PDA

View Full Version : Random number generator - please make it more random



BurningDownTheHouse
01-06-2011, 03:04 AM
Look at the combat log in the image below:

http://i433.photobucket.com/albums/qq58/natanmager/ScreenShot00003.jpg

Does anything about the rolls there strike you as being less than random?

Now, I'll be the first to admit that the kind of number clustering you see in the image works both ways, and sometimes you get 3 vorpals in a row as well.

Having said that, this is ridiculous. The chances for 3 1's in a row on a d20 is 1 in 8000, it happens way more than that. I've had the same number rolled 10 times in a row on a d100 roll as well. The chance for this happening is lower than the chance of winning the grand prize in the lottery two consecutive times in a row!!!

I know that logical random number generators tend to cluster over short instances, but games like this is the exact place where this kind of clustering is detrimental.

There is a way to have a real random number generator used (the one using atmospheric readings comes to mind). The problem is, that it costs money.
I think that it would be a worthwhile investment none the less.

Your thoughts now.

Neversignup
01-06-2011, 03:34 AM
OP... I'm not trying to knock you mate but it sounds like you are after system that offers different numbers everytime. Which ain't random. Which would be exploitable to the brilliant player base that makes up the Eberron community.

Dendrix
01-06-2011, 03:38 AM
That is random. 1 in 8000 chances happen 1 in 8000 times. the no of dice rolls made in a 4 hour gaming session is tens of thousands

BurningDownTheHouse
01-06-2011, 03:39 AM
OP... I'm not trying to knock you mate but it sounds like you are after system that offers different numbers everytime. Which ain't random. Which would be exploitable to the brilliant player base that makes up the Eberron community.

Umm, not sure what you mean...
I want the numbers not to cluster, what did you think I wanted?

BurningDownTheHouse
01-06-2011, 03:42 AM
That is random. 1 in 8000 chances happen 1 in 8000 times. the no of dice rolls made in a 4 hour gaming session is tens of thousands

This was just an example.
Try attacking the ship dummy, and record the clustering.
You'll be surprised.

TechNoFear
01-06-2011, 04:04 AM
You assume that each character, mob, chest, pool of lava, etc has their own random number generator, which I do not think is the case.

So your three '1's in a row are not consecutive values from the games/servers random number generator (ie there are lots of other numbers generated by the RNG inbetween each of your '1's so it is not 'clustering').

BurningDownTheHouse
01-06-2011, 04:12 AM
You assume that each character, mob, chest, pool of lava, etc has their own random number generator, which I do not think is the case.

So your three '1's in a row are not consecutive values from the games/servers random number generator (ie there are lots of other numbers generated by the RNG inbetween each of your '1's so it is not 'clustering').

I don't think that only one instance of a random number generator will be able to support the games needs.
I would guess that there is one per character, mob, or any other relevant game object. Makes more sense to me.

Daggaz
01-06-2011, 04:16 AM
The random generator has been tested extensively, and disregarding external code effects, has been shown to be consistently very random. In fact, random number generators in general are extremely well made these days and do NOT suffer the commonly held misconception of being "not random at all." Go ahead and wiki it...

Stop kicking this dead horse, especially with such an incredibly limited data set, especially when you admit to knowing that your data set is far too limited to be of use.

somebody want to link the appropriate threads.. I dont bother bookmarking that stuff.

BurningDownTheHouse
01-06-2011, 04:25 AM
The random generator has been tested extensively, and disregarding external code effects, has been shown to be consistently very random. In fact, random number generators in general are extremely well made these days and do NOT suffer the commonly held misconception of being "not random at all." Go ahead and wiki it...




There are two principal methods used to generate random numbers. One measures some physical phenomenon that is expected to be random and then compensates for possible biases in the measurement process. The other uses computational algorithms that produce long sequences of apparently random results, which are in fact completely determined by a shorter initial value, known as a seed or key. The latter type are often called pseudorandom number generators.

A "random number generator" based solely on deterministic computation cannot be regarded as a "true" random number generator, since its output is inherently predictable. How to distinguish a "true" random number from the output of a pseudo-random number generator is a very difficult problem. However, carefully chosen pseudo-random number generators can be used instead of true random numbers in many applications. Rigorous statistical analysis of the output is often needed to have confidence in the algorithm.


And I did not say "not random at all" I said "tend to cluster over short instances".



Stop kicking this dead horse, especially with such an incredibly limited data set, especially when you admit to knowing that your data set is far too limited to be of use.


Shouldn't be dead if it is...
/cast raise dead ;)

MrLarone
01-06-2011, 04:26 AM
i believe the OP is looking for this:

http://gamesbyemail.com/News/DiceOMatic

BurningDownTheHouse
01-06-2011, 04:33 AM
i believe the OP is looking for this:

http://gamesbyemail.com/News/DiceOMatic

Cool, just adapt it to the other D&D dice and we're golden. :D
+1

Lord_WC
01-06-2011, 05:16 AM
1. Fight for a bit.
2. Check your last 3 attack rolls (let's say it was 13,16,8)
3. Those had 1 in 8000 chance as well.

