RNGeesus!

This is going to be a controversial statement, but here it is - 

There is nothing wrong with the Arena RNG Shuffler. 

Well, that’s not quite a fair statement. More precisely, there is currently no solid evidence to justify believing that there is a problem. Peoples’ anecdotal experiences aren’t sufficient. Nevertheless, complaints and controversy over how Arena deals out cards abound. From unnatural levels of Mana Flood and Mana Screw, to rigged draws and rigged games, to the program turning on you when you’ve won “too much”. People are claiming everything from the Shuffler being broken to outright, intentional tampering of the program by Wizards. 

I’m not a numbers guy. There’s plenty of people out there who can go deep on the technical end and explain how these systems work, and how the raw data has shown that it’s functioning well within the margins of where it should be. One of them can be found here: https://blog.mtgatracker.com/debunking-the-evil-shuffler 

I want to talk in this post about what the community is actually experiencing - because it’s absolutely true that there’s an experience behind the complaints. It’s important to understand how we process and interpret these events. 

There’s three major components in play here. First, digital hand generation feels different than playing IRL. Second, many people don’t have a good grasp of what “true” randomness is. Third, We have cognitive biases affecting our perceptions of just how often certain events occur. 

I'm using quotes around “true” randomness here because I understand that even a computer isn't “truly” random. They use seed numbers and complex equations to produce practically random results. But that's another topic entirely. For the sake of what Arena's doing, and for this post, it's random. 

I’ve been playing digital Magic for many years on MTGO (Magic: the Gathering Online). When you transition from Paper Magic to Digital Magic, one of the things you’re bound to notice is that the drawing an opening hand intuitively “feels” different, although it can be hard to quantify exactly how (more on this below). We move through much of the world on feel and intuition, and even more so with something as complex as Magic. If you ask Pros about plays they made, they might not always be able to explain why they made the play that they did, other than it “felt” correct. It's just too much to process all at a higher cognitive level. 

Part of why it feels different I’ll address below, but part of it is certainly due to the absence of the physical act of shuffling. It’s a repetitive motion which we associate with the game, and with drawing an opening hand. So when we start the game without that physical sensation, or the “feeling” of having shuffled it's noticed. There's also something to be said for the idea, justified or not, that you have some control over what kind of opening hand you're getting. Like “If I only shuffle well enough, I'll have a decent hand” and “I must have gotten screwed because I didn't shuffle well enough.” As a rule, we don't like ceding control. And handing control over to a cold, unthinking, uncaring machine? Well, there’s entire Sci-Fi sagas written about that. 

Random numbers and distributions are tricky things. Look at the two images below. 

Which one is more random? Most people will intuitively think that it's the one on the right where fewer dots are clumped together. However, “true” randomness doesn't work this way. With actual randomness, things like clumping and strands are to be expected. So when we're drawing cards, we should expect there to be some number of occurrences of clumping of Lands and clumping of Spells, and even clumping of events of Screw and Flood. What we're often thinking of, when we’re randomizing a deck, is actually “even distribution”, and not actual randomization. 

As a side note, what Wizards’ considers “sufficiently randomized” in regards to tournament policy, is only that neither player can have an idea of where any give card, or set of cards is located in the deck. It's got nothing to do with actual randomization or distribution of the cards. 

Manual shuffling of cards does not really randomize them in a machine sense of the concept. We inevitably start with cards which are arranged into piles of lands and spells, and probably higher-costed, or highly-situational cards in our hands. When we do shuffle, we invariably will “clump” cards as we're doing it. 

You might think this is an argument for “pile shuffling” or “mana weaving” (ordering the deck with an even distribution of  lands and spells) before shuffling, but it's not. Pile shuffling isn't actual randomization, since you're controlling where each card is put, and so violates Wizards’ tournament policy. And if you're finding that your draws are better after mana weaving, then you're not shuffling sufficiently after having done so, because thorough shuffling should produce random distribution regardless of the starting position of any given card.Because of the inherent differences between random machine distribution and physically shuffling cards, we're invariably going to see a difference in the hands which are drawn. 

“But Dan,“ I hear you saying, “I got mana screwed like 7 games in a row, right after I went on a winning streak. It happens every time! And every time I mulligan because I'm land heavy, I always end up with another land heavy hand. The system is rigged!” Here we come to what's probably going to be the most controversial part of this whole post. Understanding and embracing our own fallibility.

Our own personal, anecdotal experiences cannot be used as data for analyzing the functionality of Arena's card randomization process. There are a variety of cognitive processing quirks and biases which impact our perception of the frequency of these events. 

First among these is our strong talent for pattern recognition. When there is an event, or object that is significant in some way, we're more likely to be aware of future instances of events or objects. This is often experienced when buying a car. We will suddenly start noticing examples of that model of car all over the place. It's not because there's suddenly more of the, only that our awareness of the existing examples has changed. There is really only so much information we can process, so there's much that's pushed off as background “noise” in our awareness centers. 

Moreover, when an event has a strong emotional component to it, like losing a game to poor draws, especially if it's one you feel you “should” have won, then we are much more aware of the instances when the screw or flood happens again. Those stay with us, while all the many games where we didn't get flooded or screwed just fall away into background noise. They aren't notable, so they aren't noted and our memories fill up with examples of when things went poorly for us. 

Once we have the idea that the shuffler might be broken or rigged, then we can fall prey to Confirmation Bias. This is the phenomenon which is the tendency to search for or interpret information in a way that confirms one's preconceptions, and discredit information that does not support their views. Now the “hits” register even more highly, and the “misses” (all the games where things went well) are not at all entered into the calculation. It's an unfortunate reinforcing cycle, and one that we're all susceptible to. 

So, I’m sorry, but no one’s anecdotal account of how they’ve been abused by the shuffler has any analytic, or statistical value. And the plural of “anecdote” isn’t “data”. It doesn’t matter how many people chime in with similar stories. At most, that might give us reason to investigate, but it’s not valid data by which to form a conclusion. 

There’s one final point I want to touch on. 

The almost conspiracy theory level belief that Wizards is intentionally screwing people over with the shuffler. Like they’ve rigged the program to intentionally hammer you some number of times. And when I’ve asked why, the only response I get is some version of “To squeeze more money out of the players.” 

HOW? Seriously, how? How does ensuring that players are punished for doing well make your players want to do anything but quit the game? That makes absolutely no sense at all. Wizards has spent the past 25 years operating on the proven premise that “If you print it, they will buy it”. Proven not only in Paper, but on MTGO. And it’s not insignificant that the Pros grind and test on MTGO. It’s clear that Wizards can make all the money they want without messing with the shuffler. 

Thanks for taking the time to read this. Whether you agree or disagree, I'd love to hear from you, and you can also find me on the MTG Arena Players Facebook group, where I share moderator duties.