Shuffler
Enabled "Smooth Shuffling" for the Play queue.
Important Note: This is for the "Play" queue only. It does not apply to Ranked Traditional Play, Traditional Ranked, or any other events and/or formats.
More Information on Smooth Shuffling: We are currently testing changes to the shuffler algorithm to decrease the number of games with extreme examples of "mana flood" or "mana screw" (drawing too many or too few lands). While players may still find themselves in these scenarios, the shuffler changes are intended to mitigate the rarest scenarios, such as drawing 8+ Land cards in a row. As noted above, we are only testing these changes in the "Play" queue, and it does not apply to any Ranked or Traditional formats. We are using this opportunity to gather more data and to fine tune these changes, as well as allow for player feedback. All shuffles are still randomly generated, The difference is we now look deeper into the decks to determine a pool of shuffles to randomly choose from. We are planning to iterate on this fairly rapidly. and will provide more information as these changes develop and solidify.
For all best of one play queues, we have increased the number of deck shuffles and starting hands we consider to three (up from two).
For all best of one play queues, we now apply the starting hand approach to mulligan hands as well.
More Information on Starting Hands: Our goal with this change is to ensure you're still on "even shuffler ground" with your starting hand, even when you mulligan. As with the other shuffler change, we are using this opportunity to gather more data and to fine tune these changes, as well as gather player feedback.
Alternate Land Art
Feb 15, 2019
Currently, the only way to select which basic land art to use is to import the deck, selecting the set and number of the basic land you want to use.
I can't speak to specific plans, but we've definitely heard clearly from players that you'd like a better way to accomplish this.
Shuffler
Jan 15, 2019
1. Please play nice here people.
2. When players report that true variance in the shuffler doesn't feel correct they aren't wrong. This is more than just a math problem, overcoming all of our inherent biases around how variance should work is incredibly difficult. However, while the feels say somethings wrong, all the math has supported everything is correct.
3. The shuffler and coin flips treat everyone equally. There are no systems in place to adjust either per player.
4. The only system in place right now to stray from a single randomized shuffler is the bo1 opening hand system, but even there the choice is between two fully randomized decks.
5. When we do a shuffle we shuffle the full deck, the card you draw is already known on the backend. It is not generated at the time you draw it.
6. Digital Shufflers are a long solved problem, we're not breaking any new ground here. If you paper experience differs significantly from digital the most logical conclusion is you're not shuffling correctly. Many posts in this thread show this to be true. You need at least 7 riffle shuffles to get to random in paper. This does not mean that playing randomized decks in paper feels better. If your playgroup is fine with playing semi-randomized decks because it feels better than go nuts! Just don't try it at an official event.
7. At this point in the Open Beta we've had billions of shuffles over hundreds of millions of games. These are massive data sets which show us everything is working correctly. Even so, there are going to be some people who have landed in the far ends of the bell curve of probability. It's why we've had people lose the coin flip 26 times in a row and we've had people win it 26 times in a row. It's why people have draw many many creatures in a row or many many lands in a row. When you look at the math, the size of players taking issue with the shuffler is actually far smaller that one would expect. Each player is sharing their own experience, and if they're an outlier I'm not surprised they think the system is rigged.
8. We're looking at possible ways to snip off the ends of the bell curve while still maintaining the sanctity of the game, and this is a very very hard problem. The irony is not lost on us that to fix perception of the shuffler we'd need to put systems in place around it, when that's what players are saying we're doing now.
Since people are asking, and it's no great secret. To shuffle decks in MTG Arena we use Fisher-Yates, pulling numbers from a Merseene Twister (MT199937), which is seeded with 256 cryptographically secure randomized bits. We use the same approach for coin tosses, only we're looking for a 1 or a 2 rather than a whole deck of cards.
In the new world, we start the match-making process by placing players into buckets based on their rank. Tiers don't matter here, just the rank you're at (Bronze, Silver, Etc). You can think of this as a progression of difficulty that you also see in tabletop Magic: from Kitchen Table up through your LGS, to PTQ, to the Pro-Tour. We want MTG Arena to serve all of these tiers of skill, and this is the way we believe best addresses the climb. By bucketing by rank we give players a chance to improve over time, rather than forcing them to start at potentially a pro-tour level of play.
After we group players together by rank we then sort them based on their W/L record. As far as I can tell no one is worried about this.