Backley
01-06-2011, 05:18 AM
Clusters aren't evidence of non-randomness.

Simple test:
Teacher asks class to write down their guesses for the results of 100 coin flips.
Teacher flips a coin 100 times and records the actual results.

The class's guesses are always less clustered (streaks of all Heads or all Tails) than the actual results.

Humans just aren't wired to think random.

MrLarone
01-06-2011, 05:21 AM
Humans just aren't wired to think random.

banana philharmonic garden tyres!

MrLarone
01-06-2011, 05:23 AM
Humans just aren't wired to think random.

banana philharmonic garden tyres!

***edit***

just to note that it was a bit of a struggle to think up 4 words without an immediate cognitive association (e.g. after banana my first instinct was to type monkey) so i guess you're right Backley =)

Keybreaker
01-06-2011, 05:27 AM
Clusters aren't evidence of non-randomness.

Yes, clustering is actually an indication of randomness. (Indication, not proof... Conversely, no clustering with a very large sample size is good evidence of non-randomness.) As Backley is saying, truly random sequences contain many clusters.

Humans tend to confuse scattered, with random. Another example is random dots on a page. Ask a human to randomly place 20 dots on a piece of paper... the usual result is scattered, well spaced-out dots. However, a truly random pattern would more than likely have 2 dots very close to each other, ie cluster.

Another popular example is the apparent birthday paradox. How many people do you need in a classroom to assure a better than 50% chance that two people have the same birthday? (Hint: It's fewer than you think.)

Anyhoo, the point is... OP does not understand randomness.

Dendrix
01-06-2011, 05:37 AM
This was just an example.
Try attacking the ship dummy, and record the clustering.
You'll be surprised.

No I won't because I understand mathematics and randomness. Frequent clustering is to be expected in any random distribution.

Here's are two example of randomness
1) How many people do you need in any random grouping to have a 50% chance that two of them share the same birthday?
2) How many people do you need in any random grouping to have a 99% chance that two of them share the same birthday?

Answers below
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
23 and 57

BurningDownTheHouse
01-06-2011, 05:43 AM
1. Fight for a bit.
2. Check your last 3 attack rolls (let's say it was 13,16,8)
3. Those had 1 in 8000 chance as well.


Clusters aren't evidence of non-randomness.

Simple test:
Teacher asks class to write down their guesses for the results of 100 coin flips.
Teacher flips a coin 100 times and records the actual results.

The class's guesses are always less clustered (streaks of all Heads or all Tails) than the actual results.

Humans just aren't wired to think random.


Yes, clustering is actually an indication of randomness. (Indication, not proof... Conversely, no clustering with a very large sample size is good evidence of non-randomness.) As Backley is saying, truly random sequences contain many clusters.

Humans tend to confuse scattered, with random. Another example is random dots on a page. Ask a human to randomly place 20 dots on a piece of paper... the usual result is scattered, well spaced-out dots. However, a truly random pattern would more than likely have 2 dots very close to each other, ie cluster.

Another popular example is the apparent birthday paradox. How many people do you need in a classroom to assure a better than 50% chance that two people have the same birthday? (Hint: It's fewer than you think.)

Anyhoo, the point is... OP does not understand randomness.

Alright, I admit that I am not a mathematician and definitely not a statistician.
But...
I was just beating on the dummy till its death, 213 hits, and the results are:
No clusters of 13,16,8
2 clusters of 3 2's
1 cluster of 3 5's
2 clusters of 3 13's
1 cluster of 3 20's
and...
1 cluster of 4 3's

OK, so no large clusters of 1's this time (there were 2 in a row, but I didn't record those)... :rolleyes:

Now tell me, is this just stupid old me, or does look just a tad weird?

BurningDownTheHouse
01-06-2011, 05:45 AM
No I won't because I understand mathematics and randomness. Frequent clustering is to be expected in any random distribution.


Look above and tell me that those results make statistical sense.

MrLarone
01-06-2011, 05:54 AM
23 and 57

how'd you work those out?

sincere curiosity since it's been a while since i played with statistics.

to my mind it's both questions are answered by the number of people minus one dived by the number of days in the (normal) year. or:

(N-1)/365

which i believe is right as you can only be 100% certain to have two people with the same birthday once you have 366 or more (as then you would have one person to match every day and then one left over to guarantee the duplicate day).

in which case you have a 50% chance at around 182 and a 99% chance around 361 (spreadsheet floating point accuracy not withstanding)

correct me if i'm wrong =)

Lord_WC
01-06-2011, 06:00 AM
Alright, I admit that I am not a mathematician and definitely not a statistician.
But...
I was just beating on the dummy till its death, 213 hits, and the results are:
No clusters of 13,16,8
2 clusters of 3 2's
1 cluster of 3 5's
2 clusters of 3 13's
1 cluster of 3 20's
and...
1 cluster of 4 3's

OK, so no large clusters of 1's this time (there were 2 in a row, but I didn't record those)... :rolleyes:

Now tell me, is this just stupid old me, or does look just a tad weird?

