If you think matchmaking is rigged, read this

Multiplayer Discussion
1

So one common piece of “evidence” cited by the “matchmaking is rigged” crowd is that if you win enough, eventually you’ll come across a wall of counter decks which will make it very difficult to advance.

Assuming that the game of Hearthstone more or less follows a “paper rock scissors” metagame, this is demonstrably true. If you don’t change your deck archetype, and you keep winning, you will face a wall of counter decks.

The “matchmaking is rigged” crowd then goes on to contend that this is evidence that the matchmaking algorithm scans your deck list to deliberately match you against counter decks. And this is the point at which the “matchmaking is rigged” crowd messes up. This effect is a natural conclusion of a winrate-based matchmaking algorithm plus a paper rock scissors metagame. No deck scanning is required; if you have those two things, the Counter Deck Wall is the inevitable result.

Because it’s important to understand that winrate based matchmaking is not random. It can be random in regards to everything BUT your rank or rating, but it isn’t random in terms of your rank or rating. Your rank/rating is winrate dependent, and your winrate is, to some degree, deck-choice and metagame dependent.

Perhaps you, dear reader, are having trouble visualizing what I’m describing here. Which is why I created evidence to share with you.

In my weaponized boredom, I created an Google spreadsheet that simulates 1024 players playing 64 rounds of strictly winrate-based matchmaking. Here are the rules of the simulation:

  1. Each player starts off randomly choosing paper, rock or scissors. They can’t change this throughout the matches; the choice is locked.
  2. Matchmaking is strictly by winrate first, with randomness only as a tiebreaker if multiple potential opponents have the same winrate as you do.
  3. Unlike normal paper rock scissors, in the event of both sides being the same type, the winner is determined by coin flip. No ties allowed.

I ran this simulation five times (each took my phone about 3 minutes just to process). And then I made a chart of how many of each choice (paper rock or scissors) was at each rank.

Here are the results:
https://imgur.com/gallery/fs5IxOC

And if you’re nerdy and want to see the spreadsheet, here it is:
https://docs.google.com/spreadsheets/d/1M_G7FbziIY3gavKK5_NYbMRWMWWt2RFCTRcuzdVjFSE/edit?usp=drivesdk

This effectively proves that standard matchmaking by winrate will clump paper rock and scissors into like groups over time, such that as you climb in ranks, you’ll tend to face favorable matchups, then even matchups, then unfavorable matchups, then (if you can break through the Counter Wall) favorable matchups again, at which point the cycle repeats.

No deck scan required.

13 Likes
2

You’re asking irrational people to consider rational and logical evidence in an attempt to educate them on how statistics work and what statistics in randomness looks like and how randomness can create what appears to be non random data.

I admire your effort, but the chances of this even convincing or educating just 1 person who is on the side of “it’s rigged!” I feel is very low.

I’ve taken full statistic classes and sat through demonstrations and rarely does someone come out of those truly able to understand why our minds reject real randomness and forcefully looks for patterns.

I just don’t see that happening here. I think statistics on randomness is one of those things you either get it right away and understand the human mind or you likely don’t ever get it.

I can have hope that it finally clicks for someone from that rigged crowd and they post here they finally get it, but I don’t think you’ll get that here.

We can always try though, so good effort.

7 Likes
3

Le sigh.

This is not about randomness. This is about the NON-random effect of matchmaking by rank/rating. Such an algorithm (without additional help from additional algorithms) takes a random chaos of a starting position and over time orders it nonrandomly into a predictable pattern.

Random in; less random out. Over a theoretically infinite timeline, no randomness remaining.

Perhaps the point of this post should be to educate the “matchmaking is NOT rigged” crowd to stop using the randomness strawman. Matchmaking by rank/rating is NOT random. This proves that it isn’t.

3 Likes
4

The “It’s Rigged” will take this as proof the game is rigged and they were right all along. Except that no one is getting more or less rigged upon than the next guy under these rules. We are all playing in the same field with the same rules. The “It’s Rigged” call these rules rigging everyone else calls these rules impartial.

1 Like
5

I think you’re mixing what the rigged argument is.

The claim isn’t that the pool of people you face is random, the claim is that the next person in that pool isn’t random.

Your efforts show that the pools of people and decks aren’t random as far as I can tell, but the problem for people is that they see patterns in their matches and then determine that the person it chose for them to face is not random.

So if they end up in D5 that’s filled with Rock and they’re playing Scissors… they aren’t saying it’s not random that rock exists a lot in D5, they are saying it doesn’t matter what else is in D5, they’ll face Rock if the program wants them to even if Paper exists in small amounts in D5.

4 Likes
6

Nothing in my evidence contradicts this. So it’d be weird and sad if anyone read the OP and concluded

But I guess we live in a weird, sad world.

