I think that the only issue that matters here is whether or not the shuffle (after ‘de-clumping’) is sufficient. If the deck gets properly randomized, then the card order will be random and what order the cards were in prior to shuffling is irrelevant. Any advantage the player believes he gets from his technique is only in his head(*), making this no different than if a player always wears his lucky underwear on tournament days and circles his chair 3x clockwise before sitting down to play.


(*) I could imagine some scenarios where ‘de-clumping’ the library before shuffling would make a difference, such as breaking apart a pair of cards that are stuck together, but if anything that is only going to improve the randomization that occurs when the deck is shuffled.

Lyle, it sounds like you believe the player knows the distribution of the cards in his deck after shuffling. That is specifically called out in the IPG, so if that is what you believe then there is definitely grounds for something more here.

The question is whether the player actually knows the distribution, or just thinks he knows. Also do they actually know the distribution or do they just know it's better than it would be without this specific shuffle. Did you ask the player what benefit they get? Do they do it to avoid mana screw/flood, to draw more key spells more often, to draw sideboard cards…what specific benefit do they get? I'm also intrigued by the suggestion that the opponent should use the same technique; clearly the player does not believe they are doing anything wrong or they would not say that in front of two judges. Maybe take them up on the offer to learn the method so you can see how it works? That would help you a lot.

If it turns out that it's mere superstition, don't try and convince them that the shuffle does nothing. As long as it's not slow play and they aren't stacking, let them have their idiosyncrasy.

Originally posted by Chris Lansdell:

The question is whether the player actually knows the distribution, or just thinks he knows. Also do they actually know the distribution or do they just know it's better than it would be without this specific shuffle. Did you ask the player what benefit they get? Do they do it to avoid mana screw/flood, to draw more key spells more often, to draw sideboard cards…what specific benefit do they get? I'm also intrigued by the suggestion that the opponent should use the same technique; clearly the player does not believe they are doing anything wrong or they would not say that in front of two judges. Maybe take them up on the offer to learn the method so you can see how it works? That would help you a lot.

1) The player believes that they get a better distribution of lands and spells in their deck than if they shuffle “normally”. Which is to say, they claim that they mana flood/screw less often than if they shuffle “normally”.

2) The problem with learning/executing this method is, wouldn't I be cheating too if I tried it? I don't want to cheat =/

3) It was actually only within earshot of 1 judge (me) and I wasn't on duty at this event. I called the store judge over to handle the situation after the fact, although the admission of what he was doing was pretty much the same when he was talking to me or talking to the store judge in front of me (he also talked to the store judge with myself not there, and I don't know what he said at that time).

I have seen plenty of players think that ‘I don’t know where any of the cards are, or the distribution' is sufficient, while at the same time believing their lands/non-lands to be more evenly distributed.

Obviously if their shuffling is resulting in a normalized distribution, it is insufficient shuffling. Personally, I'm very much against pre-shuffle weaving, because it belies what is actually allowed. It gives the impression of an advantage (possibly an actual advantage) while proclaiming no foul. Even if sufficient shuffling happens, it allows both players to attribute success of the weaver to their weaving, right or wrong. It also allows them to perpetuate the notion that this does in fact make a difference. This is where I see it differs from the ‘lucky underwear’ comment; no one will perceive lucky to be anything more than superstition or luck (even if they believe it is lucky) - it is something external to the game. How one shuffles (if one shuffles sufficiently) directly impacts the game, and thus doesn't implicitly get the same leniency.

One thing I will point out, is that often players don't shuffle at regular, they cut. It's entirely possible they are, intentionally or not, relying on their opponents' lack of additional shuffling to maintain their land distribution. This (combined with the above) makes me want to find fault with this scenario. (This is my disclaimer - I am not stating a verdict, since I need to view it objectively before making one.)

One further note, ‘you can weave as long as you shuffle sufficiently’ feels like it will be taken by players as ‘this is technically not allowed, but here’s a loophole for you (just say you shuffled enough)'. Bad mouth feel all around, not sure there's an elegant one size fits all solution.

Originally posted by Lyle Waldman:

2) The problem with learning/executing this method is, wouldn't I be cheating too if I tried it? I don't want to cheat =/

I think there's nothing wrong with learning the method solely for the purpose of understanding it. I mean, we don't penalize players who know how to mana weave for cheating, only those who purposely use it to gain an advantage. Plus, if it does turn out to be a sufficient shuffling technique, you could (if you wanted to) use it without concern since it isn't an improper randomization technique.

Either way, the important step for determining whether it is a proper shuffle or not is to know how it works. I think that would just be good investigating, not cheating.

Edited Andrew Keeler (July 8, 2017 11:40:27 AM)

As a side topic of investigating further in these situations - I feel like the average shuffle is also not a very good sufficient randomization, which is why I support the “pile shuffle” as a pre-shuffling technique once per match as the rules allow. In my case, I always pile shuffle and then start to shuffle my deck “for real,” because if I do the average man shuffle I find that I'm not separating cards that were together in the previous game, lands in particular have this problem and imo that's not really random either. In my early days of magic (over 15 years ago) I did mana weaving in casual play but I no longer support that even as a “pre-shuffle” and haven't for over 10 years.

