I, like many others, get pretty annoyed when getting mana screwed, color screwed, flooded or drawing too many of the same card on MTGO. Over the years I have frequently quit MTGO because of the shuffler but come back because of the game. I decided recently to do an analysis of opening hands on MTGO to see if the system really is rigged. Below are links to Google Docs of 100 sample hands on MTGO and the statistical implications I got from those hands in terms of how many lands are being drawn versus how many should be, duplicate copies of the same card in your opening hand, etc. What I discovered is that the MTGO shuffler isn't rigged--it's truly random. However, I've started (but not finished) a similar analysis for a real life version of drawing 100 sample hands and analyzing them the same way and have come to the conclusion that it's real life that has the rigged shuffling. What do you guys think of the attached results? Should the MTGO shuffler be changed to work more like real life? Do you know a general formula that would work for it? Should we shuffle differently in real life? Do you need an explanation on any of the stats or did I get anything wrong (probably)?
Links to Google Docs with PDFs
The reason sample real life hands appear rigged is that people are generally not all that great at actually randomizing the contents of their decks.
The rules require you to present a randomized deck. The shuffler does that properly. There's no reason to nerf the online shuffler just because people (sometimes intentionally, sometimes unintentionally) are bad shufflers.
After sampling 100 IRL hands, I can say with some degree of confidence that they are rigged. You draw more lands IRL and are more likely to draw the "desired" number of lands. It's true, the MTGO shuffler is "truly" random and IRL shuffling simply isn't, because at the end of most games your lands and spells aren't together, they get sloppily riffled into your deck, then you pile and riffle shuffle and your opponent cuts and everything is still somewhat ordered. I chose to replicate this process with a relatively thorough shuffling method and the results are below.
I don't think that if the MTGO shuffler worked like IRL shuffling it would be "nerfed" it would simply be more realistic and imperfect at randomization. Everyone is technically a bad shuffler as no one can truly independently and consciously create "randomness".
Except for cases of cheating decks that are shuffled at FNM/GP's and PTs are randomized.
It is very weird that original post uses a faulty arguement from a place not related to the computerized shuffler to say that that computerized shuffler produces random shuffles.
True randomization is impossible. If you pile lands and spells together after a game and riffle that pile into your deck, then pile and riffle shuffle the average number of times, then there will be clumps that remain together and this will result in clumps of lands and spells that are preferable to the true randomization of the online shuffler. This document indicates the statistical differences between 100 hands on MTGO and 100 real life hands with the same deck and how the real life version results in more hands. The sample size needs to be upped, of course, to obtain a lower p-value (which is currently unacceptably high) but the trends are relatively clear. My argument isn't faulty, it's supported by statistics and years of anecdotal experience.
Thanks for the reference @pizzap! That's a much better statistical analysis. I assume the sample hand and in game shuffler work off the same randomization method, though (EDIT: read the Reddit post, appears they are different but I will work off the 13,000 hand statistic from now on). I believe if I drew 13,000 hands with it my results would concur with his. As for real life shuffling, I'll continue testing the randomization but I expect it to maintain it's slightly skewed nature.
There is a way to get random shuffling in real life. Just bring a smartphone with a program that can produce random numbers. Put your cards face down in line, then let your shuffling program tell you which card to put on top of the pile next. Then maybe shuffle once or twice normally, hand your deck to your opponent, and let them do the same. Totally random (you can even pick a smartphone program that uses true sources of randomness like random.org).
The issue is not that truly randomized shuffling in real life is impossible (though it essentially is if you regularly shuffle at all, though if you're going off the random.org method, yes it would be random). The issue is that most people *don't* shuffle that way and so they have a mana weaving advantage. I have sort of outlined that in the stats but will continue to increase the sample size to make sure it's accurate.
True randomization is impossible. If you pile lands and spells together after a game and riffle that pile into your deck, then pile and riffle shuffle the average number of times, then there will be clumps that remain together and this will result in clumps of lands and spells that are preferable to the true randomization of the online shuffler. This document indicates the statistical differences between 100 hands on MTGO and 100 real life hands with the same deck and how the real life version results in more hands. The sample size needs to be upped, of course, to obtain a lower p-value (which is currently unacceptably high) but the trends are relatively clear. My argument isn't faulty, it's supported by statistics and years of anecdotal experience.
Simple: a procedure in which all possible permutations of cards in the deck are equally likely in outcome. The rules simple say "random order", which means the same thing..."true randomization" simply emphasizes that a number of procedures people use to produce a "random order" don't actually do so.
