|
7 months ago ::
Nov 06, 2012 - 8:33AM
#61
|
Date Joined:
Feb 26, 2003
|
Would you keep a hand with 6 Plains and a CoP: Black vs a mono-black deck? Not trying to be sarcastic. I think I would, but I don't consider myself smart enough to be confident in doing so.
I actually had a similar situation come up in real life. Playing at (if I remember correctly) States 2003, I was playing a match against a red burny deck (remember Blistering Firecat  ?) deck. It was game two of the match. I was playing Broodstar Affinity (the format was Standard and Mirrodin had just come out a month or so before). I had CoP: Red in the sideboard. My opening hand was six lands (almost all of them artifact lands, including a white one) and a CoP: Red. I kept and I won the game.
Of course, I went on to lose the match in game three, but that wasn't the fault of my keep in game two!
|
|
|
|
7 months ago ::
Nov 06, 2012 - 10:40AM
#62
|
Date Joined:
Jun 15, 2009
|
Yes, the bias for human shufflers to hit fewer land clumps than expected by chance while the computer shuffler to hit more land clumps than expected by chance has been well documented.
References? Well documented indeed...
Perhaps the former has been well documented (it certainly could be well documented because it is true), but the latter is definitely NOT well documented, nor could it be because it is simply not true.
I should certainly hope you don't expect us to treat this thread as such documentation...
I've documented it, whether people choose to acknowledge my documentation that land|land pairs come up more often than would be expected by chance. Another way to look at the results I saw online, which occurred over thousands of draws, is that drawing a land becomes more likely in your next draw provided that your current draw is a land. While by chance one would expect it to become less likely, since there is one less land to draw out of the deck. This bias towards lands following lands was consistant across two trials of over 1,000 samples. Drawing a land in the current draw online makes you 1.85% more likely to draw a land in the following draw, while by chance you would expect it to make you 2.56% less likely to draw a land in the following draw.
Lands following lands and spells following spells is exactly what leads to land and spell clumps.
|
|
|
|
7 months ago ::
Nov 06, 2012 - 12:16PM
#63
|
Date Joined:
Mar 20, 2003
|
Lol, bobus, you're really something. "It's well documented" implies a hell of a lot more than your own arm-chair statistics based on a prejudice. Imagine if I'd written my thesis like that. References, professor? Why, I wrote the reference material myself! Look how well documented everything is... I wrote that documentation! Yes, yes, reviewed by peers, ie me, what better authority to approve my work! 
|
|
|
|
7 months ago ::
Nov 06, 2012 - 1:10PM
#64
|
Date Joined:
Oct 27, 2012
|
From the distribution of ELO's you can perhaps glean (at least approximations of):
1) How deep/difficult a game is. For this, look at the spacing of the average ELO, the top quartile ELO, the top decile ELO, etc. If the distribution of ELO's is bunched near the center and has a long tail compared to another game (e.g. the ratio of the expected win percentage of top decile playing top quartile to the expected win percentage of top quartile playing average is large, and/or similar other ratios) we could say the game is deeper, i.e. there more room for meaningful improvement (a "higher ceiling") for a player.
Is tic-tac-toe a deeper or more difficult game than Magic? Here is the solution to it: xkcd.com/832/ Certainly "good" players will never lose a tic-tac-toe match against "bad" players, but what does that actually tell us about the game?
At tic-tac-toe, the top quartile player (player ranked, after adjusting for ties, above 75% of the population) is the same as the top decile player (player ranked, after adjusting for ties, above 90% of the population). There's no "fat tail"; there's no "tail" at all to the distribution of ELO's for tic-tac-toe. From this we can see it's not a very deep game.
Having a "fat tail" would essentially mean that the 90% player is substantially better than the 75% player, and a 95% player is substantially better than a 90% player, and so on.
You may ask what "substantially" means. This depends on the game but can also be read off of the distribution of ELO's (this is the "randomness" aspect, #2 in my list) as any sort of average or expected difference in ELO between two players.
|
|
|
|
7 months ago ::
Nov 06, 2012 - 4:07PM
#65
|
Date Joined:
Jun 15, 2009
|
Lol, bobus, you're really something. "It's well documented" implies a hell of a lot more than your own arm-chair statistics based on a prejudice.
Imagine if I'd written my thesis like that. References, professor? Why, I wrote the reference material myself! Look how well documented everything is... I wrote that documentation! Yes, yes, reviewed by peers, ie me, what better authority to approve my work!
