One more thing about the dice.

[zorba_ca]

Supermarcol... Earlier in this thread, Lack wrote:

This new set should last for 10 days before being recycled at the current rate of play.

So it appears that the file does not get changed.

Even if the file did get changed, then while yes I agree with you that over time the distribution of any single file should not make a difference, I still find it unpalatable that at any given time I have a greater or lesser chance of achieving any single outcome when rolling.

It goes back to my analogy of the 7-sided die. Would you still play against me using 7-sided die? Although neither of us would know which number is more likely to come up, we know that one number will come up more often. Both of us would be equally affected, so it could still be fun, but there is a definite appearance of inequality.

[supermarcol]

I found a way to go around the cheating though, simply order the attack and def dice from biggest to smallest. that way the order will most likely be changed compared to the file. I don't think its that hard to do, most programing languages have a sorting function. So unless you get the same numbers you can't cheat with that.

[lackattack]

I still find it unpalatable that at any given time I have a greater or lesser chance of achieving any single outcome when rolling.

You know there are more more 6;6;6;6;6 than 1;1;1;1;1 only because you looked at the file. If you did not look at the file, your chances of rolling 6;6;6;6;6 and 1;1;1;1;1 would be the same, from your perspective.

If you make any randomly generated file of 500,000 rolls there will probably be more of one or more of the other. And by looking at the file you are peeking at the outcomes before they happen.

You can't look back and say the dice are not random after a game of risk because your opponents rolled more sixes than ones.

[Marvaddin]

These dice arent fine. Again, again, AGAIN I lost in a fight 11 x 3 armies, I lost 9 and killed 1. The chances for 4 losses and 1 draw are 0,002 in 5 rolls 3x2, or 0,2%. Why this is always happening to me? And in the worst hours? My partner can tell you how the dice are ALWAYS against me. I had 2 or 3 moments of great luck since Im playing here, and almost all time I have the worst luck in the world.

I will analize the file lack put here.

[zorba_ca]

Lack:

This may be long, so please bear with me:

1. The number of 6s and the number of 1s in the file are irrelevent. If the rolls come out 6;6;6;6;6 or 1;1;1;1;1 the result is the same - defender wins two.

2. My point is that if using a given file the possible outcome of any given roll will not be perfectly distributed. The randomness is fine so long as it is a random occurrence of a non-random set (in this case, the possible dice outcomes). The right way of fixing the problem is to have 4 lists:

one with all the possible outcomes for when two dice are rolled
one with all the possible outcomes for when three dice are rolled
one with all the possible outcomes for when four dice are rolled, and
one with all the possible outcomes for when five dice are rolled

Then, for each attack, randomly select any one occurrence from the appropriate file.

3. The file is not representative of the normal distribution of dice rolls.

Further analysis can accurately determine in which way the game is skewed.

I think that you have done a remarkable job the site in general and with ensuring randomness in particular (I don't buy into the whole "sticking" thing) - we just need to make sure the right items are being randomized.

[lackattack]

one with all the possible outcomes for when two dice are rolled
one with all the possible outcomes for when three dice are rolled
one with all the possible outcomes for when four dice are rolled, and
one with all the possible outcomes for when five dice are rolled

Then, for each attack, randomly select any one occurrence from the appropriate file.

How would I randomly select the roll from these lists of all possible outcomes? Could I get a random number from 1 to 7776 (or less as the case may be) from random.org? Could I get a bunch of these random numbers in advance and put them in a file? Are you gonna look at the file and tell me that it's not evenly distributed? Aren't we back at square one?

I'll concede that it would be nicer if I didn't recycle the file. But I still don't understand how this file does not contain 500,000 perfectly random rolls of up to 5 dice.

[SMITH197]

you've definatly confused the hell out of me. But i was just curious...those of you that play other games that involve dice online, be it another risk site or another game altogether, how do those sites create random or pseudo-random rolls? Maybe they could shead some light on this...

