I know my experience is a small sample size, but it seems, more often than not, the land is totally skewed one way of the other - too few lands or too many and not enough anything else. And then you add in 2 or 3 color decks and you give the Random Brutalizer (as I like to call it) yet another weapon in its bag of tricks. Now it can starve you in one color. Very infrequently do you get a fair distribution.
It’s hard to tell but some, including myself, suspect it has a lot to do with the hand shaping code involved. There was an experiment (too lazy to look for the post now) that involves exporting a deck list, modifying its card order, then re-importing the list and it resulted in far smoother shuffles. It appears that WotC fixed this though. Alternatively, there are anecdotes that mulliganing at least once will stop the lilted land draw and return the randomness to something more normal. This is something I’ve experienced.
Or do what some people do and build decks that take advantage of the hand shaping. This is why 13-land Burn is a thing and actually works.
A lot of people, and I’m sure they’ll pop up here and say as much, say it’s just random chance.
But I don’t think randomness is playing as big as a role as it should here. We already know that Arena shapes your hand in BO1 and it’s suspected that some of that code be ure-used for BO3. The developers has said as much. As soon as anyone starts messing with statistics and randomness then... well... it’s not random, is it?
It’s weird to say this but I don’t really consider Arena the same as paper Magic. At all. It’s not even a suitable replacement. The weird hand shaping. The weird matchups. The garbage clumping. The rankings getting reset so often. The cheesy “start first” nonsense. The inability to B/S/T to get a specific deck. It plays more like its own format entirely separate from any paper format.
This is not to say that the code is completely random, due to the handshaping element in whatever format that applies. But the color distribution/screw, the either flood or starvation of lands after your opening hand, that's just randomness.
Humans suck at predicting/understanding randomness. We are also creatures of pattern recognition, even when there isn't a pattern to recognize.
When you think of a randomized deck, you likely think of one with a relatively even distribution of lands throughout the deck. But true randomness is rarely relatively even, and often filled with more clumps and other distribution weirdness. A truly randomized deck will have significantly more clumps of lands than you would expect.
This is not to say that the code is completely random, due to the handshaping element in whatever format that applies. But the color distribution/screw, the either flood or starvation of lands after your opening hand, that's just randomness.
Humans suck at predicting/understanding randomness. We are also creatures of pattern recognition, even when there isn't a pattern to recognize.
When you think of a randomized deck, you likely think of one with a relatively even distribution of lands throughout the deck. But true randomness is rarely relatively even, and often filled with more clumps and other distribution weirdness. A truly randomized deck will have significantly more clumps of lands than you would expect.
In paper Magic, it can be proven when decks are shuffled randomly, including clumps and streaks. Statistically, I can expect less than 47.6% chance of drawing 2 or more lands in my opening hand for 13 land red. Too low to be statistically reliable to be playable. Even Autumn Burchett Mono-Blue used 19 Islands for a 57.2% probability of drawing at least two Islands. But if you move over to Arena, these statistics are worthless thanks to the hand shaping involved.
In order to get that hand shaping to work reliably and consistently the deck is “shuffled” twice and an opening hand is chosen. Given that vast majority of computers don’t have TRUE RNG anyways, the hand shaping algorithm also intensely favors large amounts of clumping in order to get the correct hand shaping. In other words, the shuffle algorithm is likely purposely flawed in order to better hit the desired land counts in the opening hand. Ergo, not randomized properly.
Does it seem, with the release of M20, that they turned the land abuse up another notch?
Sort of. The card clumping that happens seems to be a bit more dramatic. It’s almost like it “flip-flops” now. For instance, I’ll either consistently draw the same group of cards or another group of cards but not cards from both groups.
To put it another way. Start with a sorted deck. Then split the deck in half. Shuffle each half separately. Then stack each half one on top of the other. Whatever cards I draw seem to be from this first half or the second half. So if I draw card X, it’s unlikely I’ll ever see card Y unless my deck is shuffled or milled during the game.
I haven't played Arena in almost a couple of months. Soon as I started playing again yesterday I immediately noticed how crappy my hands were.