My point was only that 3 numbers are just that, 3 numbers. It's only a coincidence that we think 1 has a special meaning, but it has not. 3 consecutive 1s have the exact same chance as any 3 (same or different) numbers. It doesn't matter.

And random is random. You can roll 1000000000 1s in a row, that does not tell anything about the randomness of the random generator.

My personal best was 4 20s in a row while vorpaling stuff, it looked great:) (knowing that those are actually 4 20s in 8 rolls)

Drallac
01-06-2011, 06:06 AM
http://en.wikipedia.org/wiki/Birthday_problem

statistics is a funny think that is often weird, like someone else said this is a dead horse. "Clusters" of numbers as you say, are in fact proof it is random, otherwise if there were never clusters randomness could never occur, as Tim Minchin said "as we know things that have a 1 in 64 000 000 chance if happening, happen all the time"

BurningDownTheHouse
01-06-2011, 06:10 AM
My point was only that 3 numbers are just that, 3 numbers. It's only a coincidence that we think 1 has a special meaning, but it has not. 3 consecutive 1s have the exact same chance as any 3 (same or different) numbers. It doesn't matter.

And random is random. You can roll 1000000000 1s in a row, that does not tell anything about the randomness of the random generator.

My personal best was 4 20s in a row while vorpaling stuff, it looked great:) (knowing that those are actually 4 20s in 8 rolls)

And yet I can't help but wonder.
So many clusters in such a small sample range? Some of them multiple times?
I guess that the only way to be sure will be to record each number and then analyze the entire data set.

I'm sure that in the long run it will add up just fine. But clustering like this in small samples are deterministic to this kind of game.

Keybreaker
01-06-2011, 06:11 AM
how'd you work those out?

sincere curiosity since it's been a while since i played with statistics.

to my mind it's both questions are answered by the number of people minus one dived by the number of days in the (normal) year. or:

(N-1)/365

which i believe is right as you can only be 100% certain to have two people with the same birthday once you have 366 or more (as then you would have one person to match every day and then one left over to guarantee the duplicate day).

in which case you have a 50% chance at around 182 and a 99% chance around 361 (spreadsheet floating point accuracy not withstanding)

correct me if i'm wrong =)

The "Birthday Paradox" is a well known one and quite extensively explained here: http://en.wikipedia.org/wiki/Birthday_problem

In probability theory, the birthday problem, or birthday paradox[1] pertains to the probability that in a set of randomly chosen people some pair of them will have the same birthday. By the pigeonhole principle, the probability reaches 100% when the number of people reaches 367 (including February 29 births). But perhaps counter-intuitively, 99% probability is reached with just 57 people, and 50% probability with 23 people. These conclusions are based on the assumption that each day of the year (except February 29) is equally probable for a birthday.

BurningDownTheHouse
01-06-2011, 06:17 AM
http://en.wikipedia.org/wiki/Birthday_problem

statistics is a funny think that is often weird, like someone else said this is a dead horse. "Clusters" of numbers as you say, are in fact proof it is random, otherwise if there were never clusters randomness could never occur, as Tim Minchin said "as we know things that have a 1 in 64 000 000 chance if happening, happen all the time"

This is actually fascinating, yet it doesn't explain to me how I can determine the "expected" number of clusters in a given test.

If you can come up with something like that, it will either completely defeat my arguments and I will capitulate, or it will show that I am on to something.

BurningDownTheHouse
01-06-2011, 06:24 AM
You have 213 samples from, i don't know millions of rolls a day?

The problem is you see significance in rolling the same number as before. You probably think rolling 2,2,2 is somehow different than rolling 3 numbers you just make up. But esentially those two examples are the same.

I understand that, that's why I said that I'll try to get a whole list and analyze it.

Lord_WC
01-06-2011, 06:26 AM
And yet I can't help but wonder.
So many clusters in such a small sample range? Some of them multiple times?
I guess that the only way to be sure will be to record each number and then analyze the entire data set.

I'm sure that in the long run it will add up just fine. But clustering like this in small samples are deterministic to this kind of game.

You have 213 samples from, i don't know millions of rolls a day?

The problem is you see significance in rolling the same number as before. You probably think rolling 2,2,2 is somehow different than rolling 3 numbers you just make up. But esentially those two examples are the same.

SiliconShadow
01-06-2011, 06:33 AM
We all know that some dice in DDO are weighted.

However some dice should not be, but for example when I roll to hit and strike a vorpal, if I get a 20 then most of the time the following striking will also be a 20, then looking at the values you often see the faster you strike the more numbers appear clustered as put.