[HighBorn]

all i know is i just lost 50+ armys to 9 in a game i just finished u figure it out

[zorba_ca]

Lack:

The thing about probability is that it will even out over the long term. Recycling the file is definitely the biggest issue since we no longer have the same odds of any given outcome over a period of time. We (the players) just don't know what the most likely outcomes will be.

Could I get a random number from 1 to 7776 (or less as the case may be) from random.org?

I would recommend having one file from 1-36, one file from 1-216, one file from 1-1296 and one file from 1-7776, with the right file being chosen based on how many dice are being rolled.

Could I get a bunch of these random numbers in advance and put them in a file?

Yes, but do not recycle the file. That way we can be assured that in the long run all outcomes will be equally likely.

Are you gonna look at the file and tell me that it's not evenly distributed?

Do not publish the file. So long as the file is not recycled and the numbers are randomly generated (I don't think anyone can argue this point) no one should care. The key issue is that every outcome should have an equal chance of occuring.

I still don't understand how this file does not contain 500,000 perfectly random rolls of up to 5 dice.

This is not in dispute. All rolls are equal. Some are just a little more equal than others.

[Comrade]

Yes, you can have true randomness.

But you need quantum computers for that.

[geniusxboy]

The dice are messed up because i keep losing 6 armies to kill 1. The frequency of this occuring has skyrocketed after the downtime. The proof is there. I remember when it was functioning correctly because the results matched the proper statistics of winning and losing. I should be winning 80% of the time against a single army, but instead I am losing 80% of the time.

[zorba_ca]

You should actually only be winning 66% of the time 3 against 1.

[kingwaffles]

That's the theory, but it i chance so its not always exactly what you expect. That's the idea behind probability, its what you expect not necessarily what happens...

[fishfleas]

In the end there will always be people complaining about bad rolls.... but I think that there are some real legitimate concerns here. When you generate a file and remove numbers from the file you are taking away true randomness. as the chance of those numbers being rolled are being greatly reduced. each number should always be able to be rolled.... it shouldn't be determined by a predetermined list that takes away rolls as you do it. I know you simulated the rolls in the first place but that takes away some of the chances taht you will roll something not on that list.

[lackattack]

I don't think you get it, fishfleas. I didn't remove any numbers from the file. Any combination had an equal chance to be generated.

What I've done is show everybody a preview of what the random dice rolls will be. I did this to demonstrate that the dice are quality random numbers.

It seems that all the criticisms stem from the fact that I revealed the dice rolls and not from the quality of the file.

So now I am going to generate a new file from random.org and I will keep the contents secret. I will replace the file every so often but it may also be recyled a few times. And that will be the end of it

[geniusxboy]

You said that the numbers are picked from the atmospheric pressure right?

When you generate 500,000 all at once, you are grabbing numbers from a small and constant set of atmospheric pressure fluctuation numbers, right? It takes only 3 seconds to generate the file so the flux can't be that large....

[Scarface]

All I can say is that I'm new, I've only played 2 games so far, and am losing both from ridiculous dice rolls. I've had rolls where I lost 10 armies to one, each roll having my opponent meet my exact roll, thus having me lose armies since I am attacking. Although this is a game, I still find it hard to believe that it is "random" when out of 9 rolls, my opponent meets my EXACT roll each and every single time!

[zorba_ca]

Lack:

It seems that all the criticisms stem from the fact that I revealed the dice rolls and not from the quality of the file.

So now I am going to generate a new file from random.org and I will keep the contents secret. I will replace the file every so often but it may also be recyled a few times. And that will be the end of it

Before you proceed as planned, I want to let you know that I have finished analyzed 20% of the rolls and have some findings you may find interesting.

NOTE TO OTHERS: What follows is may be confusing. If you are not sure about what is being written, please send a PM to me so as not to bog down this thread.

