Dice Complaints and Various Suggestions to Fix Them

[shocked439]

All I know is the dice are either ridiculously good are impossibly bad, there is no expected results with the dice. It is atrocious.

[maasman]

1,000,000 assaults per day. So the random number file is being recycled quite often. The "impossible" is bound to happen in that many permutations.

The hardest part about dealing with random numbers is that they are so darn unpredictable.

To get a rough estimate, 5,000,000 dice per day distributed uniformly over 24 hours is ~ 200,000 dice per hour. The dice list has 50,000 entries, so we can guess that a given list is run through about four times before it is replaced.

I could have sworn I saw somewhere that the dice are replaced when they are all used, or when the hour is up, whichever comes first. So I don't think the same numbers are used multiple times.

[blakebowling]

1,000,000 assaults per day. So the random number file is being recycled quite often. The "impossible" is bound to happen in that many permutations.

The hardest part about dealing with random numbers is that they are so darn unpredictable.

To get a rough estimate, 5,000,000 dice per day distributed uniformly over 24 hours is ~ 200,000 dice per hour. The dice list has 50,000 entries, so we can guess that a given list is run through about four times before it is replaced.

I could have sworn I saw somewhere that the dice are replaced when they are all used, or when the hour is up, whichever comes first. So I don't think the same numbers are used multiple times.

This was changed with the last update. IIRC the same file is used constantly and not refreshed.

[Juan_Bottom]

What does "high quality random numbers" even mean? I've always wondered this but didn't want to look dumb. How low quality can 6 different numbers be?

[Metsfanmax]

What does "high quality random numbers" even mean? I've always wondered this but didn't want to look dumb. How low quality can 6 different numbers be?

Even if your script produces theoretically random numbers, in the sense that it's impossible to predict what the result would be (i.e. there are no regular patterns), it's not worth much if it doesn't also produce a normal distribution; that is, you would expect that each number from 1 to 6 would occur about 1/6 of the time. If the generator is biased in such a way that, say, 5 is substantially more likely than any other number, then your random numbers are not high quality.

[Woodruff]

What does "high quality random numbers" even mean? I've always wondered this but didn't want to look dumb. How low quality can 6 different numbers be?

Metsfanmax's answer sounds much more educated, but I actually suspect that was only a modest attempt at humor.

[Ace Rimmer]

What does "high quality random numbers" even mean? I've always wondered this but didn't want to look dumb. How low quality can 6 different numbers be?

Even if your script produces theoretically random numbers, in the sense that it's impossible to predict what the result would be (i.e. there are no regular patterns), it's not worth much if it doesn't also produce a normal distribution; that is, you would expect that each number from 1 to 6 would occur about 1/6 of the time. If the generator is biased in such a way that, say, 5 is substantially more likely than any other number, then your random numbers are not high quality.

(I'm no mathematician, but I had a good night's sleep last night)
I'm pretty sure that's not true. If it follows a normal distribution then it would not be random - random means that the past and future rolls have no impact on the current roll, so you could roll a 6 for 100,000 times in a row in true random. There is no need for a normal distribution, because true randomness will not have a normal distribution. You would expect it to haev a normal distribution, but it is not required.

[trapyoung]

Yeah but any skew towards any number favors the defender if there is no normal distribution. That's why it would be more useful to just have a list of 666 numbers that list 1-6 111 times each and jumble it up then just project which might be random and skew. The distribution should not be random, only when and how they are generated show be.

[squishyg]

i think the RNG should be updated to reflect how often in real life my dice roll off the table.

[Doc_Brown]

What does "high quality random numbers" even mean? I've always wondered this but didn't want to look dumb. How low quality can 6 different numbers be?

Even if your script produces theoretically random numbers, in the sense that it's impossible to predict what the result would be (i.e. there are no regular patterns), it's not worth much if it doesn't also produce a normal distribution; that is, you would expect that each number from 1 to 6 would occur about 1/6 of the time. If the generator is biased in such a way that, say, 5 is substantially more likely than any other number, then your random numbers are not high quality.

(I'm no mathematician, but I had a good night's sleep last night)
I'm pretty sure that's not true. If it follows a normal distribution then it would not be random - random means that the past and future rolls have no impact on the current roll, so you could roll a 6 for 100,000 times in a row in true random. There is no need for a normal distribution, because true randomness will not have a normal distribution. You would expect it to haev a normal distribution, but it is not required.

Metsfan is actually correct. It is true that for a true random number generator, prior result have no bearing on the probability of the outcome of the next roll. However, because of that very fact, over a very large number of rolls, assuming the probability density function is uniform, you should expect to see roughly equal numbers of each of the possible outcomes. This principle underlies quite a bit of modern physics (particularly quantum physics).

Now, rolling a 6 100 times in a row has a low probability in and of itself, but if it happens, the probability of rolling a six on the next roll is still unchanged (1/6 probability). Also, if you roll the dice enough times, it would become extremely unlikely to not see 100 6s in a row on occasion.