The final metric we look at is MMR. And to be perfectly clear: our matchmaking rating does not force players to a 50% win rate. Stronger players will have a higher win-rate in our system. It is a loose check to see if the two players are within a certain skill range that we deliberately set to be large enough to not require an "equal match". Do great in DOM draft, but then suck it up hard in XLN/RIX and this will pair you with other people in the same boat. We believe this is a fair system where everyone will still have to earn their wins.
All of these metrics will also expand out based on time in the queue. There will be matches across ranks in some cases, just as at times there are matches with different win/loss records and distant MMRs.
All of this said, if you believe matchmaking in Limited should always be Swiss, then it's unlikely I've said anything to sway your opinion. If you want to go toe-to-toe with any Magic player in the world, we have Traditional Draft as the place for you to show your skill without climbing up the Ranks. Traditional Draft remains solely based on W/L record. As always we'll be watching how this plays out in reality, as we've only been able to do sims to this point, and continue to make adjustments.
[I]f you ever wait more than 2-3 minutes something has likely gone wrong. Right now the longest match time we've seen is 100s. There is a chance as the very first people hit a rank they may have to wait longer, but this should even out.
In looking over everything I realized we didn't mention in the primary post that we now have a window of 60s during which you will be unable to match against the person you just played. We'll watch how this current window works to prevent repeat matches, while still keeping queue times brisk and adjust as needed.
In this system when you switch decks it will be looking for similarly skilled players to you, that happened to be playing their own decks around your level.
I can say that when we're assigning deck weights we look at the strength of the card far more than the rarity. Rarity, in general, is an indicator but there are many rares/mythics that have lower values than an uncommon like lightning strike for example.
There are no plans to show any information on deck strength to players. The goal is for players to build the deck they want to play and find a match that's appropriate. Getting more stats into the Profile tab is something we'd like to see.
My Account was Hacked!
Jan 10, 2019
GDPR changed the way information is collected, and how we use and store PII(Personally Identifiable Information). Initially we collect as little as possible as in the new world we only collect what we absolutely need, and what we collect is usually enough to restore a lost account. Once you've made purchases we have more PII on your account, which we can use to restore it if it was lost for some reason. One of the myriad of reasons we have no trading is it helps make all the work of hacking accounts just not worth it as well.
Duplicate Protection
Jan 15, 2019
No plans to expanded duplicate protection out to uncommons or commons at this time. We've said from the beginning that adding in duplicate protection was going to require some balance changes, we're pretty happy where this has settled. Time, data, and players will really tell us what needs to adjust further.
Hey hrvl are you an employee of wizards? When you responded "we" I wondered.
Here's the way to organize the website/articles:
PLEASE
Give us a homepage for every format. Doesn't have to be much, but you can explain the rules et. at least. This isn't the most important.
If not that, give us a page on every format page you already have with a link to EVERY article regarding that format in date format order. I can't tell you how many times I wish I could just find the updates for Arena specifically, from the Arena homepage.
As you can see, I'm literally posting on a forum to find this stuff. I just spent multiple minutes trying to find this info. on wizard's website for a friend and just said "Look on this link right here" to finish it off.
He linked me to mtg arena's official forum for the info. eventually. People shouldn't have to search forums to find this stuff.
I'm not a Wizards employee, I just forgot to add links! Except for the evil shuffler link, all those are posts by Arena developers, mostly Chris Clay (the evil shuffler guy is even a developer now).
I agree with you about how hard it is to find info, I've just given up expecting it from playing Magic Online. Last year they made a statement about needing to be more transparent and organized on some unused tumblr at the end of a post about something completely different. They have pages they're legally required to keep updated that are often just totally wrong. On Arena, this info isn't only on a forum, it's sometimes buried hundreds of pages into a thread, and the dev tracker isn't that helpful because many of the posts aren't from developers or about development.
Had to check the fb group for my lgs for the pre-release date because it wasn't wizard's website under the "coming soon" section under the "war of the Spark" banner. Also not in their articles either.
Latest Patch Notes
Feb 14, 2019
0.12.00.00 - Major
https://forums.mtgarena.com/forums/threads/46580
-Shuffler changes for "Play."
Enabled "Smooth Shuffling" for the Play queue.
