Dice

[Agent 86]

It just somehow manages to have more 2's and 4's rolled for everyone on the site somehow..randomly pulled from the atmosphere

[degaston]

I don't know.. I do know you can't have a truly random number generator for CC games if you also want to implement something to make the players average dice rolls to be 3.5 .. that by design defeats the purpose of RNG. Without the feature to equalize rolling averages, you'd have random average dice rolls for each player. So now you have random.org using a mathematical equation to interpret atmoshperic noise (which has to be predictable) and a CC equation to manipulate those results to keep the players even. I'm just guessing at all this..

I'm sorry, but that post is almost entirely incorrect. True random dice will naturally average out close to 3.5 over the long run without any intervention. Some players will be slightly above that, and some slightly below, but the more dice you roll, the closer your average should be to 3.5. Anything else would indicate that the numbers were not uniformly distributed. (As shown by the fact that almost everyone here with a lot of dice rolls has an average of 3.51) Though it would be very suspicious and entirely unnatural if everyone's average remained at exactly 3.500000. That would be like flipping a coin and getting HTHTHTHT... forever. It would not be very interesting for a game, either.

Random.org is highly regarded as a source of true random data, and CC would have no reason to try to manipulate it. Here's an easy to see example of the difference between true random and pseudo-random: http://boallen.com/random-numbers.html

[Metsfanmax]

Random.org is highly regarded as a source of true random data, and CC would have no reason to try to manipulate it. Here's an easy to see example of the difference between true random and pseudo-random: http://boallen.com/random-numbers.html

There are non-trivial reasons to want to manipulate it. In particular, to soften streaks. It might be like the iTunes effect where they coded it so that you were less likely to get two songs from the same album in a row than what random chance would dictate. We don't do this, but it's an interesting idea.

[degaston]

There are non-trivial reasons to want to manipulate it. In particular, to soften streaks. It might be like the iTunes effect where they coded it so that you were less likely to get two songs from the same album in a row than what random chance would dictate. We don't do this, but it's an interesting idea.

I analyzed some random.org data for streaks, and the first set of 10,000 rolls I looked at contained a streak of 7 digits, which should only happen once out of 46,656 rolls. But the more data I added, the more things evened out, and eventually, everything was right about where it should be. I know that an individual's dice are not taken from the data sequentially, so longer streaks than what appear in the data are possible, but no more likely.

I don't think they should make any attempt to even out the streaks - that would only lead to more suspicion that something was fixed. But I do wish they would fix the bias so that the only valid response to a bad dice thread is that the dice are truly random and these things happen. Perhaps they could publish the sets of data after they have been used and replaced, so that we could figure out where all those missing 1's have gone.

[ntcbadabing]

degaston... if it was truly random there would have to be a higher variance between players average rolls. The entire site's rolls might average pretty close to 3.5, but it should not work out to anyone with (random number pick here) say 100 games or more gets a 3.5 or 3.51 average. That would be statistically extremely unlikely. The wider the range of samples should result in a range of average dice rolls and all the results together should average out to 3.5. Instead, we have every sample averaging to 3.5 or 3.51, which is as unlikely, if not more, than 2's and 4's being significantly more probable. There's manipulation somewhere, but that's because, and back to my original statement, a true RNG is pretty much impossible.

[degaston]

degaston... if it was truly random there would have to be a higher variance between players average rolls. The entire site's rolls might average pretty close to 3.5, but it should not work out to anyone with (random number pick here) say 100 games or more gets a 3.5 or 3.51 average. That would be statistically extremely unlikely. The wider the range of samples should result in a range of average dice rolls and all the results together should average out to 3.5. Instead, we have every sample averaging to 3.5 or 3.51, which is as unlikely, if not more, than 2's and 4's being significantly more probable. There's manipulation somewhere, but that's because, and back to my original statement, a true RNG is pretty much impossible.

Again, I'm sorry, but I don't think you have a very good understanding of what you are saying, and I can't find much logic or evidence in your statements. I think you're probably using an intuitive understanding of randomness and probability as opposed to a mathematical one.

Here are a couple of graphs of dice statistics:

On the left are players with 100 games, 5,000-30,000 rolls, and an average between 3.494 and 3.550.
On the right are players with 1000 games, 230,000-490,000 rolls, and an average between 3.504 and 3.509.
By the definition of uniform distribution, the more samples you use, the closer the averages should get 3.5. It is not only likely, it is necessary.

You can see that variance decreases as you increase the number of samples. If the dice were uniformly distributed, then the graph on the right would show all dice rolls very close to 16.66% (They would all look similar to 3 on the right).

Please cite some reliable evidence for why you think that true random number generation is impossible.