I wish we had the luxury of riffle shuffling, but cards are too expensive to put so much wear on them. I wonder though if 'side shuffling' is as random as riffle-shuffling...I usually do 10-20 of these, and I know for a fact that random shuffling doesn't happen nearly ever in older formats because of fetches. I always get accused of 'taking too long' after fetching/opponent fetching for properly shuffling. Perhaps we should go to 55 min rounds and have better shuffling enforcement.
100 samples mean *nothing* for statistical analysis of this sort.
Run a sample size of fifty thousand, and track the number of lands in each, then compute the mean, median, standard deviation, skew and curtosis of the distribution function and compare it to the calculated expectations. 50k is about the number needed to tell the difference between the shuffler correctly giving you 40% land, or wrongly giving you 39.75% or 40.25% land (assuming a 24 land/60 card deck).
I would love you guys to break down a recent experience I had recently that has me hesitant to put any more money into mtgo until I am comfortable it is fixed or explained. I didn't take screen shots for proof, but I am sure many of you have had this experience.
I entered a SOI swiss draft recently. I drafted a green/white humans deck. Low curve. I think one 5 mana card. The rest from 1-4 drops Mostly 2's and 3's.
It had 14 Green spells and 10 White. 16 Land. 9 Forest, 7 Plains.
First two games I won easily. A couple of mulligans. One match loss.
Round 3. First hand: No land mulligan. Second:1 Forest, all White spells mulligan. Third: 2 Forest. 2 green spells. 1 White. Lost around turn 8 after only drawing green spells and forest.
Second match. Keep 3 Forest. 2 Green spells. 2 White spells. Bottom forest off mulligan. Draw forest. Draw white spell. Draw white spell. Draw forest. Draw forest. Draw white spell. Concede.
I am not talking about a conspiracy, nor do I understand the arithmetic used in the programming but as a very common sense and practical person normally, something just doesn't add up about those types of games. I doubt I am the only one who gets those types of rounds. I have a decent grasp on the fundamentals of mtg and I just don't see how those types of matches happen. This is not an isolated event either. I have been playing mtgo for about 5 years on and off and there are just ruts that I go through with this type of mess. Games where you draw 7 spells in a row even though card draw spells. Or games where you draw your only non colored sources every game. (Ghost quarter in every opener)
Also, I watch a lot of draft videos (LSV) and I notice quite often, He will have similar issues. (6 land draws in a row or 3 same spells in opener)I have never noticed this issue the same way in paper magic. There are games with mana clump, but it seems online I have this issue tons during certain time periods and not during others...
Ultimately, I have just decided to not put money into it for right now until I have either forgotten about the sting or until it is addressed in some way. Perhaps scrutiny of the source code is warranted? Have they released that to anyone to check over? I know there would be issues with either people using the source to come up with cheats or with copyright issues, but to me, that would help with the integrity of the game. They may have just lost my 40 dollars a month, but for electronic cards, I guess they really don't mind.
You are sort of describing the answer to your implied question yourself ("are these events random?"), by pointing out that it is happening repeatedly. Drawing forests and only white spells is what actually may happen from time to time when you're dealing with actual* randomness. It might be a bit bitter when you're subject to it, but it may be a comfort to you in knowing that we are all equally subject to it on average.
I've argued with people over the number of cards in a deck many times, and I feel that having a high number of "good cards" is better than having a small number of "really good" cards. I laugh at the people who argue that with a smaller deck you're more likely to draw what you need, because if you get in that situation you're gonna NOT end up drawing it many many more times than you would.
I play limited and I also watch a whole lot of MtGO streams and videos (only draft and sealed). I do not believe there is any issue with the shuffler. That's just my own opinion, of course.
There are games, matches, or even drafts in which mana issues decide the outcomes but I believe it to be clearly within the nature of randomization as would occur with cards. There are also plenty of drafts I play in where mana issues do not occur much for either player.
It's the nature of luck that it will often go in streaks.
Yup, streaks don’t happen as often in paper magic because everything people do when handling their cards is meant to avoid streaks (mana screwed/flooded). In MTGO, they don’t have that option, so there’s a bigger change of having bad draws.
On the rare occasions that I play in a paper event, it seems like my opponents never really bother to shuffle. They go through the motions...I think its more ignorance than bad intent.
true, but in playing about 10 matches of 3 games each, I had some type of mana problem over 80% of the time. I'm playing RW Burn and only have 19 lands so when I get flooded 3 games in a row it's quite agitating. Oddly enough, after about 5 matches and getting flooded or color screwed (which I admit is most likely due to the mana base because it's budget) I actually put 3 more lands in and ended up getting mana screwed the next few games and on and on. I don't know what the deal is, in a deck with 19 lands you don't want to have 8 lands on turn 9. It wouldn't frustrate me if it wasn't happening over 50% of the time but every time I play MTGO it seems I'm getting mana screwed or mana flooded, no matter the deck, no matter the game. As for real life, personally, in a tournament I begin with a 9 pile, pile shuffle and then I bridge for the rest of the duration of shuffling, present no cards to anyone and maintaining an order that is as randomized as possible. This seems to work so anyone out there take it from me, that is if you don't mind bridging an $800 modern deck LOL but after all, they are just cards.