It was a very straightforward process to gather and interpret the data. I wrote down drawn cards as strings of 1s and 0s as lands and non-lands respectively. Then I wrote a spreadsheet to count the number of 00's, 01's, 10's and 11's substrings in those strings. Then I looked at the number of times each substring came up, and compared it to the number of times each substring was expected to come up by chance. 1000+ card pairs were collected and analyzed, and then the process was repeated, collecting another 1000+ pairs. The results from both trials were the same, the land-land pair comes up considerably more often than one would expect by chance and because of this, drawing a land with any given draw makes it more likely the subsequent draw will be a land as well. These results are contrary to the results one would expect by chance, where drawing a land in any given draw makes it less likely that a land will also be drawn in the following draw.
|
|
|
|
7 months ago ::
Nov 06, 2012 - 7:41PM
#66
|
Date Joined:
Jul 20, 2007
|
Lol, bobus, you're really something. "It's well documented" implies a hell of a lot more than your own arm-chair statistics based on a prejudice.
Imagine if I'd written my thesis like that. References, professor? Why, I wrote the reference material myself! Look how well documented everything is... I wrote that documentation! Yes, yes, reviewed by peers, ie me, what better authority to approve my work!
It was a very straightforward process to gather and interpret the data. I wrote down drawn cards as strings of 1s and 0s as lands and non-lands respectively. Then I wrote a spreadsheet to count the number of 00's, 01's, 10's and 11's substrings in those strings. Then I looked at the number of times each substring came up, and compared it to the number of times each substring was expected to come up by chance. 1000+ card pairs were collected and analyzed, and then the process was repeated, collecting another 1000+ pairs. The results from both trials were the same, the land-land pair comes up considerably more often than one would expect by chance and because of this, drawing a land with any given draw makes it more likely the subsequent draw will be a land as well. These results are contrary to the results one would expect by chance, where drawing a land in any given draw makes it less likely that a land will also be drawn in the following draw.
Nushae is not questioning your honesty or dedication but your grasp on reality. Or so it seems to me from reading his somewhat hillarious post. If you say something is "well documented" it typically means "by unimpeachable disinterested third parties." So while we can see from your many many posts on the very subject that you have a real thing for the shuffler and its terrible hate for you, this has not swayed us from looking for those unbiased sources to back up your somewhat outrageous claims.
Winter.Wolf
|
|
|
|
7 months ago ::
Nov 06, 2012 - 10:37PM
#67
|
Date Joined:
May 28, 2003
|
From the distribution of ELO's you can perhaps glean (at least approximations of):
1) How deep/difficult a game is. For this, look at the spacing of the average ELO, the top quartile ELO, the top decile ELO, etc. If the distribution of ELO's is bunched near the center and has a long tail compared to another game (e.g. the ratio of the expected win percentage of top decile playing top quartile to the expected win percentage of top quartile playing average is large, and/or similar other ratios) we could say the game is deeper, i.e. there more room for meaningful improvement (a "higher ceiling") for a player.
Is tic-tac-toe a deeper or more difficult game than Magic? Here is the solution to it: xkcd.com/832/ Certainly "good" players will never lose a tic-tac-toe match against "bad" players, but what does that actually tell us about the game?
At tic-tac-toe, the top quartile player (player ranked, after adjusting for ties, above 75% of the population) is the same as the top decile player (player ranked, after adjusting for ties, above 90% of the population). There's no "fat tail"; there's no "tail" at all to the distribution of ELO's for tic-tac-toe. From this we can see it's not a very deep game.
Having a "fat tail" would essentially mean that the 90% player is substantially better than the 75% player, and a 95% player is substantially better than a 90% player, and so on.
You may ask what "substantially" means. This depends on the game but can also be read off of the distribution of ELO's (this is the "randomness" aspect, #2 in my list) as any sort of average or expected difference in ELO between two players.
I suppose I meant the question more philosophically. Tic-tac-toe has no random element to it - it is a game of "pure skill". Chess is also a game of "pure skill". Magic has a random element to it, so it is not "pure skill".
So can we make the claim that MTGO has "more skill" than tic-tac-toe, a game of "pure skill"? I certainly can - but not from looking at the results of a series of matches (even hypothetically). Instead:
In order to play tic-tac-toe at a master level, you need only two skills:
1) The ability to read, comprehend, and follow rules.
2) The ability to follow links on the internet (i.e. my link above)
Both of these skills are also necessary to play MTGO - you need to find the download link and you need to know how to actually click on cards to play them. However, while having just these two skills can make you a tic-tac-toe master, if you possess no additional skills you will lose the vast majority of your MTGO games. Therefore, MTGO requires more skill than tic-tac-toe.
Now I don't doubt we could do some exhaustive studies and conclusively prove results such as "randomness plays a significant role in the outcome of a match in MTGO", and even bobus's grail "land issues will cause you to lose games", but so what? What point do we prove? What do we learn? What can we honestly say about the game besides those (blatantly obvious) conclusions?