Here's a screen shot with a .62% chance of happening. (Note that I had already used a Bond of Flourishing that allowed me to dig 3 cards deeper than the 24 cards left in the deck.) I had another game just a couple after that where I was stuck on 2 lands out of 19 cards which has a .11% chance of happening. At .11% should happen every 909 games. Another game or two goes by, draw 9 lands out of about 13 cards. This has a 1.4% chance of happening.
People that don't understand math will say things like Sharpened did "Oh it's just random, humans can't tell what's going on, we all just assign patterns where there aren't really patterns." This is absolutely true if you're talking about just using your brain without any kind of other tools. However, if you take screen shots and use a hypergeometric calculator and have an understanding of statistics, you can be nearly assured that the shuffler is broken. All it takes is playing 10 games and like 6 out of those 10 games will have these crazy unlikely hands/draws show up. Then you keep playing and notice it keeps happening. Then you see all the other people who complain about it too. I listen to the people who know how to use a hypergeometric calculator and give percentage chances and screen shots with their answer.
Fact is, you can't keep having these 1 in 300-900 games chances keep popping up at a rate of more like 1 in 2 games without the shuffler being obviously broken. Your instincts are correct and there is hard proof out there given by that buy who analyzed a million games.
Yes, the shuffler is definitely still broken it seems. There was a study done by a statistician that proved without a doubt the shuffler was broken, it analyzed over a million games. You can search for it on Reddit or find the link somewhere in this Arena forum. It really makes me sad since Arena is so great in every other way.
BTW I was the one who figured out you can manually switch your list around to fix the broken shuffler. There is a post about that too on this forum. It's rather tedious and involved, but I guarantee you that you will instantly see results with it. I figured out how to do this because of the author of that study mentioned he thought the card ordering in the list had a lot to do with the lack of randomness in the shuffler. So I took my list and evenly distributed the lands throughout it. Magically, I no longer experienced negative effects from the evil shuffler. This is an EVEN MORE obvious evidence that the shuffler is truly broken. When something truly randomizes something, it does not matter the order of the list to begin with. Even if the list is perfectly in "order" to begin with, one round of randomization will make it it totally random. The fact that changing the order of your cards in the list fixes the shuffler proves that it's broken.
Go to the Deck list screen and select a deck. At the bottom of the screen you should see an EXPORT option. Export the list and it will ask you to save the list somewhere on your hard drive. The resulting list is now a text file that can be opened in something like Notepad or SCiTE. Do NOT use MS Word.
Once you have the file open, take the listed cards and rearrange them as you see fit. Once you save the new file go back into arena at the same deck screen and choose IMPORT. Import the modified list.
In my own experiments I found that separating each card on a card by card basis (eg one line of 20 islands turns into twenty lines of 1 island) caused some Arena loading and editing problems but the deck is playable. It’s unclear to me whether doing so changes anything in the sort. Moving the cards on a line-by-line basis (one line of 20 Plains remains as one line of 20 but on a different line) seems to have better results.
However, one of the Arena update seems to have modified the EXPORT option showing the modified lists are rearranged yet again. I haven’t looked at that particular avenue since.
I have been playing Arena for a little over a month now. Constructed, Limited, special events, etc. and not once have I experienced this.
What are folks land counts? Are you using the auto feature to add lands? Sure, do I get games where I flood/screw? Absolutely. But it’s not something that I’d pin on a broken shuffler.
I'll never play Arena because I've seen plenty of evidence suggesting it's simply not random.
How hard is it to program a random card shuffler? Surely they could figure it out with all their big bucks.
Only an independent, comprehensive statistical survey saying that the shuffler is actually random could convince me. If you can't shuffle the decks, you can't play Magic.
Go to the Deck list screen and select a deck. At the bottom of the screen you should see an EXPORT option. Export the list and it will ask you to save the list somewhere on your hard drive. The resulting list is now a text file that can be opened in something like Notepad or SCiTE. Do NOT use MS Word.
Once you have the file open, take the listed cards and rearrange them as you see fit. Once you save the new file go back into arena at the same deck screen and choose IMPORT. Import the modified list.
In my own experiments I found that separating each card on a card by card basis (eg one line of 20 islands turns into twenty lines of 1 island) caused some Arena loading and editing problems but the deck is playable. It’s unclear to me whether doing so changes anything in the sort. Moving the cards on a line-by-line basis (one line of 20 Plains remains as one line of 20 but on a different line) seems to have better results.
