Determining Randomness

31 posts / 0 new
ChrissChrossed92
Joined Dec 2011
486 Posts
Hello, I have a question about "legitimate" ways to determine how to do things randomly.  Here my scenario:

My opponent has 40 something saprolings out, a few creatures making the saprolings each turn, a few enchantments, and only like 5-6 lands.  I then played a tyrant of discord and targeted him.  He proceded to take out a six-sided dice and said "Even, I'll sac land, odd, nonland"  This is where I protested.

His idea of "random" was highly in his favor, since now there's a 50% chance of him saccing a land to negate the "repeat" effect of the tyrant.  I think this is unfair because if a permanent is chosen from all his permanents, then his 5-6 lands is out of roughly 50 something permanents total, so shouldn't there be only ~10% chance for him to sac a land?  His argument was that it doesn't matter if the odds are in his favor or mine, just that he finds a "random" way to trigger the tyrant's effect.

Is there anything in the rules that proves him wrong?  Or am I looking for rules that don't exist?
ikegami
Joined Nov 2007
2268 Posts
This is where I protested.

And rightly so.

just that he finds a "random" way to trigger the tyrant's effect.

No, he needs a random way to choose a permanent.

Is there anything in the rules that proves him wrong?

You are asked to choose a permanent. "Land" and "non-land" are not valid choices as they are not permanents.

110.1. A permanent is a card or token on the battlefield. [...]

2goth4U
Joined Oct 2007
10306 Posts
random here means each permanent has an equal chance of being selected

his method would not achieve that

if there were six permanents of type land and six permanents of other types (and the sets were mutually exclusive), then his method would be fair for the first determination because there would be an equal chance of choosing one from either set, but after the first sac those odds are no longer fair

40 saprolings + a few other creatures (let's go with 4) + a few enchantments (again 4) + 6 lands = 54 permanents

6/54 <> 1/2

DCI Certified Judge & Goth/Industrial/EBM/Indie/Alternative/80's-Wave DJ
DJ Vortex

DCI Certified Judge since July 13, 2013  - If you have any concerns with my conduct as a judge, feel free to submit feedback here.
DCI #5209514320

My Wife's Makeup Artist Page <-- cool stuff - check it out

Bowshewicz
Joined Apr 2012
2078 Posts
There's nothing in the rules that we can quote to prove him wrong. Nor does Tyrant of Discord specify that the randomization method must provide an equal chance for any given permanent to be selected. However, even without rules to back it up, I'm certain that this is the case. Your friend may be satisfied to know that he has found a small loophole in the rules, but you should stress to him that all random effects assume an equal chance.
Chaikov
Joined Jun 2006
8455 Posts
Is there anything in the rules that proves him wrong?  Or am I looking for rules that don't exist?

Randomness is not specifically described in the CR, so the English definition must be used.
One thing is sure: each qualified object must have an equal chance of being chosen. Otherwise, throw a 100-sided dice and say "sacrifice land on a throw of 1-99; sacrifice non-land on a throw of 100"

Perhaps your opponent is unaware of the fact that his method does not provide equal chances; probability is a tough subject to understand (which explains why lotteries are so popular).

Insanity: doing the same thing over and over again and expecting different results.                        -Albert Einstein

So, you keep on voting, and you keep expecting different results from those elections?

How long have you tried this, over and over?

Bowshewicz
Joined Apr 2012
2078 Posts
Found this. The game probably cares more about the second definition, while your friend was attempting to apply the first.

ran·dom

[ran-duh m]

1. proceeding, made, or occurring without definite aim, reason, or pattern: the random selection of numbers.

2. Statistics. of or characterizing a process of selection in which each item of a set has an equal probability of being chosen.

ikegami
Joined Nov 2007
2268 Posts
Found this. The game probably cares more about the second definition, while your friend was attempting to apply the first.

The problem is not with his understanding of random. It matches the second definition exactly. (There is an equal probability of choosing "land" and "non-land".) The problem is that he isn't asked to choose between "land" and "non-land".

JaxsonBateman
Joined Sep 2009
4230 Posts
I'm surprised there's no mention of what random means in Magic in any part of the rules.

