group of 4-8 people together, there’s going to be SOMEONE in that group you can’t stand. And your only recourse is to either bear with it… or quit. This has caused me to change the way I play games online. When I’m on public matchmaking, I keep quiet. Don’t engage unless you have to, don’t draw attention. Just do your job and hope everyone else does theirs. Wait, did I just say “job”? Am I still having fun playing this?

We don’t have to play with randoms! We can… uh…
So why do so many of us put up with this? Because we don’t have friends that can play THIS game with us RIGHT NOW. Maybe they are available and we just don’t know it? Or maybe if they knew we were playing, they’d make plans to play with us? Or maybe we just don’t have friends that have THIS game?

Too many maybe’s, right?

Anyone that can figure out how to solve this issue will have a huge impact on gaming as a whole. So why haven’t the major platforms done so? Sure, Microsoft has their reputation system. But it’s based on negative reporting and its effectiveness is still in question. And PlayStation? Nothing. Steam. Nope. To solve this problem, you’d either have to make sure I never play with the “bad players”, or make sure I only play with the “good players”.

But what’s “good” and what’s “bad” is subjective. ONE solution is to build a tamper-proof personality profile on every player to know what they consider “good” or “bad”. Too many issues with that, right?

More player data required.
Requires constant player feedback to score other players.
Privacy concerns.
It’s time consuming and expensive.
No one else has done this. (unless you’re a dating site)

So are we on a date?
Wait, did I just say dating site? Yes, I did. Now I’m sure there’s tons of articles and posts that say, “Dating sites are just the worst”. But hear me out. I think this is workable. Every player has to describe themselves by a set list of personality traits. Combined with that person’s demographic, you can come up with a pretty full profile on that person. Then to score another player, all you have to do is rate a “like” or “dislike”. Machine Learning will sift through that data and figure out the rest. Netflix does this to figure out what movies you like. Amazon does this to suggest products. And yes, dating sites do this to figure out who you’d like to go out with.

What is Machine Learning? That’s a huge (and complex) topic that’s beyond the scope of this article. The short of it is, you take a lot of data and you give your system half of it and point out what’s statistically relevant. Then you tell it the answer. Player 1 traits + player 2 traits = “like”. Player 1 traits + player 3 traits = “dislike”. You train your system and then feed it the other half of your data and see how it scores. Then you see if what the system thinks should be a “like” or “dislike” matches up with your data’s answers. If you’re close to 85% accurate, then you’re in business.

But I am a unique snowflake!
No, you are not. Given a large enough sample set, there are many others just like you. So given enough data and asking the right questions will eventually generate consistent results.

So leave it up to Skynet?
If you have the cajones to take up THIS solution, then yes. It’s not simple, but it is doable. But like I said earlier, this is just ONE solution. Should the Microsofts or Sonys of the gaming world attempt to take on this problem, this would be their eventual approach.

Now full disclosure: I’m attempting to solve this problem from a different angle with my site: [http://ruingaming.io]. I’ll go into more details in my next post, so stay tuned!

-Albert

[Also, I kept my little story in the beginning purposefully tame. But I’m interested to hear your stories on how a public match w/ “bad players” went horribly wrong! Feel free to comment!]