This has led me to the conclusion that due to some code optimisation in order to make the random function perform better has caused this because no random generation that is true random or pseudo random has that sort of frequent clustering.

The "random" numbers in DDO do cluster more than any random set should when taken in close fast proximity, either due to weighting or optimisation it happens and is obvious, the most obvious is when using many shot with slaying arrows most of the procs when hasted double proc sometimes triple proc if there is lag.

Optimisation / Weighting... both? :rolleyes:

MrLarone
01-06-2011, 06:34 AM
The "Birthday Paradox" is a well known one and quite extensively explained here: http://en.wikipedia.org/wiki/Birthday_problem

sweet, ta!

Draccus
01-06-2011, 06:44 AM
In over 500 UMD rolls recorded, I am four times as likely to roll a 1 as I am a 20.

My theory is that Turbine uses skewed dice rolls to try to balance the game that they completely unbalanced with ridiculous loot and exploitable monster AI.

BurningDownTheHouse
01-06-2011, 06:50 AM
In over 500 UMD rolls recorded, I am four times as likely to roll a 1 as I am a 20.

My theory is that Turbine uses skewed dice rolls to try to balance the game that they completely unbalanced with ridiculous loot and exploitable monster AI.

There you go then.
Can anyone tell me if this is an acceptable sample size? And if so, is it an "acceptable"/ "expected" result?

Keybreaker
01-06-2011, 06:50 AM
Part of the issue here is that humans are not evolutionarily programmed to recognize randomness. We are programmed to recognize patterns... even when such patterns do not exist.

In the dark... given limited information... a few blurry shapes...

Caveman A: Awww, prolly nothin'. Just something random.
Caveman B: Ack! Was that a face?? Was that a bear??

Imagine 10,000 caveman As and 10,000 caveman Bs... The nervous, pattern-recognizing Bs are more likely to survive and procreate. The laid-back, randomness appreciating As might be right sometimes, heck even most of the times... but sometimes they will be at the spear-end of a jealous neighbor, or bear-food.

We are programmed with an observer bias to see patterns, even when they are not there.

Drallac
01-06-2011, 06:52 AM
This is actually fascinating, yet it doesn't explain to me how I can determine the "expected" number of clusters in a given test.

If you can come up with something like that, it will either completely defeat my arguments and I will capitulate, or it will show that I am on to something.




CHALLENGE ACCEPTED!

Think of it this way, the value of pi, is irrational meaning it in theory goes on forever. Lets look at pi, pi to 1000 places which is here: http://c2.com/cgi/wiki?ValueOfPi

within pi to 1000 places there is:
1 set of 000
2 sets of 111
1 set of 555
1 set of 999999 (six nines)

and no more sets of triples, granted these arent randomly generated numbers, but it is from a small sample of what could effectively be random numbers as it is just a list.

Repetition is almost required in an infinite sequence of numbers otherwise it cannot be random, if there is no randomness an algorithm must be applied, similarly if you want to remove clusters of numbers the randomness is removed.

I gotta head to an exam but i will be back and try to explain better

Lord_WC
01-06-2011, 06:52 AM
We all know that some dice in DDO are weighted.

However some dice should not be, but for example when I roll to hit and strike a vorpal, if I get a 20 then most of the time the following striking will also be a 20, then looking at the values you often see the faster you strike the more numbers appear clustered as put.

This has led me to the conclusion that due to some code optimisation in order to make the random function perform better has caused this because no random generation that is true random or pseudo random has that sort of frequent clustering.

The "random" numbers in DDO do cluster more than any random set should when taken in close fast proximity, either due to weighting or optimisation it happens and is obvious, the most obvious is when using many shot with slaying arrows most of the procs when hasted double proc sometimes triple proc if there is lag.

Optimisation / Weighting... both? :rolleyes:

If you had EVERY roll EVER made by ddo probably you could analyze it. Otherwise the samples you take are just too insignificant compared to the vast numbers. Statistically it can happen that in the first 2 years of ddo noone ever rolled a 20, and in the next two years the frequency of 20s was 30%, it still wouldn't say anything about the randomness.

(and by randomness I mean computer generated randomness, which is not true random strictly speaking)

By analyzing samples you could state, that this sample of say 10.000 results had only 0,01% of appearing in a random environment. That would mean that happens in every 100 million rolls. Considering how many people are online, how many servers are and how many rolls are made, this really couldn't take long.

BurningDownTheHouse
01-06-2011, 06:56 AM
CHALLENGE ACCEPTED!

Think of it this way, the value of pi, is irrational meaning it in theory goes on forever. Lets look at pi, pi to 1000 places which is here: http://c2.com/cgi/wiki?ValueOfPi

within pi to 1000 places there is:
1 set of 000
2 sets of 111
1 set of 555
1 set of 999999 (six nines)

and no more sets of triples, granted these arent randomly generated numbers, but it is from a small sample of what could effectively be random numbers as it is just a list.