[ntcbadabing]

"I don't think you have a very good understanding of what you are saying" - guilty as charged. I have no background in mathematical or statistical analysis, it just interests me which is why I posted here. No need to apologize.. The idea of a true RNG being next to impossible is from all the science channel shows I watch. I can't cite the exact show or who said it, but it would have been one of the scientists you see on those kinds of shows. The same show also pointed to all the patterns you find in nature as a supporting fact of how difficult true randomness is. It really just doesn't exist in our universe, according to the show. To think we've developed a truly random RNG using mathematical equations designed to manipulate data is a fundamentally flawed concept, in my mind anyway. It does seem odd to me that you use the term 'necessary' that true randomness should result in uniform distribution (if I understand you correctly). It seems to me that by definition of random, you shouldn't expect uniform distribution. Perhaps our 'true' RNG's are designed to produce results consistent with the average, which would in turn make the RNG not random. Or maybe I have no idea what I'm talking about

[degaston]

"I don't think you have a very good understanding of what you are saying" - guilty as charged. I have no background in mathematical or statistical analysis, it just interests me which is why I posted here. No need to apologize.. The idea of a true RNG being next to impossible is from all the science channel shows I watch. I can't cite the exact show or who said it, but it would have been one of the scientists you see on those kinds of shows. The same show also pointed to all the patterns you find in nature as a supporting fact of how difficult true randomness is. It really just doesn't exist in our universe, according to the show.

I admit, it did not occur to me to take into account the philosophical views of Determinists who believe that randomness cannot exist in the universe because everything is causal. I might even be inclined to agree with them, but I can't find any practical application for that idea in real life, so I don't often think about it.

To think we've developed a truly random RNG using mathematical equations designed to manipulate data is a fundamentally flawed concept, in my mind anyway.

Yes, pseudo-random number generators tend to have a problem with predictability, but even if CC did use a PRNG, I don't think it would be that big an issue because the dice are not given out sequentially in large batches (or, at least, I think that's what happens). In any case, the Random.org website has a lot of information about true random number generation if you're interested.

It does seem odd to me that you use the term 'necessary' that true randomness should result in uniform distribution (if I understand you correctly). It seems to me that by definition of random, you shouldn't expect uniform distribution.

Actually, by definition, you should:

ran•dom (ˈræn dəm)
adj.
1. occurring or done without definite aim, reason, or pattern: random examples.
2. Statistics. of or characterizing a process of selection in which each item of a set has an equal probability of being chosen.

The whole point of a die is that each side should have an equal chance of coming up. If you find that one side consistently appears more or less than 1/6th of the time, then that is evidence that the die is not "fair".

Perhaps our 'true' RNG's are designed to produce results consistent with the average, which would in turn make the RNG not random. Or maybe I have no idea what I'm talking about

It appears that your ideas about what random numbers should and should not do are being colored by one or more cognitive biases. Lots of people have them regarding the probability of random events (see: Las Vegas), but if you're interested you might want to read more on the subject.

What applies here is the "law of large numbers" which states that "the average of the results obtained from a large number of trials should be close to the expected value, and will tend to become closer as more trials are performed." (http://en.wikipedia.org/wiki/Law_of_large_numbers) This is saying that the more rolls you have, the closer you should be to having each number come up 1/6th of the time. As expected, the CC dice do converge as you include more samples, but the fact that 1's, 2's and 4's do not converge towards 16.66% shows that there is a bias in the dice output.

[ntcbadabing]

Hmm.. it seems to me the 2 definitions of random conflict with each other. The first definition states without aim, reason, or pattern but the 2nd definition, when applied to statistics, sets out to put order to the process of selection. However, by definition, the 2 definitions conflict.. which could also be the result of a cognitive bias, the need to find order in something designed to be random. In the example of dice, each time the dice is rolled, there is a 1/6th chance of getting any specific number. By the first definition of random, this could result in all 5's as easily as it could and average of 3.5.. by the 2nd definition of random, all numbers should be rolled equally 1/6th of the time. Then if you apply mathematical formulas to predict outcomes, it becomes highly improbable to roll all 5's. It is correct to say the 2nd definition is more applicable in real life..

I'll read up a little more on it, thanks for the links!

[degaston]

...By the first definition of random, this could result in all 5's as easily as it could and average of 3.5

This is not true because the order of events matters when figuring out combinations, but not when calculating an average. If you roll a die 6 times, there are 46,656 possible combinations, each of them equally likely to occur. But while there is exactly one way to get all 5's, there are 720 ways to get one of each number (and many other combinations that would also average out to 3.5).

For a simpler example, think of flipping a coin four times (Heads=1, Tails=2). There are 16 possible outcomes: (1,1,1,1), (1,1,1,2), (1,1,2,1), etc. From these, there is 1 way to get an average of 1, 4 ways to get 1.25, 6 to get 1.5, 4 to get 1.75, and 1 way to get an average of 2. So you can see that getting an average of exactly 1.5 is 6 times more likely than flipping all tails.