The closest rules support I can come up with is from the Flipping a Coin section:

705.3. A coin used in a flip must be a two-sided object with easily distinguished sides and equal likelihood that either side lands face up. If the coin that’s being flipped doesn’t have an obvious “heads” or “tails,” designate one side to be “heads,” and the other side to be “tails.” Other methods of randomization may be substituted for flipping a coin as long as there are two possible outcomes of equal likelihood and all players agree to the substitution. For example, the player may roll an even-sided die and call “odds” or “evens,” or roll an even-sided die and designate that “odds” means “heads” and “evens” means “tails.”

Of course, the fact that it refers to only coin flips means that it's a soft rule for randomization as a whole at best, but you could make the same argument without this rule that you could use any method for a coin-flip randomization, even if the chances weren't equivalent. But the rule above says that you can't, and each outcome must be equal - it follows from that that when dealing with randomization of 3 or more outcomes, they must still be equal.
I'm all about super-control in MTG. If you're able to stop my shenanigans, then there aren't enough shenanigans. Lv 1 Judge Current Decklists Sweeping Beauty (Casual) A Vision of Clones (Casual) Coming soon... more decks! :-O
Esterdi
Joined Nov 2012
1517 Posts
I find this rather shocking. They should seriously think about updating the rulebook with a definition of random, at least as clarification.
Strago_Magus
Joined Oct 2009
2025 Posts
"Life is chock-full of lies, but the biggest lie is math.  That's particularly clear in the discipline of probability, a field of study that's competely and wholly fake.  When push comes to shove--when you truly get down to the core essence fo existence--there is only one mathematical possibility:  Everything is 50-50.  Either something will happen, or something will not.

"When you flip a coin, what are the odds of it coming up heads?  50-50.  Either it will be heads, or it will not.  When you roll a six-sided die, what are the odds that you'll roll a three?  50-50.  You'll either get a three, or you won't.  That's reality.  Don't fall into the childish 'it's one-in-six' logic trap.  That is precisely what all your adolescent authority figures want you to believe.  That's how they enslave you.  That's how they stole your conviction, and that's why you will never be happy.  Either you will roll a three, or you will not; there are no other alternatives.  The future has no memory.  Certain things can be impossible, and certain things can be guaranteed--but there is no sliding scale for maybe.  Maybe something will happen, or maybe it won't.  That's all there is.  What are the chances that your sister will die from ovarian cancer next summer?  50-50 (either she'll die of ovarian cancer or she won't).  What are the chances that your sister will become America's most respected underwater welding specialist?  50-50.  It will happen, or it won't.  There are two possibilities, and both are plausible and unknown.  The odds are 2:1.  These facts are irrefutable.

"Quasi-intellectuals like to claim that math is spiratual.  They are lying.  Math is not religion.  Math is the antireligion, because it splinters the gravity of life's only imperative equation:  Either something is true, or it isn't.  Do or do not; there is no try."

Chuck Klosterman - Sex, Drugs, and Cocoa Puffs; page 148

Joined Dec 2011
2548 Posts
That quote makes me cry.

Perhaps a rule outlining randomness could be written as follows:

"If an effect requires the selection of one outcome from a set of possible outcomes at random, any method of selection may be used such that each outcome has an equal likelihood of being selected. For example, if a player is asked to discard a card at random, that player may shuffle the cards in his hand into a pile, then discard the top card."
Zoidberg
Joined Aug 2005
3042 Posts
The problem is that he isn't asked to choose between "land" and "non-land".

This.

And basically, you are all mad for discussing this. It leaves no doubt as to how things should be.

X permanents means, 1 chance out of X for a specific permanent to be destroyed. ANY permanent, not separating land and non land. It is not what is on the card.

The Basic rulebook, read it! A lot of basic questions are answered there!

How to autocard :
Type [c]Black Lotus[/c] to get Black Lotus.
Type [c=Black Lotus]The Overpowered One[/c] to get The Overpowered One.