Repetition is almost required in an infinite sequence of numbers otherwise it cannot be random, if there is no randomness an algorithm must be applied, similarly if you want to remove clusters of numbers the randomness is removed.

I gotta head to an exam but i will be back and try to explain better

And yet my (almost 5 times) shorter list has shown more clustering than you have shown here...
Good luck with the test. :)

Lord_WC
01-06-2011, 06:58 AM
And yet my (almost 5 times) shorter list has shown more clustering than you have shown here...
Good luck with the test. :)

The example was just an example to show that repetition happens. You cannot compare an irrational number (which is NOT random digits following each other) with 10 different outcomes (digits from 0-9) to a supposedly random sample with 20 different outcomes.

Just for fun, here is 213 d20 rolls from ddo:
10,10,19,6,16,11,16,6,6,1,19,9,17,7,5,8,18,8,18,9, 9,6,15,1,12,14,9,18,9,5,18,18,10,18,7,16,10,2,15,1 4,8,17,17,14,
15,17,15,13,12,17,19,12,11,20,16,8,8,9,15,5,3,7,6, 17,12,13,2,8,13,13,3,7,8,13,14,8,18,20,2,1,9,17,5, 16,2,3,2,5,4,
6,6,12,10,14,1,12,4,14,12,17,8,9,11,7,17,5,4,8,12, 9,10,3,19,17,12,4,15,9,17,3,2,11,3,1,13,2,6,12,5,1 1,20,10,10,2,
19,20,4,20,20,16,7,7,17,14,4,7,5,14,12,2,8,3,18,4, 14,5,6,15,10,14,16,12,10,1,7,4,7,18,17,12,16,8,1,1 5,15,6,5,19,
19,18,19,18,12,5,16,18,9,8,7,5,13,1,2,9,5,19,18,17 ,2,8,15,17,19,20,1,7,6,14,5,10,10,18,4

No triplets here as far as i can see. Does this mean anything? I don't think so

sweez
01-06-2011, 07:05 AM
The thing I find most interesting about this thread is the fact that you decided to remove your HP bar from the screenshot.

SiliconShadow
01-06-2011, 07:15 AM
If you had EVERY roll EVER made by ddo probably you could analyze it. Otherwise the samples you take are just too insignificant compared to the vast numbers. Statistically it can happen that in the first 2 years of ddo noone ever rolled a 20, and in the next two years the frequency of 20s was 30%, it still wouldn't say anything about the randomness.

(and by randomness I mean computer generated randomness, which is not true random strictly speaking)

By analyzing samples you could state, that this sample of say 10.000 results had only 0,01% of appearing in a random environment. That would mean that happens in every 100 million rolls. Considering how many people are online, how many servers are and how many rolls are made, this really couldn't take long.

It is impossible for these patterns to happen and they do happen for 14 months I have played this game they are obvious.

As I stated it is one of two things occuring, an optimisation in which results are only refetched from the randomiser every x ms, or the numbers come from a lookup table which means patterns repeat obviously on that CPU thread (in which case Jo blogs on another instance will have a different thread for the randomiser to himself) OR the dice are weighted.

Pseudo random and true random are different and there are dozens of ways of optimising something like this that is a key element of such a system that could produce this clustering effect, the dice rolls are not, cannot be and will never be truely random and will always produce clustering in patterns if there is clustering in the first place.


The example was just an example to show that repetition happens. You cannot compare an irrational number (which is NOT random digits following each other) with 10 different outcomes (digits from 0-9) to a supposedly random sample with 20 different outcomes.

Just for fun, here is 213 d20 rolls from ddo:
10,10,19,6,16,11,16,6,6,1,19,9,17,7,5,8,18,8,18,9, 9,6,15,1,12,14,9,18,9,5,18,18,10,18,7,16,10,2,15,1 4,8,17,17,14,
15,17,15,13,12,17,19,12,11,20,16,8,8,9,15,5,3,7,6, 17,12,13,2,8,13,13,3,7,8,13,14,8,18,20,2,1,9,17,5, 16,2,3,2,5,4,
6,6,12,10,14,1,12,4,14,12,17,8,9,11,7,17,5,4,8,12, 9,10,3,19,17,12,4,15,9,17,3,2,11,3,1,13,2,6,12,5,1 1,20,10,10,2,
19,20,4,20,20,16,7,7,17,14,4,7,5,14,12,2,8,3,18,4, 14,5,6,15,10,14,16,12,10,1,7,4,7,18,17,12,16,8,1,1 5,15,6,5,19,
19,18,19,18,12,5,16,18,9,8,7,5,13,1,2,9,5,19,18,17 ,2,8,15,17,19,20,1,7,6,14,5,10,10,18,4

No triplets here as far as i can see. Does this mean anything? I don't think so

Dice number:Times - 1:9, 2:11, 3:7, 4: 9, 5:14, 6: 11, 7:12, 8:14, 9:12, 10:12, 11:5, 12:14, 13:7, 14:11, 15:10, 16:9, 17:15, 18:14, 19:10, 20:7
Mean 10.5493
Median 10
Mode 17
St Dev 5.643803
Clusters 14
Non Clusters 198
% clustered 7.07
Max cluster spacing 42

Is this random? It is pseudo random definitely the distribution is almost perfect
Is it weighted? No
Is it optimised in some way that makes this clustering happen? We can't tell from this set but from the small sample it could be worth checking if anyone is ever bored enough.

