I was referring to computation required to rig a subset of games in a way that “hides” the action.
How can there be wild variations? I can virtually guarantee there has never been the same sequence of 30 cards drawn by any set of hearthstone players ever.
Doesn’t the internet have a name for this now? Seal something? I’d look it up but the flying spaghetti monster told me I’m destined to stub my toe tomorrow, so it looks like I’m eating steak tonight.
They don’t have to because third parties inspecting have no information at all about the relation of the company with the players; they can not label a group “these players pay the company” for example because they simply don’t have the data; they only have random IDs of players with no much identification other than that.
Put it this way (which answers your other reply too); yes of course the variations of draws are massive between games but they are also massively random and sometimes that randomness is literally unfair to the player; how could you possibly know if the unfairness was random or intentional if you can’t group the players?
Say you grouped all the players who had unfair draws in their games; it says absolutely nothing unless you grouped them also by a relation with the company; that’s because there are ALWAYS people who RANDOMLY get unfair draws ANYWAY.
you moved them, though.
But we aren’t talking about that scenario.
If the draws are not random there will likely be collisions - the same draw occuring more than once.
Deck trackers have access to this information. Sounds like obtainable evidence to me 
1 Like
Nah. Even if you want to rig it you can add a random element in it to be both random and favorable to a player. And as I said even if you found individually favored players it means NOTHING; that’s because there are randomly favored players ANYWAY; you have to know also what kind of relationship they have with the company to put them in groups otherwise they just remain “randomly lucky” or “randomly unlucky” (which is natural to have both).
You need to think outside the box carnivore. If you are manipulating draws then you will produce statistically unlikely outcomes that can be measured. I’ve given one example with 2 minutes of thought. If a trained analyst took their time I’m sure they’d find many more.
1 Like
Sounds too me like you have a rudimentary understanding of Hard Determinism. Saying it’s self evidently untrue just tells me you haven’t really researched much or heard the reasoning.
That’s fine. To each his own.
Your example is “fixed” for the manipulator by just adding a rand() function in there (which is a 2 minute job). Even if we assume INDIVIDUAL variability could be shown to be non-random: the game is so extremely random BY DEFAULT that you’ll have to prove first that it’s even DETECTABLE (mathematically).
The easiest way to prove they rig it (if they do) is to just group players who you believe would be favored because then they’ll just have more favorable draws or opponents; instead now you’ll only find some dude with a “weirdly more win rate” together with some others dudes; hard to know if they were lucky.
But how will you group them?
And adding a random function obviously destroys the otherwise defined draw.
Can I suggest instead of trying to think why you can’t do something, try and think of ways you can.
1 Like
The main theory of the “rigged!” people is that the favored players pay Blizzard money at the shop, hence the easiest way to prove the theory wrong (or right) would be to just have data on all players related to the shop; the third parties have none of that information; they only know their own accounts (very few to matter) and they can’t trust the word of random players without direct access to the accounts (which only Blizzard has).
For the record: I believe it’s unlikely they rig it; but I don’t like it when people claim we are 100% sure; we will never be sure unless it was open sourced and we run the servers ourselves on top.
I agree that would be a great method, but unless you construct an experiment with 2 controlled groups, you don’t have that data on hand.
But there are other ways to detect anomalies, one of which I highlighted and you’ve dismissed without really thinking about it properly. The fundamental truth here is that if you directly manipulate outcomes to the point where they affect the meta, you will produce measurable statistical anomalies. While it would be nice to attribute them to a group, the fact that they exist at all is significant enough to know that draws are manipulated, or at least “not random”.
1 Like
Do you have a reference to the methodology that would do that in THIS CONTEXT? This context has extremely high randomness BY DEFAULT; people get very lucky all the time; they may get lucky for multiple games in a row because that’s how randomness works.
My Statistics is rusty but I’m pretty sure there is a THRESHOLD of randomness that would make this an impossible task (mathematically); i.e. if it’s already very random by default: after a point you might not even be able to detect anomalies; hence target groups help.
A simple example, the chance of a given draw sequence is 1 in n factorial, where n is the # of cards drawn. If it’s not a highlander deck it’s going to be divided by 2 (2 factorial technically, which is 2).
So if you look at as many games as you can stomach, isolate games where the sequence was the same up to n, and calculate the number of occurrences vs the likelihood. If draws are determined by a non random method then there will be anomalies.
Obviously you need to look at a lot of games.
You missed that if they wanted to rig: they may not rig that way; they may rig by giving unfavorable opponents; you may return and say “1 over factorial of the classes or netdecks” in that case. But there’s another thing they can do you can’t control at all,
rig what MMR you get below Legend (it’s hidden entirely to third parties); in fact the MMR of Legend is also hidden as I understand it; but even if we assumed Legend rank is the same as MMR: the pre-Legend players are the most customers anyway.
1 Like
For the unfavored opponent (assume you mean deck) hypothesis, you can compare matchups by deck compared to the meta average. You’ll have anomalies.
People have done that work previously and the outcome was that rigging based on deck type could not be taking place in more than 1-2% of games.
Note, respectfully, I’m not going to provide an endless list of methods for every scenario . Its reasonable for me to expect you to think of your own methods.
1 Like
And so the posts move.
Blizzard can barely balance the game, but for sure they are competent enough to create a algorithm to analyze deck composition and find an appropriate player with the appropriate deck amongst illion of other players, each hitting play in different times, all in the span of the 10 or so seconds you get matched with someone.
Or, and bear with me here, they can just match you with one of the many players in your overall bracket and not have to worry about the technichal dificuties in implementing such.
HSreplay tracks this. Deck matchup spreads mostly mimic the meta reports, even across websites. I say mostly because the meta in most data aggregrators is defined by tiers of play, and deck matchups in HSreplay is logged across all the all ladder at once, so there are deviations.
Perhaps we covered that case (given enough games (we may not have enough games to cover the extreme default randomness but let’s assume we have for now)),
but you didn’t answer what we’d do to detect rigging that happens on the MMR level which is completely hidden to most if not all players (to most for sure).
It’s so CRUCIALLY easier to group the people by “if they paid at the shop” that a better study is to find a few dozens of players to show their accounts.
For someone so confident to mock others you missed that isn’t the only way to do it.
They can rig it by MMR, and people are oblivious to the MMRs especially pre-Legend.