Dice query...

[Renee_W]

Ran across this in another thread...

The dice currently work like this:

CC gets a string of 50 000 numbers between 1-6 from random.org (which gets it's random numbers from atmospheric noise). This string is updated once an hour, ie. CC gets a new string each hour, regardless of the amount of rolls within that time.

For each individual assault, a random location is chosen from that string of 50 000, and the required 2-5 dice are read from that position.

Maybe I'm rehashing something old and tired but something about this doesn't seem right. It was always my understanding that the string from random.org was intended to be inherently random when used in a sequential, use once manner. But this method seems to completely negate the point of using random.org. Wouldn't the use of the other random system contaminate the data from random.org? ie If the random number generator used to pick the string index has a disproportionate tendency toward results it already used wouldn't that tendency negate the randomness of the random.org data to whatever extent it is flawed?

If the random number generator used doesn't have flaws in it's randomness why do we need random.org?

If the random number generator used does have flaws how do they not degrade the randomness from random.org?

Is this moot because it's an incorrect description of how we use random.org(only description I've seen)?

[shieldgenerator7]

+1 another dice thread! interesting...
yes, it seems odd that we'd need another site to give us our randomness if we already have a random number generator. Maybe this explains the streakiness of the dice. Using this information, maybe we could modify the random number generators to make a list of the indices that it has already used and make sure that it doesn't use them again before it's used the other ones. That would be a good solution, however, it might take a bit of programming and might take longer for each die roll, slowing down the system. Speed vs. Randomness. Take your pick.

[Renee_W]

maybe we could modify the random number generators to make a list of the indices that it has already used and make sure that it doesn't use them again before it's used the other ones. That would be a good solution, however, it might take a bit of programming and might take longer for each die roll, slowing down the system. Speed vs. Randomness. Take your pick.

Given the assumption that random.org numbers are truly random the results should be equally random whether we use a (truly) random index once or sequentially pass through the list and discard it at the end. For it to matter whether you go through random.org numbers via indices 1, 2, 3, 4 or 4342, 454, 5432, 53 the data from random.org would have to contain a pattern and be... non random. Therefore it should be suitable to just run sequentially through the list then get a new list when it's used up and save the processing time noting which indices have been used. You shouldn't need complex processing to use random.org numbers.

But it seems to me using another source of random numbers to pick the index from random.org would either be redundant(random selection of a random number being random) and the source used to pick the index is suitable by itself or the source of the random indices contains a source of non-randomness that then contaminates the randomness of the result.

[natty dread]

Look at it this way... random.org gives CC a string of 50000 numbers. Those numbers are all truly random.

For each roll a position in the string is chosen and the 2-5 numbers are read from that position. Now, to be random, the numbers chosen must not be affected by the previous numbers. The current system satisfies that requirement. It doesn't matter which position you start reading the numbers from, they are still random.

In other words... using a PNRG to pick numbers from a string of TRN:s does not "contaminate" the numbers. This is because each roll must be independent of the previous ones, so you can think of every roll of being the "first" one being read, and it doesn't matter where in the string you start reading, since it's all equally random.

As for streaks, they are a part of random sequences - humans just want to see patters in everything.

[Renee_W]

In other words... using a PNRG to pick numbers from a string of TRN:s does not "contaminate" the numbers. This is because each roll must be independent of the previous ones, so you can think of every roll of being the "first" one being read, and it doesn't matter where in the string you start reading, since it's all equally random.

What is the nature of the non-randomness in the PNRG that it can be nullified by the selection of a TRN based on the PNRG result? Lets say the the PNRG was prone to streakiness that streakiness should carry to the TRN results. Lets say for a given seed the PNRG was 1% more likely to select the number 1000. Then for a given auto attack it would be expected that the 5 numbers following index 1000 would be 1% more likely to be rolled. Granted such a problem would be so small it probably couldn't be clearly detected over varience by an individual player. But that would also be the case with simply using PNRG numbers in the first place.

The problem with a PNRG is that the numbers produced aren't independent. The problem is slight but it's the reason something like random.org exists. If the basis for index selection is not independent and index's can be reused(I can't imagine CC is using a table to prevent reuse on a 50k index string lol) then the resulting numbers should also not be independent.

As for streaks, they are a part of random sequences - humans just want to see patters in everything.

True and non-relevant. I seriously doubt even the PNRG standard distro in any programming language would produce results sufficiently far from random any player could tell for sure the results were non-random. But I'm not seeing how not quite random selection of random numbers doesn't result in not quite random numbers. It seems random.org is just a name to throw out and the actual benefits of using the source are being negated.

[MeanestBossEver]

I'm really confused. This appears to be an intelligent, knowledgeable and well thought out question about the dice. I'm pretty sure it doesn't belong here.