We do not know how the servers are put together, how "instances" are seperated, how and where each dice roll goes to for a number, but it is not true random.

What we are most likely looking at is something like:
http://www.random.org/analysis/randbitmap-wamp-section.png

bokaboka
01-06-2011, 07:17 AM
Screenshot or it didn't....... nvm

;) :)

BurningDownTheHouse
01-06-2011, 07:22 AM
The example was just an example to show that repetition happens. You cannot compare an irrational number (which is NOT random digits following each other) with 10 different outcomes (digits from 0-9) to a supposedly random sample with 20 different outcomes.

Just for fun, here is 213 d20 rolls from ddo:
10,10,19,6,16,11,16,6,6,1,19,9,17,7,5,8,18,8,18,9, 9,6,15,1,12,14,9,18,9,5,18,18,10,18,7,16,10,2,15,1 4,8,17,17,14,
15,17,15,13,12,17,19,12,11,20,16,8,8,9,15,5,3,7,6, 17,12,13,2,8,13,13,3,7,8,13,14,8,18,20,2,1,9,17,5, 16,2,3,2,5,4,
6,6,12,10,14,1,12,4,14,12,17,8,9,11,7,17,5,4,8,12, 9,10,3,19,17,12,4,15,9,17,3,2,11,3,1,13,2,6,12,5,1 1,20,10,10,2,
19,20,4,20,20,16,7,7,17,14,4,7,5,14,12,2,8,3,18,4, 14,5,6,15,10,14,16,12,10,1,7,4,7,18,17,12,16,8,1,1 5,15,6,5,19,
19,18,19,18,12,5,16,18,9,8,7,5,13,1,2,9,5,19,18,17 ,2,8,15,17,19,20,1,7,6,14,5,10,10,18,4

No triplets here as far as i can see. Does this mean anything? I don't think so

This list doesn't even have many doubles, is this from DDO?

Lord_WC
01-06-2011, 07:22 AM
It is improbable these patterns to happen and they do happen for 14 months I have played this game so I feel like they are obvious.

As I stated I feel maybe it is one of two things occuring, an optimisation in which results are only refetched from the randomiser every x ms, or the numbers come from a lookup table which means patterns repeat obviously on that CPU thread (in which case Jo blogs on another instance will have a different thread for the randomiser to himself) OR the dice are weighted.

Pseudo random and true random are different and there are dozens of ways of optimising something like this that is a key element of such a system that could produce this clustering effect, the dice rolls are not, cannot be and will never be truely random and will always produce clustering in patterns if there is clustering in the first place.

Corrected a bit. I don't think you made extensive testing or wrote the code of the game. While you probably notice 4 1s or 20s in a row, may I ask you what was the frequency of numbers 11,12,13,14 appearing in your rolls in this order?

You probably don't know it, because they are absolutely insignificant for the game. You will notice a 1, but a 11 will be hidden.

Until a dev confirms it or someone gets me numbers I say you two post some rare occurances while 10.000 other people don't post about rolling 9 on their saves 4 times in a row.

BurningDownTheHouse
01-06-2011, 07:29 AM
The thing I find most interesting about this thread is the fact that you decided to remove your HP bar from the screenshot.


Lol, the HP bar is hiding behind the enlarged combat log window. :)
OK, this is my ac build rogue :
http://my.ddo.com/character/thelanis/assacinate/
This image is a bit old (please fix MyDDO :( ) , but he has only 20 hp more now, so what the heck. :p

Lord_WC
01-06-2011, 07:31 AM
This list doesn't even have many doubles, is this from DDO?

Yes, as I already said that.