... by the 2nd definition of random, all numbers should be rolled equally 1/6th of the time. Then if you apply mathematical formulas to predict outcomes, it becomes highly improbable to roll all 5's. It is correct to say the 2nd definition is more applicable in real life.

(1,4,6,4,2,5) is just as improbable as (5,5,5,5,5,5), but we would think nothing of the first sequence, and be very surprised by the second. It's just more cognitive bias.

As I see it, the definitions compliment each other rather than contradict. The first states that there should be no pattern to the data, the second, that there should be no bias favoring one outcome over another. Rolling 1,2,3,4,5,6,1,2,3,4,5,6, etc. meets the second definition, but not the first. CC dice meet the first definition, but not the second.

[chapcrap]

CC does not currently intentionally tamper with the results to alter the distribution of rolls.

I've been a mod for quite a while now and I see no evidence of tampering, but I can't say for certain that it wasn't done.

[degaston]

CC does not currently intentionally tamper with the results to alter the distribution of rolls.

I've been a mod for quite a while now and I see no evidence of tampering, but I can't say for certain that it wasn't done.

I'm not saying it's intentional, but the evidence is everywhere you look:

[ntcbadabing]

The way I see it, a 'true' RNG could be defined as either having absolutely no order by the first definition, but to meet the 2nd definition, you'd have to have order.. to quote you, it is necessary. Back to how all this relates to CC, due to we see a common trend in the 2's and 4's among all samples of different players results, the results are too consistent to match the first definition and not consistent enough to match the 2nd. To match the first, there should be a greater variance from one player's results to another. To match the 2nd definition, all the results for individual numbers amongst all players rolls should be equal. This would suggest manipulation, or at least something other than either definition of random.. or random.org is not as random as claimed, for either definition. Fun stuff to think about, I may have missed my calling, lol.

[degaston]

I'm not saying it's intentional...

...and I don't usually go in for conspiracy theories, but DAMN, how does it happen that the site owner and the "Head Thinker" have stats like this? Where are all your 2's?

[degaston]

The way I see it, a 'true' RNG could be defined as either having absolutely no order by the first definition, but to meet the 2nd definition, you'd have to have order.. to quote you, it is necessary.

I think you're veering back into philosophy and confusing "pattern" with "order". What does it mean to you (and how is is even possible) to have absolutely no order? The dice follow rules that limit what you will get to an integer from 1-6. You will never get a 7 or pi, so obviously there is some "order" in that each roll has a limited range of outcomes. All that is required to meet the first definition is for it to be impossible to know the next outcome ahead of time.

To match the first, there should be a greater variance from one player's results to another.

Go back and look at the law of large numbers. With a small number of rolls, there is plenty of variance between players, but as the number of rolls increases the results for everyone should converge.

To match the 2nd definition, all the results for individual numbers amongst all players rolls should be equal.

Go back and look at the law of large numbers. As the number of samples increases, the absolute difference between the count for individual numbers is also likely to increase, but not as fast as the total number of rolls. So as a percentage of the total rolls, they should all get closer to 16.66%.

[ntcbadabing]

I agree with everything you just said.. but as someone who does not have the understanding of statistics and how to calculate them as you do, to me, you're trying to apply the #2 definition to the #1. Your opinion is they compliment each other, but if you take the definitions literally, they completely contradict each other, like polar opposites. I understand the #1 definition pretty much only exists in philosophy.. and I was not even aware the #2 definition did exist before this conversation. It just seems odd to me that in the quest to build a true RNG, we look for consistencies and rules that can be predicted and explained in mathematical formulas rather than variations from them to support how random the RNG is. As far as absolutely no order, that's a difficult idea to even conceptualize much less produce, thus why a 'true' RNG is all but impossible to build. And just fyi, I'm not intending to be argumentative with you in this conversation, you obviously have a much greater understanding of this.. especially considering my knowledge of random has been increased by multiples in this conversation. I'm more thinking out loud and learning as I go. Thank you for the insight.

[Gweeedo]

"I don't think you have a very good understanding of what you are saying" - guilty as charged. I have no background in mathematical or statistical analysis, it just interests me which is why I posted here. No need to apologize.. The idea of a true RNG being next to impossible is from all the science channel shows I watch. I can't cite the exact show or who said it, but it would have been one of the scientists you see on those kinds of shows. The same show also pointed to all the patterns you find in nature as a supporting fact of how difficult true randomness is. It really just doesn't exist in our universe, according to the show.

I admit, it did not occur to me to take into account the philosophical views of Determinists who believe that randomness cannot exist in the universe because everything is causal. I might even be inclined to agree with them, but I can't find any practical application for that idea in real life, so I don't often think about it.

