## fork from locked: shuffler

8 posts / 0 new
zenbitz
Joined Mar 2013
1603 Posts

In a perfect shuffle paradigm. Which, this isn't. It's fake math man, get over it.

What does this mean?

Also, in another SB bitch thread someone commmented on "the shuffler" -- what does THAT mean?

rstnme: "Everything looks good when your opponent passes 4 turns in a row."
silentbobus
Joined Jun 2009
8307 Posts
The shuffler means the mechanism for cards that are dealt in MtGO. People are generally critical of it, and it never underwent the same level of rigorous testing that the Poker sites went through after most of their original shufflers were found to be non-random. The moderators do not like people discussing it outside of the single dedicated 1740 post thread on the Online forum.
rstnme
Joined Aug 2012
4263 Posts
MTGO has a shuffler to randomize cards. I guess "the shuffler" is how we refer to it.

The math people use to determine the percentage of drawing a certain card hinges on an even distribution of cards... meaning it as likely to get one card as another, and so you can "calculate" the likelihood of, say, drawing a land. The perfect shuffle is an even distribution of cards based on the percentage of drawing them, which is almost a statistical impossibility.

I found Carmen Sandiego before you were born unless you're Zlehtnoba.

zenbitz
Joined Mar 2013
1603 Posts
From what I know of modern computer programming, there is almost no chance that that MTGO shuffler is buggy.   I know people who play with automatic die servers and built in stuff and they CONSTANTLY complain about the "randomness".  Poppycock.   Random doesn't LOOK distributed or perfect (like your point) it looks clumpy.

But I dispute your point about the math being fake.  It's not fake, it just only applies over a very large number of hands/games/draws.

A "perfect shuffle" i.e, a perfectly spell/land curved deck DOES occur.   You can actually calculate the chance it DOES occur..

You want:
3-4 lands to start
4 lands by turn 4
and spells that maximally use your mana.  Probably occurs about 20-30% of the time depending on your deck.  It goes down if you want to curve perfectly to 5 or 6 mana, obviously.

This is why a perfect player can be expected to lose at least 20% of the time.  If you add up your mandated mulls to 5 and the times villain curves perfectly and you don't, you usually lose these.
rstnme: "Everything looks good when your opponent passes 4 turns in a row."
rstnme
Joined Aug 2012
4263 Posts
From what I know of modern computer programming, there is almost no chance that that MTGO shuffler is buggy.   I know people who play with automatic die servers and built in stuff and they CONSTANTLY complain about the "randomness".  Poppycock.   Random doesn't LOOK distributed or perfect (like your point) it looks clumpy.

But I dispute your point about the math being fake.  It's not fake, it just only applies over a very large number of hands/games/draws.

A "perfect shuffle" i.e, a perfectly spell/land curved deck DOES occur.   You can actually calculate the chance it DOES occur..

You want:
3-4 lands to start
4 lands by turn 4
and spells that maximally use your mana.  Probably occurs about 20-30% of the time depending on your deck.  It goes down if you want to curve perfectly to 5 or 6 mana, obviously.

This is why a perfect player can be expected to lose at least 20% of the time.  If you add up your mandated mulls to 5 and the times villain curves perfectly and you don't, you usually lose these.

The fact that a perfect shuffle "does happen" and therefor this math is accurate is ludicrous, I tell you, ludicrous! The chance for a perfect distribution is too tiny for it to be the basis of your draw percentages. It's a lazy way of thinking.

I found Carmen Sandiego before you were born unless you're Zlehtnoba.

zenbitz
Joined Mar 2013
1603 Posts

The fact that a perfect shuffle "does happen" and therefor this math is accurate is ludicrous, I tell you, ludicrous! The chance for a perfect distribution is too tiny for it to be the basis of your draw percentages. It's a lazy way of thinking.

OK, I will convince myself.   Post a deck and a set of perfect shuffles and I will tell you the odds.  It will only take about 5 minutes to edit the excel sheet I have.

But it seems likely that we are defining "perfect shuffle" differently - so please be precise.

OLD MAN RANT:
I find the use of the word "curve" in MTG to be maddeningly imprecise as well.  I mean,  I kinda know what  y'all mean when you say "curve", kinda.  You know.

The "curve" of a deck is a property of it.  That's simple.  It's just the distribution of cards as  a function of CMC.   So this cannot be "perfect".  That would be like a perfect height.  It has to be perfect for some PURPOSE.   Similarly if a deck "curves at 4" WTH does that mean?  The mode of the distribution is CMC 4? The mean?  Does it only apply if the curve  is somehow "perfect".

Drives me nuts.  It's like no one who plays magic seriously ever bothered to pick up a text book and learn the names for things.

/old man rant
rstnme: "Everything looks good when your opponent passes 4 turns in a row."
silentbobus
Joined Jun 2009
8307 Posts
From what I know of modern computer programming, there is almost no chance that that MTGO shuffler is buggy.   I know people who play with automatic die servers and built in stuff and they CONSTANTLY complain about the "randomness".

The arc for modern computer programming is finding out that processes people previously thought were random are in fact not random and deterministically predictable. Randomness is not something that happens by accident, to be considered truly random a process must pass a battery of 15 specific tests, and the ones that don't pass often have errors that are easily visibile if you look at the generated numbers in the right manner. For example, if you plot them on a grid, a random process should look like snow, but an errant RNG will organize the numbers into lines, squares and other geometric patterns.
zenbitz
Joined Mar 2013
1603 Posts
Yes, I know.  And I presume you also know that there is no such thing as a truely random (computer) number generator, because computers are deterministic.   They are only pseudorandom, BUT typically off-the-shelf ones have very, very, very long repeat times.

I used to run monte carlo chemical physics simulations where you need at least 10^9 or 10^10 in sequence run.  That's probably sufficient for MTG.  You will likely only play with the same limited deck 10-20 times.   And even if you have a "bad" RNG if it's used on a rotating basis for different decks you will never catch it shuffling badly.
rstnme: "Everything looks good when your opponent passes 4 turns in a row."