However, one of the Arena update seems to have modified the EXPORT option showing the modified lists are rearranged yet again. I haven’t looked at that particular avenue since.
Looks like this is no longer an option. Too bad because I am getting exceptionally land abused this morning.
Go to the Deck list screen and select a deck. At the bottom of the screen you should see an EXPORT option. Export the list and it will ask you to save the list somewhere on your hard drive. The resulting list is now a text file that can be opened in something like Notepad or SCiTE. Do NOT use MS Word.
Once you have the file open, take the listed cards and rearrange them as you see fit. Once you save the new file go back into arena at the same deck screen and choose IMPORT. Import the modified list.
In my own experiments I found that separating each card on a card by card basis (eg one line of 20 islands turns into twenty lines of 1 island) caused some Arena loading and editing problems but the deck is playable. It’s unclear to me whether doing so changes anything in the sort. Moving the cards on a line-by-line basis (one line of 20 Plains remains as one line of 20 but on a different line) seems to have better results.
However, one of the Arena update seems to have modified the EXPORT option showing the modified lists are rearranged yet again. I haven’t looked at that particular avenue since.
Looks like this is no longer an option. Too bad because I am getting exceptionally land abused this morning.
Yeah. Since it’s such a PITA to fight the Arena system; I find it’s easier just to build decks that take advantage of it. The old 13 land red is a prime example of this. I’ve seen deck lists for a UW Flyer deck that uses a mere 21 or 22 lands on Arena but the paper version uses 24 or 25.
One thing I have yet to see is a detailed study/discussion on how the hand shaping affects the generally accepted statistics formulas on drawing cards.
To outline what I’m talking about. Using the above UW deck with 21 lands and an average CMC of 2.18, I know that I have a probability of 55.9% of drawing at least one Island and one Plains in my opening hand. I have roughly a 25-28% possibility of drawing some combination of any three lands in the same hand. And roughly 17.7% probability of drawing just one in my opening hand. On Arena, the theoretical weighted land sequence for 21 lands is 2>3>1>4>0>5>6>7. almost exactly in line with what the UW deck actually wants. So my probabilities for drawing a hand of 2 lands vs 3 lands is increased substantially. Ignoring color screw entirely of course.
There is an article that outlines this over at Channel Fireball. I haven’t bothered to verify the math it does illustrate my opening point well enough.
However, now we get to the meat of my point. How are the statistic influenced for the other 53 cards as you draw them? In other words, if Arena drew you a two land hand and discarded the three land hand, how often were an arbitrary number of cards following the chosen hand land or not land compared to the discarded hand? This is something we can’t know because we have virtually no information about the discarded hand. I mean, we know that if we have a three land hand that Arena never drew a two land hand but that’s about it. This is where I suspect is one potential avenue of the “outlier” draws that WotC colloquially refers to.
To put my example into the real world.
How often have you mulliganned due to mana screw/flood that you looked at the top three or so cards to see what you could have drawn before you shuffle and redraw? Try it next time and note whether the next few cards coincide with the mana in your hand. Or if you’re afraid of getting in trouble at your FNM, just goldfish the deck at home. Shuffle the deck. Draw a hand, then flip the next three cards. Based on the opening hand, what was the distribution like across the next few cards?
Played 3 games today. One of them had a .088% chance or a 1 in 1,136 game chance of happening. 4 lands out of 19 cards with 17 lands in the 40 card deck. This game is obviously rigged I feel more and more. I don't believe that Wizards can't hire someone to make them a random shuffler. The longer this goes on, the more I feel it's intentional. Screen shot for proof of their bull*****.
BTW I started the game with 3 lands in hand. Drew a 4th on turn 5 or so. That means I didn't draw a land for 8 turns in a row. This is obviously utter bull*****. This kind of thing has happened to me maybe a few times in my entire life of paper Magic. On Arena it's about every 3rd or 4th game with this kind of nonsense.
Or do what some people do and build decks that take advantage of the hand shaping. This is why 13-land Burn is a thing and actually works.
A lot of people, and I’m sure they’ll pop up here and say as much, say it’s just random chance.