Basically, the file CC uses is, for the most part, statistically fair. The probability of every battle outcome does approximate the expected probability of battle outcomes. That is not to say that there are losing and winning streaks we all experience. However, there are minor variances between the real and expected battle outcomes. While the level of approximation is very reassuring, the minor variations may be a problem, especially due to the recycling of the file.

The most glaring issue with the file is with the probability of getting any particular roll combination. For example, when attacking 3 v 1, it is almost twice as likely that the attacker rolls a 2, 4, 1, to the defender's 1, than it is for the attacker to roll a 4, 5, 1 to the defender's 1.

The result for each of these roll combinations is the same (defender loses 1 army), but these variances can account for problems, both real and imagined. For example, because certain roll combinations are more likely, players may feel the dice are "sticking". This gives CC the apperance of impropriety. Furthermore, the fact that each roll combination does NOT have an equal chance of occuring can account for the differences between real and expected battle outcomes.

In sum, I recommend two courses of action:
1. First and foremost, the dice file should never be recycled. Probability is based on results over a period of time, and any anomalies in any file will only be exacerbated if the file is recycled.

2. I renew my position that a seperate file be created for each of the six possible battle scenarios (1v1, 1v2, 2v1, 2v2, 3v1 and 3v2). This would ensure each roll combination has an equal chance of arising. However, even if you choose to do this, this does not negate the need to never recycle the files.

If you did not look at the file, your chances of rolling 6;6;6;6;6 and 1;1;1;1;1 would be the same, from your perspective.

The issue is not the player's perception that the chances of each dice combination is the same. The issue is that the chances of each dice combination should be the same.

Thank you for your time and dedication to this website. I look forward to your reply.

[Scarface]

Ok, this is happening again! Every time I roll the dice, the opposing player meets me! In 5 turns, I've been able to conquer all of 3 countries, with no armies left anywhere to show for it!!! (1 army on each country). The dice are flawed, I don't care what kind of stupid scientific calculation you can come up with, they're flawed and nothing you can say will make me change my mind.

[geniusxboy]

The dice are flawed, I don't care what kind of stupid scientific calculation you can come up with

i agree.

THE










DICE
















ARE
















BROKEN

[smloh]

i have another theory...

You see, because we are approaching true randomness in using numbers from random.org, we are moving away from the algorithmically enforced "perfect distributions" of pseudo-randomness.

As a result, the wider factors that play a role in the greater scheme of things are able to exert their indirect influence (like the butterfly with regards to chaos theory), and affect the outcomes of the battles.

So, we are left with the hypothesis that when players experience unfortunate streaks of the dice, perhaps it is the result of a greater power working against them. Call it luck, call it God, call it karma...

So, geniusxboy, Scarface, and everyone else who's complaining about the dice, to prove/disprove this postulate, we need you to answer a question:

WHEN WAS THE LAST TIME YOU HELPED AN OLD LADY ACROSS THE STREET?




...

more seriously, zorba's analysis sounds great. i thought maybe that earlier suggestion of just keeping a file of numbers (53125643251...) and taking as many numbers as needed each time might sound easier and less messy though?

[zorba_ca]

Yes, smloh, that works too (we should give props to TROY for the suggestion) - so long as that file is not recycled either.

I'm just partial to my suggestion because it helps retain the "feel" of rolling dice (you know: equal chance of all possible roll combinations flashing on screen = equal chance of all possible roll combinations appearing on dice).

But if it is too hard to do then a sequence of numbers should work.

Thanks for the serious post. It's good to steer away from the "dice are flawed" theories to something more constructive.

[Scarface]

I helped an old lady accross the street this morning actually...

I am not basing this on a spurr of bad luck. I am basing this on my personal opinion, and if you disagree with it, that's your choice. I'm not a rocket scientist, and can't make out half the crap that's being said on here. However, do stand by my statement. You might consider it bad luck, karma, god, etc. I call it fact.

Fact: You have a file with 500,000 pre-determined rolls that are intended to resemble true randomness.