Fun fact: I found exactly 14 doubles in this sample. The average amount of doubles in a 213 result sample should be 10,65. So The sample has about 35% higher double rate than the average, yet you comment on how few it has. I'm not attacking you, just would like to show you that feelings and first impressions are unreliable in statistics:)

just for fun here is 213 results from excel with trunc(rnd()*20+1) function:


6 12 2 17 8 20 19 2 1 13 9 4 12 6 16 10 15 16 2 10 7 2 5 8 17 18 16 18 19 20 3 6 2 7 18 8 4 13 6 16 20 16 9 20 10 4 2 1 9 19 3 4 19 20 11 16 5 4 3 7 5 16 13 15 14 8 12 2 11 20 12 13 12 8 6 15 11 18 10 9 19 7 20 11 14 5 8 10 1 10 8 3 14 4 8 6 5 6 9 10 6 2 13 18 5 20 4 2 17 13 19 8 10 8 7 20 18 9 7 3 10 14 9 10 9 10 11 5 4 16 10 1 2 20 5 16 12 8 17 3 1 2 5 14 12 4 12 18 9 2 12 3 2 17 10 4 8 16 8 20 14 5 20 16 2 4 10 18 4 6 7 20 11 5 3 4 20 15 6 17 4 8 14 11 8 5 1 15 3 13 12 1 5 8 11 12 7 17 14 3 6 9 13 1 19 12 6 8 12 10 8 16 13

as far as I can see not even one double. That is highly unlikely.

MrLarone
01-06-2011, 07:33 AM
other people don't post about rolling 9 on their saves 4 times in a row.

it was 5 and i did ask you keep that secret! ;-)

BurningDownTheHouse
01-06-2011, 07:38 AM
Fun fact: I found exactly 14 doubles in this sample. The average amount of doubles in a 213 result sample should be 10,65. So The sample has about 35% higher double rate than the average, yet you comment on how few it has. I'm not attacking you, just would like to show you that feelings and first impressions are unreliable in statistics:)

Actually my impression is probably derived from my experience with my list. I didn't even bother to note the doubles because there were so many...

So what you saying is that the list that I was willing to accept as a "counterbalance" to my own is also skewed?

I'd say that my point just got a bit stronger. :)

OK, that last one was a joke.

On the other hand you claim that no amount of testing we can do, will provide us with enough data to make any claims.
I can't accept this verdict that easily, sorry.

BurningDownTheHouse
01-06-2011, 07:41 AM
just for fun here is 213 results from excel with trunc(rnd()*20+1) function:


6 12 2 17 8 20 19 2 1 13 9 4 12 6 16 10 15 16 2 10 7 2 5 8 17 18 16 18 19 20 3 6 2 7 18 8 4 13 6 16 20 16 9 20 10 4 2 1 9 19 3 4 19 20 11 16 5 4 3 7 5 16 13 15 14 8 12 2 11 20 12 13 12 8 6 15 11 18 10 9 19 7 20 11 14 5 8 10 1 10 8 3 14 4 8 6 5 6 9 10 6 2 13 18 5 20 4 2 17 13 19 8 10 8 7 20 18 9 7 3 10 14 9 10 9 10 11 5 4 16 10 1 2 20 5 16 12 8 17 3 1 2 5 14 12 4 12 18 9 2 12 3 2 17 10 4 8 16 8 20 14 5 20 16 2 4 10 18 4 6 7 20 11 5 3 4 20 15 6 17 4 8 14 11 8 5 1 15 3 13 12 1 5 8 11 12 7 17 14 3 6 9 13 1 19 12 6 8 12 10 8 16 13

as far as I can see not even one double. That is highly unlikely.

So are you saying that the RND that is used by Microsoft is even worse than the one we have here?

This one will actually not take me by surprise? :p

Lord_WC
01-06-2011, 07:47 AM
Actually my impression is probably derived from my experience with my list. I didn't even bother to note the doubles because there were so many...

So what you saying is that the list that I was willing to accept as a "counterbalance" to my own is also skewed?

I'd say that my point just got a bit stronger. :)

OK, that last one was a joke.

On the other hand you claim that no amount of testing we can do, will provide us with enough data to make any claims.
I can't accept this verdict that easily, sorry.

The main point of randomness is that it is random. It can contain any outcome with some probabilites. You can only tell us that your sample's result is highly unlikely, but will never be able to say it 100% sure that it is not random, unless a dev confirms it, or the guy who wrote the code.

If we are rolling enough, sooner or later we will have a sample of 213 1s. If you show this to someone, he will say the rng is flawed, however if you tell him that it took 10^80 rolls to get it, it's a whole different story:D Just too many rolls happening in this game to prove something about the rng, sorry.

Shoal
01-06-2011, 07:47 AM
This is actually fascinating, yet it doesn't explain to me how I can determine the "expected" number of clusters in a given test.

If you can come up with something like that, it will either completely defeat my arguments and I will capitulate, or it will show that I am on to something.

I'm pretty sure that this is true:

A "run" includes clustering of one number, e.g., 1,2,2,3,3,3 has 3 runs. In *multiple* large sequences of random numbers, the number of runs is distributed as a normal (Gauss) distribution, with a mean of (2N-1)/3 and a variance of (16N-29)/90. For example, if you have 40 rolls, you would expect 26 +/- 2.6 runs with 95% likelihood.

However, I only scanned very old notes from a probability course, so take it for whatever it's worth.

EDIT: 26 +/- 5.2 runs with 95% likelihood. Accordingly, 26 +/- 2.6 with 67%.