To think we've developed a truly random RNG using mathematical equations designed to manipulate data is a fundamentally flawed concept, in my mind anyway.

Yes, pseudo-random number generators tend to have a problem with predictability, but even if CC did use a PRNG, I don't think it would be that big an issue because the dice are not given out sequentially in large batches (or, at least, I think that's what happens). In any case, the Random.org website has a lot of information about true random number generation if you're interested.

It does seem odd to me that you use the term 'necessary' that true randomness should result in uniform distribution (if I understand you correctly). It seems to me that by definition of random, you shouldn't expect uniform distribution.

Actually, by definition, you should:

ran•dom (ˈræn dəm)
adj.
1. occurring or done without definite aim, reason, or pattern: random examples.
2. Statistics. of or characterizing a process of selection in which each item of a set has an equal probability of being chosen.

The whole point of a die is that each side should have an equal chance of coming up. If you find that one side consistently appears more or less than 1/6th of the time, then that is evidence that the die is not "fair".

Perhaps our 'true' RNG's are designed to produce results consistent with the average, which would in turn make the RNG not random. Or maybe I have no idea what I'm talking about

It appears that your ideas about what random numbers should and should not do are being colored by one or more cognitive biases. Lots of people have them regarding the probability of random events (see: Las Vegas), but if you're interested you might want to read more on the subject.

What applies here is the "law of large numbers" which states that "the average of the results obtained from a large number of trials should be close to the expected value, and will tend to become closer as more trials are performed." (http://en.wikipedia.org/wiki/Law_of_large_numbers) This is saying that the more rolls you have, the closer you should be to having each number come up 1/6th of the time. As expected, the CC dice do converge as you include more samples, but the fact that 1's, 2's and 4's do not converge towards 16.66% shows that there is a bias in the dice output.

[Jdsizzleslice]

I'm not saying it's intentional...

...and I don't usually go in for conspiracy theories, but DAMN, how does it happen that the site owner and the "Head Thinker" have stats like this? Where are all your 2's?

Ohhhhhhhh Lolololol......... Just do my dice stats.

[degaston]

Ohhhhhhhh Lolololol......... Just do my dice stats.

Yours are the last column of the big graph on page 2. They look like everyone else's - not many 1's, lots of 2's & 4's. That was from the "What number do you roll the most" thread.

[Jdsizzleslice]

Ohhhhhhhh Lolololol......... Just do my dice stats.

Yours are the last column of the big graph on page 2. They look like everyone else's - not many 1's, lots of 2's & 4's. That was from the "What number do you roll the most" thread.

I see a little below everyone else. That should throw any theory about moderators getting better dice out the window.

[degaston]

I see a little below everyone else. That should throw any theory about moderators getting better dice out the window.

Maybe you're just not high enough in the CC hierarchy to rate getting the "special dice"

[chapcrap]

I see a little below everyone else. That should throw any theory about moderators getting better dice out the window.

Maybe you're just not high enough in the CC hierarchy to rate getting the "special dice"

He just hasn't been a mod long enough. He's still in the probationary period.

[degaston]

I agree with everything you just said.. but as someone who does not have the understanding of statistics and how to calculate them as you do, to me, you're trying to apply the #2 definition to the #1. Your opinion is they compliment each other, but if you take the definitions literally, they completely contradict each other, like polar opposites.

I wasn't taking this as an argument of any kind, and I do not claim to be an expert on this. I have some background knowledge and I can fill in some of the blanks with a little research, which I find interesting.

I'm not sure I can help you with your idea that there is a contradiction in those definitions, because I don't see it. I take them at face value and as I think they are intended. When I roll a die, I think that each side should have an equal probability, and there should be no pattern in the numbers that come up. Where's the conflict in those statements?

Probability and statistics are just the math developed to explain and analyze what was already occurring in the natural world. When you're trying to simulate something that happens in the real world (like rolling a die) it's helpful to be able to test whether your results match what you could expect if you were rolling a real die.

[Trevor33]

I was beaten 1vs3 i think it was 13 times in a row once and around 10 on another occasion. I've played quite a few games with people throughout the years and never seen anything like that. Normally real dice are much better than the CC dice.

Now is it fair, as in it's the same system for everyone but they dice are predefined and streaky. That's why you see so many good and bad runs of dice, they're not random.

[osok68]

Sometimes you get in a downswing (just like in poker).Best remedy is to play more and alot...
But i must admit the luckfactor (shorttime,in longterm its fanished) is really annoying in a strategy called-game.Thats why we (me and my friends) started playing without the dice on the boardgame.
I already asked for something like that here (forum;no dice),but theres a strong lobby against it sadly enough.With no dice you can keep alot new players playing at cc whom otherwise return to the cardboard game.