Investigation could help discover if something like a mana weave or a pile shuffle is intended as an advantage, and I personally think the biggest factor in that investigation is making sure any player that shuffles in these ways also then shuffles in an acceptable randomizing way afterwards. As long as they are doing that, I don't see it as suspicious.

Originally posted by Ryan Phillips:

once per match as the rules allow

Hi Ryan,

actually, rules allow pile “shuffling” once per game (i.e. once in each game of a match).

On Fri, Jul 7, 2017 at 4:54 PM, Brian Ross <
forum-36232-b3fc@apps.magicjudges.org> wrote:

> Obviously if their shuffling is resulting in a normalized distribution, it
> is insufficient shuffling.
>

Technically, we *want* normalized distribution,. but that does not mean
that lands and spells are evenly spaced. Rather, it means that the number
of “runs”* of lands and spells after shuffling is not so many as to suggest
that the player is forcibly separating them. Given a deck of sufficient
size (*Magic* decks qualify), the number of runs in properly randomized
decks approximates a normal distributtion, whose mean and standard
deviation are calculable. (I can send an Excel file privately that I
created which will calculate the curve, as well as give the probability of
a specific number of runs as well as where it fits on the scale) So if a
60-card control deck with 26 lands boasts 40 runs, the chances of getting
that many runs (or more) is less than 1%.

[*A “run” is a set of consecutive cards of the same type, which for our
purposes would be lands vs. spells. Runs can be any size from 1 to however
many of that type are in the deck. So a distribution of L, S, S, S, S, L,
S, L would count as five runs (three land, two spell). Note that
“declumping” is meant to separate the cards, thus creating more runs, while
mana screw/flood are created when there are larger runs (and thus the
number of runs is lower).]

We want players to shuffle their decks sufficiently and randomly. If you're performing any kind of statistical analysis or using the words “hang on, let me get my spreadsheet” you're probably looking too far past the problem.

Watch them shuffle; if you believe from what you've seen that they could have an idea of the distribution of the cards in their library or the locations of any particular types of cards at the end of this process, then you have a problem. That's all is expected of Judges and all that is expected of players.

The spreadsheet was primarily for my benefit, but has educational uses as
well. I know what I consider to be randomized; the sheet merely confirms or
denies what I suspect.

I love me some statistics! As long as we don't bring them to the tournament floor or use them as arguments in discussions with players, statistics are awesome; we also don't allow those combo's that with probability approaching to 1 will result in a certain combination of cards, like milling an opponent with 2 Emrakuls in their deck.

As Sophie said: we don't want players to have a deck that meets certain statistical criteria; we want players to randomise properly, and we determine that by looking at their efforts while shuffling.

Also, using statistics here is very dangerous. We're not interested in the likelihood of this number of runs given a proper randomisation; we want to know the likelihood of a proper randomisation given this distribution, and that is a completely different question that can't be answered with some simple normal distribution approximation…

On 3 Sep 2017 08:40, “Dustin De Leeuw” <
forum-36232-06cd@apps.magicjudges.org> wrote:


Also, using statistics here is very dangerous. We're not interested in the
likelihood of this number of runs given a proper randomisation; we want to
know the likelihood of a proper randomisation given this distribution, and
that is a completely different question that can't be answered with some
simple normal distribution approximation…

Dustin I love you for saying this! Could you go to lawschool and become my
collegue, so we could try to explain this to people at my work together? :)

There's only one stat you need: A proper randomization takes place after
log(2)+1 imperfect riffle shuffles (or mash equivalents). That happens to
be equal (except for massive decks such as those involving Battle of Wits)
to one more than the number of Swiss rounds needed for a tournament of that
many players. A tournament of 60 players takes 6 rounds, so a proper
randomization of 60 cards takes 7 shuffles. Pile-counting is not a shuffle.

If someone mana weaves and then shuffles a number of times seriously below
this, I may very well point out the discrepancy between what the deck
turned out and what “should” happen between rounds. For the record, the
expected number of runs is (2 x Lands x Spells)/(Deck Size) +1.

Could you elaborate on why, if I have read correctly, the number of shuffles sufficient for randomisation is dependent upon the number of rounds in the event (though log(2)+1 is a constant, not a variable, so I expected I have either misread or there is a typo)?

Should I shuffle 16 times each round at a GP? That seems excessive and also counter-intuitive; surely if a shuffle is sufficient in one event, an equivalent shuffle should be sufficient at any event?

Originally posted by Dominick Riesland:

If someone mana weaves and then shuffles a number of times seriously below
this, I may very well point out the discrepancy between what the deck
turned out and what “should” happen between rounds. For the record, the
expected number of runs is (2 x Lands x Spells)/(Deck Size) +1.

So you are saying the randomness isnt random because it doesn't fit your kind of random? Or did I get that wrong?