Important Note: This is for the "Play" queue only. It does not apply to Ranked Traditional Play, Traditional Ranked, or any other events and/or formats.
More Information on Smooth Shuffling: We are currently testing changes to the shuffler algorithm to decrease the number of games with extreme examples of "mana flood" or "mana screw" (drawing too many or too few lands). While players may still find themselves in these scenarios, the shuffler changes are intended to mitigate the rarest scenarios, such as drawing 8+ Land cards in a row. As noted above, we are only testing these changes in the "Play" queue, and it does not apply to any Ranked or Traditional formats. We are using this opportunity to gather more data and to fine tune these changes, as well as allow for player feedback. All shuffles are still randomly generated, The difference is we now look deeper into the decks to determine a pool of shuffles to randomly choose from. We are planning to iterate on this fairly rapidly. and will provide more information as these changes develop and solidify.
For all best of one play queues, we have increased the number of deck shuffles and starting hands we consider to three (up from two).
For all best of one play queues, we now apply the starting hand approach to mulligan hands as well.
More Information on Starting Hands: Our goal with this change is to ensure you're still on "even shuffler ground" with your starting hand, even when you mulligan. As with the other shuffler change, we are using this opportunity to gather more data and to fine tune these changes, as well as gather player feedback.
Feb 20, 2019
0.12.01.00 - Minor
https://forums.mtgarena.com/forums/threads/52931
-Fixed long matchmaking times
-Fixed frequent disconnections, being monitored
Alternate Land Art
Feb 15, 2019
Currently, the only way to select which basic land art to use is to import the deck, selecting the set and number of the basic land you want to use.
I can't speak to specific plans, but we've definitely heard clearly from players that you'd like a better way to accomplish this.
Deck Builder: Advanced Search Terms
Feb 14, 2019
https://forums.mtgarena.com/forums/threads/46567
Debunking the Evil Shuffler -Unofficial
Aug 6, 2018
https://blog.mtgatracker.com/debunking-the-evil-shuffler
Shuffler
Jan 15, 2019
1. Please play nice here people.
2. When players report that true variance in the shuffler doesn't feel correct they aren't wrong. This is more than just a math problem, overcoming all of our inherent biases around how variance should work is incredibly difficult. However, while the feels say somethings wrong, all the math has supported everything is correct.
3. The shuffler and coin flips treat everyone equally. There are no systems in place to adjust either per player.
4. The only system in place right now to stray from a single randomized shuffler is the bo1 opening hand system, but even there the choice is between two fully randomized decks.
5. When we do a shuffle we shuffle the full deck, the card you draw is already known on the backend. It is not generated at the time you draw it.
6. Digital Shufflers are a long solved problem, we're not breaking any new ground here. If you paper experience differs significantly from digital the most logical conclusion is you're not shuffling correctly. Many posts in this thread show this to be true. You need at least 7 riffle shuffles to get to random in paper. This does not mean that playing randomized decks in paper feels better. If your playgroup is fine with playing semi-randomized decks because it feels better than go nuts! Just don't try it at an official event.
7. At this point in the Open Beta we've had billions of shuffles over hundreds of millions of games. These are massive data sets which show us everything is working correctly. Even so, there are going to be some people who have landed in the far ends of the bell curve of probability. It's why we've had people lose the coin flip 26 times in a row and we've had people win it 26 times in a row. It's why people have draw many many creatures in a row or many many lands in a row. When you look at the math, the size of players taking issue with the shuffler is actually far smaller that one would expect. Each player is sharing their own experience, and if they're an outlier I'm not surprised they think the system is rigged.
8. We're looking at possible ways to snip off the ends of the bell curve while still maintaining the sanctity of the game, and this is a very very hard problem. The irony is not lost on us that to fix perception of the shuffler we'd need to put systems in place around it, when that's what players are saying we're doing now.
Since people are asking, and it's no great secret. To shuffle decks in MTG Arena we use Fisher-Yates, pulling numbers from a Merseene Twister (MT199937), which is seeded with 256 cryptographically secure randomized bits. We use the same approach for coin tosses, only we're looking for a 1 or a 2 rather than a whole deck of cards.
