PDA

View Full Version : Dual Color Draw Test! (test results only)



dwebber88
11-01-2013, 08:37 AM
Others have tested (http://forums.cryptozoic.com/showthread.php?t=29508) drawing resources. I would like to do a test that can be analyzed a little bit better.

from my observations i've seen that actually drawing 1/1 1/2 or 2/2 resources is significantly harder than drawing 2, 3 or 4 resources of the same color (0/2 0/3 or 0/4).

Testing decks must have 12/12 resources in the deck!

Please enter your test results in this format:

7 card draws - <number starting hands>
0/0 - <number>
0/1 - <number>
0/2 - <number>
0/3 - <number>
0/4+ - <number>

1/1 - <number>
2/1 - <number>
2/2 - <number>
3/2 - <number>
3/3+ - <number>

6 card draws - <number of mulligan to 6>
0/0 - <number>
0/1 - <number>
0/2 - <number>
0/3 - <number>
0/4+ - <number>

1/1 - <number>
2/1 - <number>
2/2 - <number>
3/2 - <number>
3/3 - <number>

jtatta
11-01-2013, 01:26 PM
I really need moderator status on these forums so I can ban winners like sagar. I think you must be the reincarnation of EmraldArcher. Webber is trying to help out and you're just being toxic. Really not necessary, dude.

KiraForce
11-01-2013, 01:37 PM
Also, just saying, but there was another thread that almost entirely proved that something may be wrong with their randomizing algorithm. If I had any time, I would help you out, but I have work and school...

nicosharp
11-01-2013, 01:53 PM
I tend to agree with Sagar here. If there is a problem, hopefully the data collection CZE is doing from Hex games being played is pulling these numbers, and they can revise as needed.

When I was more forum gunho, I might have tried an endeavor like this out of boredom. Unless this becomes a competitive issue once competitive play is available, I'd let them handle it.

sagar52983
11-01-2013, 01:59 PM
I tend to agree with Sagar here. If there is a problem, hopefully the data collection CZE is doing from Hex games being played is pulling these numbers, and they can revise as needed.

When I was more forum gunho, I might have tried an endeavor like this out of boredom. Unless this becomes a competitive issue once competitive play is available, I'd let them handle it.

Thank you. You have presented my case in a more articulate manner :) (no I'm not being sarcastic)
I honestly think at this early in the stage of alpha, this is probably not something that they are even considering to look at in full dept. Stability and getting these sets out is probably more of a priority. I'm sure the more sets and cards they add...the more they'll have to look into the algorithms.
(sp?)

Also loving the guild Nico. A few friends and I were looking into it.

Niedar
11-01-2013, 02:13 PM
Also, just saying, but there was another thread that almost entirely proved that something may be wrong with their randomizing algorithm. If I had any time, I would help you out, but I have work and school...

No there has not been. Nothing has been prove to be statistically relevant.

Mr.Funsocks
11-01-2013, 11:52 PM
A "randomization algorithm" would be a little odd... You don't need an algorithm. Just assign each card a random number between 1 and 60, and put them in order...

Statistical significance on this would take a little while though. As in 200+ draws.

Kroan
11-02-2013, 01:36 AM
A "randomization algorithm" would be a little odd... You don't need an algorithm. Just assign each card a random number between 1 and 60, and put them in order... So how do you get that random number?

shocker455
11-02-2013, 02:41 AM
Assuming they didn't decide to create their own random number generater (which there is no reason 2), the randomness will be correct.

Soooo im sure it is fine.

-------

edit; Humans never seem to trust computers when it comes to randomness. I dont know why, i am sure there is some reason in psychology.

Maybe it has something to do with how people think randomness works and how it actually works. E.X. if you have 1 person flip a coin 100 times and record the results, and 1 person write h/t 100 times. you can tell from the list which one is fake, and which one is real with statistics.

Willd
11-02-2013, 04:12 AM
Assuming they didn't decide to create their own random number generater (which there is no reason 2), the randomness will be correct.

Soooo im sure it is fine.

-------

edit; Humans never seem to trust computers when it comes to randomness. I dont know why, i am sure there is some reason in psychology.