Joined Feb 2007
12269 Posts
If someone just says "at random", it's safe to assume that they mean a uniform distribution. It would take the worst kind of rules lawyer to argue that the absence of an explicit rule implies that any other distribution is equally valid.
Joined Dec 2011
2548 Posts
And yet, as pointed out in this very thread, the rules go out of their way to explain that flipping a coin means that there must be two equally likely outcomes. If the rules see fit to specify that much, I don't see why they shouldn't also specify what "at random" means in a general sense.
Strago_Magus
Joined Oct 2009
2025 Posts
The issue isn't about what "random" means, it's about how the card works.  Rolling a die and assigning odds to lands and evens to nonlands is in and of itself random, but that's not what the card is asking the player to do.  The person who played the card tried to resolve it in a very poor way.  It has nothing to do with the definition of "random."

Chaikov
Joined Jun 2006
8455 Posts
And basically, you are all mad for discussing this. It leaves no doubt as to how things should be.

This can be said about MANY discussions, here & elsewhere.
But there are those who just won't recognize the truth even when it's in plain sight. Perhaps they don't like what they see?
"Ajustable" game rules has been a very effective way of winning since the dawn of time...
There's also some weird conception of pride which forbids admitting being wrong.

Insanity: doing the same thing over and over again and expecting different results.                        -Albert Einstein

So, you keep on voting, and you keep expecting different results from those elections?

How long have you tried this, over and over?

Strago_Magus
Joined Oct 2009
2025 Posts
Dec 26, 2012 -- 5:14PM, The Fonz wrote:
There's also some weird conception of pride which forbids admitting being wr-wr-wr-wr...

Chaikov
Joined Jun 2006
8455 Posts
Let's hope people will realize that this part is a joke:
"Life is chock-full of lies, but the biggest lie is math.  That's particularly clear in the discipline of probability, a field of study that's competely and wholly fake.  When push comes to shove--when you truly get down to the core essence fo existence--there is only one mathematical possibility:  Everything is 50-50.

...and let's hope people will realize this part is true:
Math is not religion.  Math is the antireligion, because it splinters the gravity of life's only imperative equation:  Either something is true, or it isn't.

Insanity: doing the same thing over and over again and expecting different results.                        -Albert Einstein

So, you keep on voting, and you keep expecting different results from those elections?

How long have you tried this, over and over?

ikegami
Joined Nov 2007
2268 Posts
And yet, as pointed out in this very thread, the rules go out of their way to explain that flipping a coin means that there must be two equally likely outcomes.

The rules go out of their way to explain that flipping a coin may be done without flipping a coin. This is necessary, since MTG's definition of "flipping a coin" differs from English's. (Admittedly, the example is overkill.)

Adding a rule to define random when the definition of random is clear and can be obtained from a dictionary has no merit.

2goth4U
Joined Oct 2007
10306 Posts
Wow

Chuck Klosterman sure is a hard determinist.

DCI Certified Judge & Goth/Industrial/EBM/Indie/Alternative/80's-Wave DJ
DJ Vortex

DCI Certified Judge since July 13, 2013  - If you have any concerns with my conduct as a judge, feel free to submit feedback here.
DCI #5209514320

My Wife's Makeup Artist Page <-- cool stuff - check it out

00brak
Joined Dec 2012
75 Posts
At my local FNM when we come across something like this we have all agree'd to dice system to solve it.

-Line up all cards in question of being targeted.
-figure out how many of what kind of dice would best be used to give a random number
-Assign numbers to each card starting with the lowest possible number X dice can go and going up.

Generally we use this for discard a random card and a d6 works fine. In this case it seems like it would take 3 or so d20 to get the job done. We also reroll any number that didn't have a target since that does happen.

Just our way of handling problems like that, perhaps if you continue using the tyrant you could put it to use. I'm also very shocked that there isn't at least a suggestive rule on how to handle random effects.
Joined Dec 2011
2548 Posts
In this case it seems like it would take 3 or so d20 to get the job done.

That doesn't give an even distribution either, unless I'm misunderstanding what you mean.

thatmarkguy
Joined Sep 2011
2585 Posts
3d20.  When you absolutely, positively have to choose one random target out of 8000 candidates - accept no substitutes.