But I don’t think randomness is playing as big as a role as it should here. We already know that Arena shapes your hand in BO1 and it’s suspected that some of that code be ure-used for BO3. The developers has said as much. As soon as anyone starts messing with statistics and randomness then... well... it’s not random, is it?
It’s weird to say this but I don’t really consider Arena the same as paper Magic. At all. It’s not even a suitable replacement. The weird hand shaping. The weird matchups. The garbage clumping. The rankings getting reset so often. The cheesy “start first” nonsense. The inability to B/S/T to get a specific deck. It plays more like its own format entirely separate from any paper format.
This is not to say that the code is completely random, due to the handshaping element in whatever format that applies. But the color distribution/screw, the either flood or starvation of lands after your opening hand, that's just randomness.
Humans suck at predicting/understanding randomness. We are also creatures of pattern recognition, even when there isn't a pattern to recognize.
When you think of a randomized deck, you likely think of one with a relatively even distribution of lands throughout the deck. But true randomness is rarely relatively even, and often filled with more clumps and other distribution weirdness. A truly randomized deck will have significantly more clumps of lands than you would expect.
In paper Magic, it can be proven when decks are shuffled randomly, including clumps and streaks. Statistically, I can expect less than 47.6% chance of drawing 2 or more lands in my opening hand for 13 land red. Too low to be statistically reliable to be playable. Even Autumn Burchett Mono-Blue used 19 Islands for a 57.2% probability of drawing at least two Islands. But if you move over to Arena, these statistics are worthless thanks to the hand shaping involved.
In order to get that hand shaping to work reliably and consistently the deck is “shuffled” twice and an opening hand is chosen. Given that vast majority of computers don’t have TRUE RNG anyways, the hand shaping algorithm also intensely favors large amounts of clumping in order to get the correct hand shaping. In other words, the shuffle algorithm is likely purposely flawed in order to better hit the desired land counts in the opening hand. Ergo, not randomized properly.
Sort of. The card clumping that happens seems to be a bit more dramatic. It’s almost like it “flip-flops” now. For instance, I’ll either consistently draw the same group of cards or another group of cards but not cards from both groups.
To put it another way. Start with a sorted deck. Then split the deck in half. Shuffle each half separately. Then stack each half one on top of the other. Whatever cards I draw seem to be from this first half or the second half. So if I draw card X, it’s unlikely I’ll ever see card Y unless my deck is shuffled or milled during the game.
Here's a screen shot with a .62% chance of happening. (Note that I had already used a Bond of Flourishing that allowed me to dig 3 cards deeper than the 24 cards left in the deck.) I had another game just a couple after that where I was stuck on 2 lands out of 19 cards which has a .11% chance of happening. At .11% should happen every 909 games. Another game or two goes by, draw 9 lands out of about 13 cards. This has a 1.4% chance of happening.
People that don't understand math will say things like Sharpened did "Oh it's just random, humans can't tell what's going on, we all just assign patterns where there aren't really patterns." This is absolutely true if you're talking about just using your brain without any kind of other tools. However, if you take screen shots and use a hypergeometric calculator and have an understanding of statistics, you can be nearly assured that the shuffler is broken. All it takes is playing 10 games and like 6 out of those 10 games will have these crazy unlikely hands/draws show up. Then you keep playing and notice it keeps happening. Then you see all the other people who complain about it too. I listen to the people who know how to use a hypergeometric calculator and give percentage chances and screen shots with their answer.
Fact is, you can't keep having these 1 in 300-900 games chances keep popping up at a rate of more like 1 in 2 games without the shuffler being obviously broken. Your instincts are correct and there is hard proof out there given by that buy who analyzed a million games.
Yes, the shuffler is definitely still broken it seems. There was a study done by a statistician that proved without a doubt the shuffler was broken, it analyzed over a million games. You can search for it on Reddit or find the link somewhere in this Arena forum. It really makes me sad since Arena is so great in every other way.
BTW I was the one who figured out you can manually switch your list around to fix the broken shuffler. There is a post about that too on this forum. It's rather tedious and involved, but I guarantee you that you will instantly see results with it. I figured out how to do this because of the author of that study mentioned he thought the card ordering in the list had a lot to do with the lack of randomness in the shuffler. So I took my list and evenly distributed the lands throughout it. Magically, I no longer experienced negative effects from the evil shuffler. This is an EVEN MORE obvious evidence that the shuffler is truly broken. When something truly randomizes something, it does not matter the order of the list to begin with. Even if the list is perfectly in "order" to begin with, one round of randomization will make it it totally random. The fact that changing the order of your cards in the list fixes the shuffler proves that it's broken.