But it isn’t random. It’s someone whose rank or rating is as close to yours as feasible (with limited time spent in queue as a secondary objective).

No one at Blizzard is saying that matchmaking is random. What they’re saying is that it’s NON-random with specific NON-random criteria. What the conspiracy theorists believe is that DIFFERENT criteria would be necessary to produce observable results, like the Counter Deck Wall. These other criteria are not necessary.

1 Like
7

I mean I for one think it’s highly likely Blizzard screwed up their RNG or biased it somehow and that trickles down to pretty much any application of randomness in each of their games (assuming they use the same technique for all games.) I know this technically doesn’t count as rigging the game, due to lack of intent, but I would never expect the team at Blizzard to be competent enough to be able to rig anything anyways lol.

Here is why:

  • for a game of Hearthstone scale, you probably don’t want to rely on clients performing random generation as this is prone to hacking/cheating and issues with vendor implementation of randomness

  • this means in their implementation of the game, the correctness will greatly be determined how careful they are. For example, it’s very likely they abstract the random generator as part of The concept of a self contained instance of a Hearthstone game

  • this is already problematic as then every game requires it’s own seed value that will allow concurrent random generators to work without biasing each other. I’ve seen many software implementations fall victim to this and ship products with biased randomness by not allowing distinct seeds per concurrently running process (game in this case.)

  • this type of bias is very difficult to detect by simple statistical analysis as the bias will be only correlated in time, and any aggregate statistic will almost certainly cancel out this bias

  • everything I’ve experienced in Hearthstone and possibility other Blizzard games seems to suggest to me they have this bias: unlikely events are likely to repeat with multiple rolls if those rolls happen within a given time window, for example you could go one hundred games as thief rogue and never discover rune of the arch mage, but then in one game you discover 3 (common observation in Hearthstone games in my experience.)

Anyhow, generating randomness accurately is actually an open research topic (ie not as simple as you may think.) There is a reason why there exist services that guarantee randomness by providing it via sampling guaranteed random physical processes as opposed to relying on a pseudo random number generator/algorithm to do so. If you don’t believe me look it up. See random.org for ex.

2 Likes
8

I hear what you’re saying, but you’re only going to be confusing them more like what was stated above. They will see that and say “see! It’s not random!” But not understand the problem with the conclusion because they don’t agree that you’re matched with the closest person next to them.

They don’t believe they are being matched based on mmr only and based on who is next to them, they believe they are being matched, purposefully, to create a loss on purpose.

Thus, you’re just arguing to show how mmr works to place you and how that placement can cause your next matchup(s) to be hard, but they aren’t buying the mmr only thing.

They believe if they are mmr 500 and the closest people are MMR 499 and 501, and they are on a win streak of X, it won’t choose person 499 or 501, it will choose person xxx where person xxx number is irrelevant but instead it’s that deck code that is chosen to play against because the System wants you to lose. So if deck code X is your counter, it finds player xxx in your pool with that deck code X, which might be player mmr 671, and matches you to them so you can likely lose.

9

There is no evidence that matchmaking is based on RNG. None.

Here are a list of things Blizzard says matchmaking is based on:

  • rank or rating (depending on winstreak or Legend rank) or win-loss record in Arena/Duels
  • latency (“ping” between players)
  • time spent in queue (as time in queue increases, a larger rank/rating gap is considered acceptable)

It is pseudorandom to a degree which players happen to be queuing up at about the same time you are. But there’s absolutely no RNG involved in that kind of randomness.

The rest of your post does not concern matchmaking and is thus outside the purview of this thread.

1 Like
10

Good point.

I’m not entirely familiar with how they match people and I just assumed a random roll was involved somewhere, but need to check my assumptions.

11

But The System does want you to lose. I mean, as you yourself win more, it matches you against opponents who have won more. That’s what matchmaking by rating is, on a fundamental level, and one obvious consequence of that design is that, the more you win, the more difficult it becomes to continue winning.

I don’t think we can do a good job trying to convince people of the truth if we stick to a false narrative of randomness. You’re acting as if the actual matchmaking algorithm would “flip a coin” using RNG to determine whether you’d go against 499 or 501 in your example. This is simply an inaccurate depiction. It will most likely match you with whichever of those two has been in queue for a longer time, even if it’s mere milliseconds of difference. There’s no roll to be had, except in cases of literal simultaneity — and even then, only if latency is also tied.

Like I told yellow, there’s no evidence that RNG is involved in matchmaking. If Blizzard uses a system where MMR ties are impossible (as implied by Legend numbers being unique), then there’s no need for RNG in the system at all.

