RNG and the evil shuffler

One of the frequently asked questions that’s being brought up pertains to the randomness aspect. A good real life example that could illustrate the problem was the apple and spotify shuffle debate. I’ll quote from the linked article:

“When Spotify launched, the company explained in a blog post, it made its random playlists using the “Fisher-Yates shuffle”. That is an efficient algorithm that takes only three lines of code to make “the optimal amount of operations and optimal amount of randomness”, creating random playlists without using much computing power.

But users complained that the playlist wasn’t genuinely random. Artists or genres tended to appear next to each other, giving the sense of the playlist being unfair – even though it is likely that different artists will bunch up as that they will be evenly distributed throughout.”

It’s a well known problem, but how does that concern MTG and MTGA in particular?

A deck is being shuffled by the random algorithm. You get no lands for 5 turns - you’re mana screwed. Or you draw only lands 12 times in a row - you’re mana flooded. This is variance and the nature of the game.

Traditionally to deal with this, MTG had employed first a limit of the same cards (maximum 4 in a constructed deck) and a bo3 structure that allows variance to even out somewhat. This is one of the reasons why bo3 is played in most major professional MTG tournaments.

In bo1 there are a number of ways to help you build your deck to lower the effect of variance on your gameplay:

• For Aggro decks play low-cost spells that allow you to win on 2-3 mana
• For midrange decks have some ways of generating card advantage (draw or filter cards)
• For control decks generating card advantage and filtering is of paramount importance

There are a number of resources that explain the current random algorithm and why it won’t change.

Third-party verification of the shuffler working as intended: https://blog.mtgatracker.com/debunking-the-evil-shuffler

Game Director Chris Clay confirming the shuffler working as intended: https://mtgarena.community.gl/forums/threads/19592/comments/85747

Chris Clay expanding on the shuffler working as intended, and why they won't change it: https://mtgarena.community.gl/forums/threads/20344

Comments from Chris Clay regarding the shuffler before RNA launches

https://www.reddit.com/r/MagicArena/comments/agjqkg/chris_clay_about_mtga_shuffler/

Chris Clay and Megan O'Malley discussing the shuffler (working as intended) on a recent stream:

https://www.twitch.tv/videos/335929967?t=01h02m58s

Another important difference between shuffling in paper MTG and MTGA is that you can’t mana weave. Mana weaving is when you put a land card after two cards in your deck before you shuffle. Such thing might be ok for games between friends but it’s illegal in a tournament setting exactly because it reduces variance in an unfair manner.

Furthermore, according to the official rules a deck is considered shuffled if you have done such an action thirty times. Every time. Very few people do that in a real life setting, but it becomes easy in MTGA for the program to do it for you.

As such, arguments that you get less mana flooded/screwed in paper is usually moot.

If you still find yourself willing to debate about the shuffler, it’s best if you have a good sample size of games (1000 minimum) and read through the material above.

One other topic that seems to still be debated hotly is the impact of the algorithm that chooses the better starting hand in best of one matches. The following should be noted:

• This applies only to best of one game modes
• If you mulligan even once it stops applying
• We don’t know exactly how it leans toward the better hand

There had been theories how it’s best to use only 22 or 26 lands in MTGA since closed beta.

These theories are oversimplified and have been debunked by the developers as wrong. Based on the experience of thousands of games with 16 to 27 lands the usual deckbuilding rules to land count apply.

• 20 to 22 lands for aggro
• 22 to 24 lands for aggro-midrange/midrange/tempo/combo
• 24 to 26 lands for control

You might always go up or down one land from your current list. Just be mindful of variance and keep track of your games. We'll be discussing it in greater detail in the deckbuilding section of the guide that is still under construction.

On the topic of tracking your progress, you should definitely check out the available tools and trackers.