Go to the Deck list screen and select a deck. At the bottom of the screen you should see an EXPORT option. Export the list and it will ask you to save the list somewhere on your hard drive. The resulting list is now a text file that can be opened in something like Notepad or SCiTE. Do NOT use MS Word.
Once you have the file open, take the listed cards and rearrange them as you see fit. Once you save the new file go back into arena at the same deck screen and choose IMPORT. Import the modified list.
In my own experiments I found that separating each card on a card by card basis (eg one line of 20 islands turns into twenty lines of 1 island) caused some Arena loading and editing problems but the deck is playable. It’s unclear to me whether doing so changes anything in the sort. Moving the cards on a line-by-line basis (one line of 20 Plains remains as one line of 20 but on a different line) seems to have better results.
However, one of the Arena update seems to have modified the EXPORT option showing the modified lists are rearranged yet again. I haven’t looked at that particular avenue since.
What are folks land counts? Are you using the auto feature to add lands? Sure, do I get games where I flood/screw? Absolutely. But it’s not something that I’d pin on a broken shuffler.
How hard is it to program a random card shuffler? Surely they could figure it out with all their big bucks.
Only an independent, comprehensive statistical survey saying that the shuffler is actually random could convince me. If you can't shuffle the decks, you can't play Magic.
Low-power cube enthusiast!
My 1570 card cube (no longer updated)
My 415 Peasant+ Artifact and Enchantment Cube
Ever-Expanding "Just throw it in" cube.
Looks like this is no longer an option. Too bad because I am getting exceptionally land abused this morning.
Yeah. Since it’s such a PITA to fight the Arena system; I find it’s easier just to build decks that take advantage of it. The old 13 land red is a prime example of this. I’ve seen deck lists for a UW Flyer deck that uses a mere 21 or 22 lands on Arena but the paper version uses 24 or 25.
One thing I have yet to see is a detailed study/discussion on how the hand shaping affects the generally accepted statistics formulas on drawing cards.
To outline what I’m talking about. Using the above UW deck with 21 lands and an average CMC of 2.18, I know that I have a probability of 55.9% of drawing at least one Island and one Plains in my opening hand. I have roughly a 25-28% possibility of drawing some combination of any three lands in the same hand. And roughly 17.7% probability of drawing just one in my opening hand. On Arena, the theoretical weighted land sequence for 21 lands is 2>3>1>4>0>5>6>7. almost exactly in line with what the UW deck actually wants. So my probabilities for drawing a hand of 2 lands vs 3 lands is increased substantially. Ignoring color screw entirely of course.
There is an article that outlines this over at Channel Fireball. I haven’t bothered to verify the math it does illustrate my opening point well enough.
However, now we get to the meat of my point. How are the statistic influenced for the other 53 cards as you draw them? In other words, if Arena drew you a two land hand and discarded the three land hand, how often were an arbitrary number of cards following the chosen hand land or not land compared to the discarded hand? This is something we can’t know because we have virtually no information about the discarded hand. I mean, we know that if we have a three land hand that Arena never drew a two land hand but that’s about it. This is where I suspect is one potential avenue of the “outlier” draws that WotC colloquially refers to.
To put my example into the real world.
How often have you mulliganned due to mana screw/flood that you looked at the top three or so cards to see what you could have drawn before you shuffle and redraw? Try it next time and note whether the next few cards coincide with the mana in your hand. Or if you’re afraid of getting in trouble at your FNM, just goldfish the deck at home. Shuffle the deck. Draw a hand, then flip the next three cards. Based on the opening hand, what was the distribution like across the next few cards?
BTW I started the game with 3 lands in hand. Drew a 4th on turn 5 or so. That means I didn't draw a land for 8 turns in a row. This is obviously utter bull*****. This kind of thing has happened to me maybe a few times in my entire life of paper Magic. On Arena it's about every 3rd or 4th game with this kind of nonsense.