The truth is a NON-random matchmaking algorithm. The falsehood that we’re combating is the idea that this algorithm considers criteria that it doesn’t. Randomness doesn’t even have a side in this fight.

12

WE understand what that context means, but they don’t. We understand that MMR “wants you to lose to match you in the right bracket” means matching you with equivalent skill based on the MMR you are in, but they think “wants you to lose” means something completely differerent - it wants to match you based on deck to stop you from winning and advancing… until you do something like pay money so it can “stop wanting you from losing”.

Again… they aren’t claiming MMR is rigged via a rating system… they are claiming matchups are rigged based on deck types, usually based on the amount of money you’ve invested in the game and if you’re winning when you shouldn’t be.

What you’re trying to argue for is already proven in high legend ranks. A rank 6 legend player isn’t going to get matched up against a bronze player. It will get matched up against the next closest ranked player that is queueing at the same time. WE know that.

What the rigged crowd argues is that the rank 6 player won’t get the next closest ranked player queuing at the same time… they are arguing that if the rank 6 player is on a win streak, it will skip looking for the next closest ranked player and instead look for the next closest ranked player with a deck that counters their deck.

if the rank 6 player is on a win streak, it will skip looking for the next closest ranked player and instead look for the next closest ranked player with a deck that counters their deck <----- your data has to show that this isn’t happening, otherwise, you’re doing nothing to change their minds. All this data is a good effort to give them a perspective of the MMR system only if the MMR system works the way we think it does. They don’t believe it does.

1 Like
13

Yes. We do.

That is pretty much what my data does. I mean, you can’t show evidence that proves something isn’t happening — like a photo of no Bigfoot — so the next best thing is to show how things that are happening can happen without the mystical cause. Like explaining how the rain cycle actually works, to someone who believes in a rain god.

Counter Deck Walls are real, and the more polarizing and paper rock scissors the meta becomes, the more real they are. I’m showing how matchmaking by skill rating creates them. I think if anything the “game isn’t rigged” crowd has mostly denied the existence of such things rather than bothering to explain how they come into being.

14

The thing is that they don’t only rig matchmaking to drop players winrate.

They also rig matchmaking to decrease the winrate of certain decks and promote another. you play a deck that have weakness against blood dk such as murloc shaman you will most likely to play against blood dk so blood dk winrate increase.

Also the game is not a rock paper scissor game in rock paper scissor games how you play it doesn’t matter at all your win or lose totally up to deck you choose.

in hearthstone even if the game match you against a deck that can counter you, you can still win the game because your opponent will make mistakes. so your win is not certain.

1 Like
15

I’m not paying for the higher ranks data like I used to, but let’s go take a look at Bronze through Gold data in HSReplay really quick.

Murloc Shaman
Games vs Blood DK: 3900
Total Murloc Shaman games: 37952 41,000
% vs Blood DK: 10.27 9.51

All classes
Games vs Blood DK: 557,770 650,000†
Total games recorded: 2,830,686
(x2 'cause 2 players per game = 5,661,372)
% vs Blood DK: 9.85 11.48

Hmm. Seems like I’ve got a few thousand games that say you’re factually, objectively incorrect. Murloc Shaman doesn’t play against Blood DK significantly more than usual. It’s about 10% like everyone else.

† Edit: Forgot to count mirror matches. Fixed.

16

I am giving an example not giving a particular case. in order to collect accurate data one needs to collect data between certain periods of time not all will do.

17

Scrotie probably forgot that rigging based on deck type only shows up in anecdotal evidence. Like ghosts or psychic powers.

1 Like
18

For other games were the difference in skill is visible this is already dificult to make people aware of.

For hearthstone your mission is just near impossible.

That because the easiest way in most games is just to put those people against people of higher ELO(friendly matches)and let they loss to their hearts content.

If that happens here people will use their 1 in 647477 matches against you they did win to talk about how everything is RNG.
Basically being a card game not helps.

19

Skynet has obviously taken over the world. Get those pool chalkboard ladders up, invite your ranked opponents on video and watch each other shuffle cards.

Sorted. Cavemen style.

20

First off, thank you for the post; it helps me at least, one who does believe the MMR rigs me to make it impossible after Diamond 5 to hit legend after playing the game all this time.

(I never have hit legend)

I somewhat miss the old system of matchmaking if it even was legit? Where it would fit you against someone in the same rank as you (Bronze = Bronze, Silver = Silver, etc). Maybe that never attributed to it positively or negatively (The mentioned “Decks you beat, then decks you equal at, then decks that counter you, repeat”).

I still feel a player should face any other player. Does it counter them or not, winstreak or not, because that’s how someone plays a card game.

And I’ll blatantly admit that I have little concern for the newer player experience and I do actively tell outside interested parties to not play Hearthstone.