Fact: Random is defined as not having a pre-set parameter, meaning the file could have 5,000,000,000,000 pre-determined rolls, and it STILL would not be random.

Fact: I will keep playing because even though I am getting upset with losing all my armies when faced with a 9vs1 situation, I still think this site is great and highly addictive.

I would call it bad luck if I actually rolled the dice and no one had determined ahead of time what my roll will be. However, if a person has decided ahead of time what the roll will be, then it is not random, and thus my excuses of LUCK, KARMA or even GOD himself are useless as they are not the cause for my bad roll, the creator of the file is.

Note: I still think lackattack is doing an awesome job for a one-man (or possibly woman) team!

[fishfleas]

I don't think you get it, fishfleas. I didn't remove any numbers from the file. Any combination had an equal chance to be generated.

What I've done is show everybody a preview of what the random dice rolls will be. I did this to demonstrate that the dice are quality random numbers.

It seems that all the criticisms stem from the fact that I revealed the dice rolls and not from the quality of the file.

So now I am going to generate a new file from random.org and I will keep the contents secret. I will replace the file every so often but it may also be recyled a few times. And that will be the end of it

No I do get it... I'm just looking at it simply and logically..... if you take any number of predetermined rolls then you are affecting what is able to be rolled at any one given point in time. When you then have it remove a roll from the list that you randomly generated you are reducing and affecting the chance that that number is rolled again....


To make things a little clearer on a simple level.... say I rolled 10 numbers 5 are 6s and 5 are 1s this is about as simple as I can get... yo uhave a 50/50 shot at getting a 6..... that in itself skews the true randomness of the roll... then to top it off... you start taking away a roll each time you roll..... This is what I understood you were doing with your roll sets.....

Say you rolled a 6 the first roll..... Then you have 55% chance of rolling a 1, and a 44% chance of rolling a 6. say you hit the 6 again....
Your next roll you would have a 33% chance of hitting a 6.....

This shows in a very basic way how using fixed numbers skews true randomness

With more numbers and more random rolls you can tame down the drastic affects that this method can have, however the method is still there and the consequences are still there...

Trying to simulate randomness on a computer is a very daunting task and I believe that the only real way to try to simulate it is by an animated rolling system.... That is just my opinion though. If implemented in a correct manor the randomness is affect by a human and therefor would greater reflect ones ability to roll random numbers. This might not be even a possibility without major bandwidth issues.... I'm not sure.... but this I believe would be a much better and simpler way to eliminate the "stickyness" you get when doing it in the current method.

So lack yes I do understand your method, unless somehow I misread something you were saying at one point. I don't believe you letting people evaluate the rolling schemes was a bad idea but you should definately listen to the feedback.... when so many are complaining about rolls and it is occuring so often things need to be done about it.

[lackattack]

Firstly, thanks for the feedback. I want to to be transparent about the inner workings of the dice, because you put your trust in them.

For the sake of arguement, let's say the list of 500,000 rolls is kept secret and is NOT recylcled. I agree recycling the file is bad. In zorba's words, "any anomalies in any file will only be exacerbated if the file is recycled"

On the issue of fixed / pre-determined rolls:

When the list was generated, it was written line-by-line with true random numbers. Later it is read line-by-line, as players attack each other. All the list really is is a buffer, or queue, so that random.org can be consulted in advance. If I pop a 5 3 2 4 1 off the list, that does not reduce the chance of the next line being the same.

On zorba's suggestion of enumerating all possible combinations and then choosing one:

When generating each line on the list, I randomly picked a number from 1 to 6 five times. That does not favour any combination, does it? Since each line is independent and is used once, each roll combination has an equal chance of arising the moment you roll the dice.

On the dice favoring defender / a certain time of day / players who click refresh / players with blue eyes:

People pay attention more when they lose. There is no favoritism. Luck is chaotic. It is normal to get crazy results now and then.