[trapyoung]

i think the RNG should be updated to reflect how often in real life my dice roll off the table.

It'd be cooler if it reflected how often my brother would throw the dice on the board and wipe any amount of troops off Africa and part of Europe. Like nuke spoils but more random and more devestating.

[squishyg]

i think the RNG should be updated to reflect how often in real life my dice roll off the table.

It'd be cooler if it reflected how often my brother would throw the dice on the board and wipe any amount of troops off Africa and part of Europe. Like nuke spoils but more random and more devestating.

haha!!! so true!!

[PLAYER57832]

(I'm no mathematician, but I had a good night's sleep last night)
I'm pretty sure that's not true. If it follows a normal distribution then it would not be random - random means that the past and future rolls have no impact on the current roll, so you could roll a 6 for 100,000 times in a row in true random. There is no need for a normal distribution, because true randomness will not have a normal distribution. You would expect it to haev a normal distribution, but it is not required.

Not all populations follow a bell curve, but getting into all that is more than my tired brain can handle at the moment. And, people are having a hard enough time understanding random/normal distribution already, without throwing other issues into the mix.

[PLAYER57832]

Now, rolling a 6 100 times in a row has a low probability in and of itself, but if it happens, the probability of rolling a six on the next roll is still unchanged (1/6 probability). Also, if you roll the dice enough times, it would become extremely unlikely to not see 100 6s in a row on occasion.

QFT.

That last sentence is what people have a hard time understanding and accepting, both.

[Metsfanmax]

Now, rolling a 6 100 times in a row has a low probability in and of itself, but if it happens, the probability of rolling a six on the next roll is still unchanged (1/6 probability). Also, if you roll the dice enough times, it would become extremely unlikely to not see 100 6s in a row on occasion.

QFT.

That last sentence is what people have a hard time understanding and accepting, both.

While it's true, the expected probability of that outcome is on the order of 10^-78. So you couldn't roll the dice enough times in the current age of the Universe for it to become extremely unlikely not to have a few of those streaks ;P

[Tennisie]

Now, rolling a 6 100 times in a row has a low probability in and of itself, but if it happens, the probability of rolling a six on the next roll is still unchanged (1/6 probability). Also, if you roll the dice enough times, it would become extremely unlikely to not see 100 6s in a row on occasion.

QFT.

That last sentence is what people have a hard time understanding and accepting, both.

While it's true, the expected probability of that outcome is on the order of 10^-78. So you couldn't roll the dice enough times in the current age of the Universe for it to become extremely unlikely not to have a few of those streaks ;P

It all depends on the state of the universe. Considering the universe as a pseudorandom number generator with an infinitely long cycle, if it was started at the right spot in the cycle, it could produce many of those streaks because the streaks themselves are randomly dispersed throughout the cycle. Apparently we've hit a spot in the cycle where there are lots of streaks. For example, I once lost 87 - 0 during a single auto-attack, which I'm told involves a probability of around 10^-20.

[Metsfanmax]

It all depends on the state of the universe. Considering the universe as a pseudorandom number generator with an infinitely long cycle

No.

, if it was started at the right spot in the cycle, it could produce many of those streaks because the streaks themselves are randomly dispersed throughout the cycle. Apparently we've hit a spot in the cycle where there are lots of streaks.

Also, no.

For example, I once lost 87 - 0 during a single auto-attack, which I'm told involves a probability of around 10^-20.

That's slightly more reasonable. The age of the Universe is on the order of 10^21 seconds.

[Donlarry]

u can still get your wins in with bad dice

[Woodruff]

u can still get your wins in with bad dice

If I weren't just recently a private, I would greatly enjoy responding to this.

[Donlarry]

i do lose about 20% of my games to bad dice u get doubled off in the beginning of the game and its over or a important point in the game ull get doubled off sometimes ull get 4 doubles in a row

[nebsmith]

And why would these things be randomized anyways? If a true army of 1000 men attacked one of 200 would the 200 randomly defeat them?

If you don't like the basic premise that CC (and the game it is based upon) is built on then you are free to go and invent your own game that suits your own preferences.

Yes! If you don't like risk, find another game you do like. Don't try and change this one into something that is no longer risk.

[TheForgivenOne]

And why would these things be randomized anyways? If a true army of 1000 men attacked one of 200 would the 200 randomly defeat them?

If you don't like the basic premise that CC (and the game it is based upon) is built on then you are free to go and invent your own game that suits your own preferences.

Yes! If you don't like risk, find another game you do like. Don't try and change this one into something that is no longer risk.

But I like monopoly...

[Thezzaruz]

So do I. Lets petition for a mandatory "Free Parking" territory on all maps. \o/\o/\o/

[mgourley]

People, you don't seem to know what random is.

Random does not mean extended hot or cold streaks won't happen, it mean's ANYTHING can happen!

[ben79]

NEW DICE SYSTEM FAST