Maybe it has something to do with how people think randomness works and how it actually works. E.X. if you have 1 person flip a coin 100 times and record the results, and 1 person write h/t 100 times. you can tell from the list which one is fake, and which one is real with statistics.

Actually most pseudo random number generators don't have nearly enough possible starting states for every ordering of a deck to be possible. Most only use 32 bits but a deck of 60 cards requires > 256 bits to be able to produce every possible combination (this assumes all cards are unique, which obviously isn't normally the case but TCGs also potentially require shuffling of much larger decks). That means the shuffle wouldn't be truly fair. However it is still highly unlikely that the specific types of bias that people think they seeing would come about as a result of this shuffle.

Any shuffle that wants to be truly fair should use a truly random number generator based on environmental factors. This might not be necessary for a TCG if you can show that some amount of bias doesn't actually alter gameplay at all but is vital for things like online poker and online casinos.

dwebber88
11-02-2013, 04:45 AM
So much for test results only. Thanks a lot guys! I'll report my findings later, when i have enough data.

People saying there is another thread about this, just really didn't read my fipo. Great job!

franky
11-02-2013, 07:13 AM
Actually most pseudo random number generators don't have nearly enough possible starting states for every ordering of a deck to be possible. Most only use 32 bits but a deck of 60 cards requires > 256 bits to be able to produce every possible combination (this assumes all cards are unique, which obviously isn't normally the case but TCGs also potentially require shuffling of much larger decks). That means the shuffle wouldn't be truly fair.

That's assuming the shuffler works by first generating a list of every possible deck shuffling, and then picking a random one from that list. That seems extremely unlikely, as it would be terribly inefficient to have to calculate that list every time you shuffled your deck. It's much more likely that it works simply by numbering the cards, and then randomly selecting one card that goes on the bottom, another card that goes on top of that, and so on.

EDIT: Actully, it would be impossible. To generate such a list would take up more memory than any computer ever build has. Let's say each card gets a unique identifying number that takes one byte of memory. A 60 card deck therefore uses 60 bytes of memory. A list of every possible configuration of those 60 cards would consist of 60*59*58... and so on items, times 60 bytes per deck. That's almost 5*10^71 terabytes.

Mr.Funsocks
11-02-2013, 08:31 AM
edit; Humans never seem to trust computers when it comes to randomness. I dont know why, i am sure there is some reason in psychology.

Because most people don't understand "random". They think it means "always patternless". Fun experiment you can do: Have 2 people write down the results of 10 coin flips. One person makes up the coin flips, the other person actually flips a coin. Leave the room while they do this. Come back in, and you can almost always tell which is which. The actually random coinflips will have streaks of 2-3, the one someone made up will not.

Soldack
11-02-2013, 08:52 AM
So what were the results? I am interested in this. thanks for doing it OP.

mudkip
11-02-2013, 09:00 AM
So how do you get that random number?

Magic.

DeusPhasmatis
11-03-2013, 12:25 AM
Actually most pseudo random number generators don't have nearly enough possible starting states for every ordering of a deck to be possible. Most only use 32 bits but a deck of 60 cards requires > 256 bits to be able to produce every possible combination (this assumes all cards are unique, which obviously isn't normally the case but TCGs also potentially require shuffling of much larger decks). That means the shuffle wouldn't be truly fair. However it is still highly unlikely that the specific types of bias that people think they seeing would come about as a result of this shuffle.

There are order randomization algorithms that achieve random shuffles in both efficient run time and space complexity. Brute force isn't the only solution.

Bossett
11-03-2013, 01:52 AM
The shuffler could work very simply - just be constantly generating new random numbers server size, put it into a common stack and pop off the top x bytes whenever you have to make a random choice. For a deck shuffle, order the deck however you want (card #? alphabetical?) pop off a byte, take it modulo deck size and the result is the first card. Pop off another byte, modulo deck size - 1 and you have card number 2. Keep going.

Considering you don't actually need to generate shuffled decks all that often (once, twice maybe three times a game per player?) and that a shuffled deck order isn't actually all that large a number (it'll fit in, what, 60 bytes for a 255 card deck?), a high-quality, slow source of entropy isn't going to be significantly taxed.