Indeed, I hope people aren't doing any "add dice together" math in any of these random determinations.  What makes sense is something like "if D20a shows 1-to-10, we use the value on D20b.  If d20a shows 11-to-20, we use the value on d20b plus twenty.  That's a random number between 1 and 40.  Since we only have 37 targets, we will reroll entirely if this procedure results in 38, 39, or 40."
Glimmereyes
Joined Jul 2008
315 Posts
www.random.org/
Or if you want to stick with WotC stuff, www.wizards.com/dnd/dice/dice.htm
00brak
Joined Dec 2012
75 Posts
I done a poor job explaining myself there.

Lets say in the op case you have 54 targets. It would take 3 d20 to have all of them encluded. You would start with 3 as the lowest number and number up to 57. So you roll 3d20 add them up and destroy what ever card the number came up for, rerolling a 58-60 total.

Now I understand its far from a perfect way to randomize them. Generally when we use this method it's no more than a single d20 having to be roled. Aside from shuffling all the cards in question together and drawing, I can't think of another way to randomize it fairly. Tokens also don't do well with that method lol.
Bowshewicz
Joined Apr 2012
2078 Posts
3d20 doesn't work, because you'll never roll a 1 or 2, and you're more likely to get numbers in the area of 30 than you are to get numbers in the area of 60. If you absolutely must use multiple dice to generate a random result, I recommend thatmarkguy's suggestion, as it gives an even distribution of values in your range.

On that note, the most likely reason that I can come up with for the comp rules' omission of a definition of "random" is so that plays like this are not illegal. Most players won't have an issue agreeing on a random number generation method, and it's not the comp rules' problem to make those who can't get along.

@below: XD
cyphern
Joined Jan 2003
19350 Posts
3d20 can work if you do it in base 20 with each die representing a different digit.

For example, if you rolled I, 6, C (18, 6, 12 in more familiar base 10), you would have chosen the I6Cth permanent (the 7332th permanent).

thatmarkguy
Joined Sep 2011
2585 Posts
I done a poor job explaining myself there.

Lets say in the op case you have 54 targets. It would take 3 d20 to have all of them encluded. You would start with 3 as the lowest number and number up to 57. So you roll 3d20 add them up and destroy what ever card the number came up for, rerolling a 58-60 total.

Wow.  I really shouldn't have given you the benefit of the doubt.

Now I understand its far from a perfect way to randomize them.

Indeed.  In that it's far from uniform, far from proper, far from fair.  Far from accceptable.  Who gets to decide which permanents are assigned the least probable numbers and which ones are assigned the most probable ones?

Generally when we use this method it's no more than a single d20 having to be roled.

If it's no more than a single die, there's no issue.  Once you start adding dice together it's a significant issue (because while the result of one fair die is a uniform distribution, the sum of two or more dice is not).  Suppose you had 11 items to randomize - you'd think 2d6 is fair?  The permanent assigned '7' is six times more likley to be 'randomly chosen' than the permanent you assigned to 2.
Esterdi
Joined Nov 2012
1517 Posts
Isn't the best way to assign a random target beyond 10 to use two d10, one for digits and one for tens?
thatmarkguy
Joined Sep 2011
2585 Posts
Isn't the best way to assign a random target beyond 10 to use two d10, one for digits and one for tens?

Whether that's the 'best' way is a matter of opinion, but it is a valid and fair way (unlike 00Brak's method).
Joined Jun 2011
1004 Posts
Isn't the best way to assign a random target beyond 10 to use two d10, one for digits and one for tens?

Whether that's the 'best' way is a matter of opinion, but it is a valid and fair way (unlike 00Brak's method).

Assuming that you 'fully' reroll both dice in the event that there is only a partial match (say 40-50 on the tens die and there are more than 40 but fewer than 50 permanents, but the units die number gave a number higher than the total permanents). Otherwise some permanents may still be more likely to be selected.

In the event that there are less than or equal to 100/x permanents, it may be worth assigning each permanent to x numbers (say 1 and 51 in the event of 50 or fewer permanents (x=2)) to reduce the probability of rerolling.

Robvalue: Copy target bug or glitch. You may choose a new card for the copy...