"I would win more if there was less randomness" seems to be the desired inference, however it would not actually follow from any line of thought proposed. For example, regardless of the results of any statistical analysis, can you* defeat the argument "The ability to use resources effectively is one of the core skills of MTGO - by removing mana issues you reduce the importance of this skill, and thus reduce the ability to differentiate a good player from a bad one, resulting in an even tighter distribution of wins"?
*"You" meaning anyone, not "you" cten.
|
|
|
|
7 months ago ::
Nov 07, 2012 - 5:00AM
#68
|
Date Joined:
Mar 20, 2003
|
Nushae is not questioning your honesty or dedication but your grasp on reality.
That's more or less correct. I wouldn't have said "grasp on reality". Let me put it this way, I think bobus is caught in a kind of reverse duck typing: he's showing all the behavioral indicators of scientific rigour, including jargon, but that doesn't make him rigorous. I don't think he's doing that to mislead us, I think he is somewhat mislead himself.
|
|
|
|
7 months ago ::
Nov 07, 2012 - 5:07AM
#69
|
Date Joined:
Jun 15, 2009
|
Nushae is not questioning your honesty or dedication but your grasp on reality. Or so it seems to me from reading his somewhat hillarious post. If you say something is "well documented" it typically means "by unimpeachable disinterested third parties." So while we can see from your many many posts on the very subject that you have a real thing for the shuffler and its terrible hate for you, this has not swayed us from looking for those unbiased sources to back up your somewhat outrageous claims.
Lack of verification that the shuffler is working as intended from an independant third party source is one of my primary issues with the MtGO shuffler. Most major online sites that deal with shuffled cards have such verification, most sites that never received such verification had serious, documented issues with their shufflers. My second issue is the sheer volume of unlikely card draws I see in online limited play. 9 land draws in a row, 11 spell draws in a row, and a 14 land, 7 spell split in the first 21 card draws after keeping a 3 land hand have all happened in the last 100 MtGO games I've played. The frequency of these unlikely draws prompted me to collect data regarding card|card correlations, the data showed that land|land pairs came up much more often than would be expected by chance. I repeated this data collection and found the same results during the second trial.
So after all of this we have for online limited (draft):
1) A documented land|land correlation bias
2) No independant verification there is no land|land correlation bias
|
|
|
|
7 months ago ::
Nov 07, 2012 - 5:38AM
#70
|
|
|
Nushae is not questioning your honesty or dedication but your grasp on reality. Or so it seems to me from reading his somewhat hillarious post. If you say something is "well documented" it typically means "by unimpeachable disinterested third parties." So while we can see from your many many posts on the very subject that you have a real thing for the shuffler and its terrible hate for you, this has not swayed us from looking for those unbiased sources to back up your somewhat outrageous claims.
Lack of verification that the shuffler is working as intended from an independant third party source is one of my primary issues with the MtGO shuffler. Most major online sites that deal with shuffled cards have such verification, most sites that never received such verification had serious, documented issues with their shufflers. My second issue is the sheer volume of unlikely card draws I see in online limited play. 9 land draws in a row, 13 spell draws in a row, and a 14 land, 7 spell split in the first 21 card draws after keeping a 3 land hand have all happened in the last 100 MtGO games I've played. The frequency of these unlikely draws prompted me to collect data regarding card|card correlations, the data showed that land|land pairs came up much more often than would be expected by chance. I repeated this data collection and found the same results during the second trial.
So after all of this we have for online limited (draft):
1) A documented land|land correlation bias
2) No independant verification there is no land|land correlation bias
Dude, seriously? You continue to act like your test is the only one and pretend that the dozens of other findings from people who were just as earnest as you in collecting and analyzing data (mainly in the shuffler thread) don't exist. Their work was at least as scientific if not much more than yours in most cases. I used to admire your persistence even if i disagreed with your findings, but at this point it's just reached ridiculous proportions. You aren't some martyr here who valiantly continues your crusade for what's right despite unjust persecution; rather you just appear more and more unreasonable and delusional and are completely disregarded by those who know better. Many don't even bother to rebut you anymore because you are so clearly disconnected from reality on this and you do a decent job of proving it on your own.
Magic is a game with an inherent and significant element of chance/luck. That's the way it is. Take your claimed vast skill at games elsewhere to a non-chance game like chess if you don't like it. You starting a new thinly-veiled shuffler thread every few weeks is getting very old and it's sad that some newcomers engage with you because they don't know better. I for one am getting tired of it being allowed.
|
|
|