Private Mod Note
():
Rollback Post to RevisionRollBack
To post a comment, please login or register a new account.
Links to Google Docs with PDFs
100 Sample hands on MTGO with GW Devotion (1st page), related statistical analysis (2nd page) and in-progress sample hands IRL (3rd page, ignore for now)
https://drive.google.com/file/d/0ByNR1o2kAOymOWxYdTJpU1V5ZVU/view?usp=sharing
The rules require you to present a randomized deck. The shuffler does that properly. There's no reason to nerf the online shuffler just because people (sometimes intentionally, sometimes unintentionally) are bad shufflers.
https://drive.google.com/file/d/0ByNR1o2kAOymM2FIWjVxcm1vODA/view?usp=sharing
I don't think that if the MTGO shuffler worked like IRL shuffling it would be "nerfed" it would simply be more realistic and imperfect at randomization. Everyone is technically a bad shuffler as no one can truly independently and consciously create "randomness".
It is very weird that original post uses a faulty arguement from a place not related to the computerized shuffler to say that that computerized shuffler produces random shuffles.
https://drive.google.com/file/d/0ByNR1o2kAOymM2FIWjVxcm1vODA/view?usp=sharing
"True randomization", what is that ?
The mtg rules never speak of this.
Run a sample size of fifty thousand, and track the number of lands in each, then compute the mean, median, standard deviation, skew and curtosis of the distribution function and compare it to the calculated expectations. 50k is about the number needed to tell the difference between the shuffler correctly giving you 40% land, or wrongly giving you 39.75% or 40.25% land (assuming a 24 land/60 card deck).
I entered a SOI swiss draft recently. I drafted a green/white humans deck. Low curve. I think one 5 mana card. The rest from 1-4 drops Mostly 2's and 3's.
It had 14 Green spells and 10 White. 16 Land. 9 Forest, 7 Plains.
First two games I won easily. A couple of mulligans. One match loss.
Round 3. First hand: No land mulligan. Second:1 Forest, all White spells mulligan. Third: 2 Forest. 2 green spells. 1 White. Lost around turn 8 after only drawing green spells and forest.
Second match. Keep 3 Forest. 2 Green spells. 2 White spells. Bottom forest off mulligan. Draw forest. Draw white spell. Draw white spell. Draw forest. Draw forest. Draw white spell. Concede.
I am not talking about a conspiracy, nor do I understand the arithmetic used in the programming but as a very common sense and practical person normally, something just doesn't add up about those types of games. I doubt I am the only one who gets those types of rounds. I have a decent grasp on the fundamentals of mtg and I just don't see how those types of matches happen. This is not an isolated event either. I have been playing mtgo for about 5 years on and off and there are just ruts that I go through with this type of mess. Games where you draw 7 spells in a row even though card draw spells. Or games where you draw your only non colored sources every game. (Ghost quarter in every opener)
Also, I watch a lot of draft videos (LSV) and I notice quite often, He will have similar issues. (6 land draws in a row or 3 same spells in opener)I have never noticed this issue the same way in paper magic. There are games with mana clump, but it seems online I have this issue tons during certain time periods and not during others...
Ultimately, I have just decided to not put money into it for right now until I have either forgotten about the sting or until it is addressed in some way. Perhaps scrutiny of the source code is warranted? Have they released that to anyone to check over? I know there would be issues with either people using the source to come up with cheats or with copyright issues, but to me, that would help with the integrity of the game. They may have just lost my 40 dollars a month, but for electronic cards, I guess they really don't mind.
https://www.reddit.com/r/magicTCG/comments/2o73b4/incontrovertible_fact_of_the_unfairness_of_the/
You are sort of describing the answer to your implied question yourself ("are these events random?"), by pointing out that it is happening repeatedly. Drawing forests and only white spells is what actually may happen from time to time when you're dealing with actual* randomness. It might be a bit bitter when you're subject to it, but it may be a comfort to you in knowing that we are all equally subject to it on average.
Let me know if I misunderstood your question.
*approximated
There are games, matches, or even drafts in which mana issues decide the outcomes but I believe it to be clearly within the nature of randomization as would occur with cards. There are also plenty of drafts I play in where mana issues do not occur much for either player.
It's the nature of luck that it will often go in streaks.