The Magic Tournament Rules specifically state 'For the first game of a match, the winner of a random method (such as a die roll or coin toss) chooses either to play first or play second'. My method of choice has often been Rock-Paper-Scissors, but it never occurred to me until recently that this might not be considered a valid method under the rules. Obviously this question is more hypothetical than not; if my opponent disagrees with my method of randomization then a different option, or a judge, would be called for. But at least in theory, spectators or judges watching our match could complain, so the question holds: is RPS (likely to be) considered a 'random method' for determining play/draw under the tournament rules?


My understanding is that it's really up to the judge.

The tournament rules just say this:

For the first game of a match, the winner of a random method (such as a die roll or coin toss) chooses either to play first or to play second.

They don't really specify a more detailed criterion for "random." (For example, cutting to a random card of each other's decks and then comparing their mana costs is a method players use some times. It's random, but it's not actually fair: the CMC of various cards in your deck skews the results.)

My first assumption was no rock-paper-scissors, since RPS is a game that permits some level of psychological skill. However, here's a quote from an old tournament report:

The both of them shuffled up and scanned about for a way to figure out who went first; turns out neither of them had dice. "That's the problem with being up here," deadpanned Jensen. "You play against a random? They always have a die."

It was suggested that perhaps the two of them play rock, paper, scissors to decide who would go first, but Jensen said that it wasn't random enough. "Rock, paper, scissors is strategy," he said earnestly. "I think I could psych someone out at rock, paper, scissors." Eventually, someone donated not only a twenty, but a twelve-sider and a ten-sider, and they rolled all three and added them up to see who went first. Jensen snagged it with a 22.

Note that it was the opponent who decided RPS was insufficient, not a judge. This implies that RPS might've been acceptable if both players were fine with it.

In practice, I would expect low-level events not to care about the exact quality of the method you use as long as it's mutually agreeable to both parties (which is all the comp rules really call for), unless they've had to set up a specific system due to previous issues with cheating.

  • 2
    If you cut to highest CMC of the same deck, I think it would be fair. But then you'd gain some knowledge of that person's deck. I think the odds of not having a coin or die available are fairly slim… :) – ghoppe Feb 26 '12 at 5:28
  • 2
    @ghoppe Cutting even like that isn't really fair either. If I cut to some card I only have one of, then I now know its position in my deck and that could affect my mulligan decision. Or, if we cut before we shuffle, the cut may not be truly random if I have my deck stacked. We could shuffle, then cut, then shuffle again, but that's a lot of work. A better method: Get a random card (could be from one of the decks, I suppose), and then ask your opponent if the Collector Number printed at the bottom of the card is odd or even. If they get it right, they can go first. – Cory G. Feb 27 '12 at 16:32
  • 1
    @Pwngulator An interesting idea, but are the collector numbers randomly distributed? That is, are there the same number of odd collector number common cards as even? What pool is the card being drawn from? Popular sideboard cards may have 60% odd collector numbers, for examle… I realize of course, this is pedantic to the extreme and the person choosing odd or even is unlikely to make such considerations. – ghoppe Feb 27 '12 at 17:22
  • 1
    @ghoppe Don't underestimate how far people will go for a small edge. A tournament regular was recently banned (in part) for drawing 8 cards in his initial hand when he was on the draw and then skipping his first-turn draw to conceal it. All this cheat affords you is one card of additional information for making mulligan decisions, yet he still felt it was worth the risk. – Alex P Feb 27 '12 at 17:53
  • Oh, I definitely don't expect low-level events to care about the method used - I was more curious about the theoretical implications here. :-) Still, this earns the check mark... – Steven Stadnicki Feb 27 '12 at 18:15

A method isn't random if it is asymmetrical in any way or its outcome comes from a mixture of random and non-random factors.


  • Cutting decks for highest mana cost is asymmetrical if you are not cutting from the same deck.
  • Cutting decks for highest mana cost is still inadvisable if you are cutting the same decks you are playing with, as it could be a breach of other rules. It might be acceptable if you use an unrelated third deck, provided that it is in a randomized state and its cards are not marked.
  • Cutting for even/odd collector's number is not random, since the distribution of evens and odds may be known to one of the players, or the distribution may not be symmetrical. Odd numbers also occur slightly more often than even numbers (sets with an odd number of cards contain 1 more odd card than even). There is a somewhat fair way to approach it (i.e. use your own deck and ask the opponent to pick odd or even), but since you might have the opportunity to rig the deck to contain only odd or even collector numbers, this could become a psychological guessing game similar to rock-paper-scissors. I.e. most novice RPS players lead off with rock, so you might have an asymmetrical skill advantage if you throw paper. Most analytical even-odd number guessers should guess odd, since it occurs slightly more frequently than even, so you could rig a deck to have all even collector numbers and essentially guarantee that you win the option.
  • You roll a d12 and the opponent rolls a d10. This is obviously asymmetrical, but it helps highlight less obvious examples that give either player a statistical edge by a similar amount.
  • Each player rolls a different d20. Your d20 might be fair, but how do you know that your opponent's d20 isn't imbalanced?
  • Rolling a spindown d20. It may be outside the realm of real human control, but you could attempt to roll it similar to spinning a top so that it settles on a high number in the 14-20 ballpark.
  • Playing any game of skill. Playing a game of skill is not a random method. For similar reasons, since RPS has some skill, it is not a random method. RPS strategy integrates elements of psychologically reading your opponent to deduce which symbols they favor, using predefined strategies so that you aren't shooting reactively (and predictably), and sometimes reacting to predictable novice behavioral patterns.
  • Playing high number on a random number generator app could be adequate, provided that it isn't rigged. You could neutralize the chance of rigging it by letting the opponent decide whether the first number is yours or theirs. In theory, this choice could be very difficult to psychologically game, but possible.

The seriousness of randomness enforcement will be greater at higher REL events. However, even at REL casual, most opponents would require you to abide by common sense randomness practices (i.e. no spindown dice, no cutting decks).

  • 1
    This is incorrect, random doesn't have to be fair and in truth random is rarely fair. For your dice examples even if two people roll the exact same die/dice how they perform the roll can have a massive impact on the results and reduce the randomness of the outcome. – Joe W May 11 '18 at 16:34
  • Another way to think of it is we could use a random method to determine who starts 100 matches and I could win the right to start all (or most) of them which wouldn't be fair. Fair would be me starting 50 and you starting 50 but that wouldn't be random. – Joe W May 11 '18 at 16:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.