Matchmaking
Dec 14, 2018
https://forums.mtgarena.com/forums/threads/43383
In the new world, we start the match-making process by placing players into buckets based on their rank. Tiers don't matter here, just the rank you're at (Bronze, Silver, Etc). You can think of this as a progression of difficulty that you also see in tabletop Magic: from Kitchen Table up through your LGS, to PTQ, to the Pro-Tour. We want MTG Arena to serve all of these tiers of skill, and this is the way we believe best addresses the climb. By bucketing by rank we give players a chance to improve over time, rather than forcing them to start at potentially a pro-tour level of play.
After we group players together by rank we then sort them based on their W/L record. As far as I can tell no one is worried about this.
The final metric we look at is MMR. And to be perfectly clear: our matchmaking rating does not force players to a 50% win rate. Stronger players will have a higher win-rate in our system. It is a loose check to see if the two players are within a certain skill range that we deliberately set to be large enough to not require an "equal match". Do great in DOM draft, but then suck it up hard in XLN/RIX and this will pair you with other people in the same boat. We believe this is a fair system where everyone will still have to earn their wins.
All of these metrics will also expand out based on time in the queue. There will be matches across ranks in some cases, just as at times there are matches with different win/loss records and distant MMRs.
All of this said, if you believe matchmaking in Limited should always be Swiss, then it's unlikely I've said anything to sway your opinion. If you want to go toe-to-toe with any Magic player in the world, we have Traditional Draft as the place for you to show your skill without climbing up the Ranks. Traditional Draft remains solely based on W/L record. As always we'll be watching how this plays out in reality, as we've only been able to do sims to this point, and continue to make adjustments.
[I]f you ever wait more than 2-3 minutes something has likely gone wrong. Right now the longest match time we've seen is 100s. There is a chance as the very first people hit a rank they may have to wait longer, but this should even out.
In looking over everything I realized we didn't mention in the primary post that we now have a window of 60s during which you will be unable to match against the person you just played. We'll watch how this current window works to prevent repeat matches, while still keeping queue times brisk and adjust as needed.
In this system when you switch decks it will be looking for similarly skilled players to you, that happened to be playing their own decks around your level.
I can say that when we're assigning deck weights we look at the strength of the card far more than the rarity. Rarity, in general, is an indicator but there are many rares/mythics that have lower values than an uncommon like lightning strike for example.
There are no plans to show any information on deck strength to players. The goal is for players to build the deck they want to play and find a match that's appropriate. Getting more stats into the Profile tab is something we'd like to see.
My Account was Hacked!
Jan 10, 2019
GDPR changed the way information is collected, and how we use and store PII(Personally Identifiable Information). Initially we collect as little as possible as in the new world we only collect what we absolutely need, and what we collect is usually enough to restore a lost account. Once you've made purchases we have more PII on your account, which we can use to restore it if it was lost for some reason. One of the myriad of reasons we have no trading is it helps make all the work of hacking accounts just not worth it as well.
Duplicate Protection
Jan 15, 2019
No plans to expanded duplicate protection out to uncommons or commons at this time. We've said from the beginning that adding in duplicate protection was going to require some balance changes, we're pretty happy where this has settled. Time, data, and players will really tell us what needs to adjust further.
January State of the Beta
Duplicate Protection (Overview)
https://forums.mtgarena.com/forums/threads/44955
Event Reward Changes
https://forums.mtgarena.com/forums/threads/44953
Preseason 2
https://forums.mtgarena.com/forums/threads/46022
Here's the way to organize the website/articles:
PLEASE
Give us a homepage for every format. Doesn't have to be much, but you can explain the rules et. at least. This isn't the most important.
If not that, give us a page on every format page you already have with a link to EVERY article regarding that format in date format order. I can't tell you how many times I wish I could just find the updates for Arena specifically, from the Arena homepage.
As you can see, I'm literally posting on a forum to find this stuff. I just spent multiple minutes trying to find this info. on wizard's website for a friend and just said "Look on this link right here" to finish it off.
He linked me to mtg arena's official forum for the info. eventually. People shouldn't have to search forums to find this stuff.
I agree with you about how hard it is to find info, I've just given up expecting it from playing Magic Online. Last year they made a statement about needing to be more transparent and organized on some unused tumblr at the end of a post about something completely different. They have pages they're legally required to keep updated that are often just totally wrong. On Arena, this info isn't only on a forum, it's sometimes buried hundreds of pages into a thread, and the dev tracker isn't that helpful because many of the posts aren't from developers or about development.