Lord_WC
01-06-2011, 07:52 AM
So are you saying that the RND that is used by Microsoft is even worse than the one we have here?

This one will actually not take me by surprise? :p

Don't blame M$ now, pls :D Actually generating high-quality random numbers with computers is not easy, and they will never be random strictly speaking. Sure, it would be great to know what function turbine did call for the rng, but for the time being lets just accept that it's truly random.

BurningDownTheHouse
01-06-2011, 07:57 AM
I'm pretty sure that this is true:

A "run" includes clustering of one number, e.g., 1,2,2,3,3,3 has 3 runs. In *multiple* large sequences of random numbers, the number of runs is distributed as a normal (Gauss) distribution, with a mean of (2N-1)/3 and a variance of (16N-29)/90. For example, if you have 40 rolls, you would expect 26 +/- 2.6 runs with 95% likelihood.

However, I only scanned very old notes from a probability course, so take it for whatever it's worth.

Can you elaborate on the *size* of the runs? I assume that the longer the run, the less likely it is to appear.

BurningDownTheHouse
01-06-2011, 07:58 AM
Don't blame M$ now, pls :D Actually generating high-quality random numbers with computers is not easy, and they will never be random strictly speaking. Sure, it would be great to know what function turbine did call for the rng, but for the time being lets just accept that it's truly random.

Well, the whole point of this thread is that I don't accept that it's truly random... ;)

Does any dev feel like intervening and giving us an unforgettable quote? :D

Lord_WC
01-06-2011, 08:05 AM
Well, the whole point of this thread is that I don't accept that it's truly random... ;)

Does any dev feel like intervening and giving us an unforgettable quote? :D

Most likely the difference of the outcome between a true random and computer generated random numbers amounts to one wasted or gained Battle Coin during the lifetime of the universe. It's pointless really.

Maldavenous
01-06-2011, 08:06 AM
Well, the whole point of this thread is that I don't accept that it's truly random... ;)

You're just seeing patterns and discarding the information between them. The reason clusters seem to happen so often is because they're easily recognizable while a string of unrelated numbers is easily dismissible. Even in a P&P session players frequently see clusters, it's part of being random. The number you roll one time has no effect on what you will roll next.

Shoal
01-06-2011, 08:13 AM
Can you elaborate on the *size* of the runs? I assume that the longer the run, the less likely it is to appear.

The size of the run is only relevant to the extent that it reduces the total number of runs. Also, I made a calculation error in my previous post, but I'll attach an edit.

There are more factors in analyzing random sequences, such as classifying the runs into those followed by higher or lower numbers, but this doesn't change the basic distribution.

BurningDownTheHouse
01-06-2011, 08:20 AM
You're just seeing patterns and discarding the information between them. The reason clusters seem to happen so often is because they're easily recognizable while a string of unrelated numbers is easily dismissible. Even in a P&P session players frequently see clusters, it's part of being random. The number you roll one time has no effect on what you will roll next.

Fact: Most P&P dice are slightly to highly weighted. You can usually pick one that "rolls better".

What I'm saying is that the random number generator that is used here might also be weighted (intentionally or is a debate for a different thread).

Keybreaker
01-06-2011, 08:31 AM
Fact: Most P&P dice are slightly to highly weighted. You can usually pick one that "rolls better".

What I'm saying is that the random number generator that is used here might also be weighted (intentionally or is a debate for a different thread).

Fact: You are making a claim with no evidence whatsoever.

Fact: You are categorically disregarding attempts to educate you about probability and statistics of which you are repeatedly demonstrating your ignorance.

Broldin
01-06-2011, 08:32 AM
In over a decade of playing MMO's I have seen this complaint come up multiple times relating to every game I've played. Generally from people who have no understanding of statistics what so ever. No offence.

There is one simple concept people need to understand. Using an example of 5 rolls of a d20, you are just as likely to get 1/1/1/1/1 as you are 2/19/16/17/5, or 10/2/5/11/2, or 1/2/3/4/5, or any other combination of 5 specific numbers. Statistically, there is nothing different between getting five 1's or five different numbers. Every roll of the dice has a 1/20 chance of being a 1, 2, 3, etc... It's the human mind that give the importantce to a patern.

That said, if you find you are getting a specific roll/patern at specific places/times, it may be more than a statistical anomoly. Proving this would requrie a statistical test to be completed within a reasonable margin of error. It's credibility would then need validated by calculating the power of the statistical test. Basically statistics are a mess. It's not hard math like algebra, trig, or geometry. There's no proff the can illustrate the basis for statistics as it's mostly rules man has made up in order to quantify probabilities and predictions.

There was a point in time where there were legitimate arguements about randomization in computer algorythms. Those days have passed, it's time to move on to something else.

BurningDownTheHouse
01-06-2011, 08:51 AM
Fact: You are making a claim with no evidence whatsoever.


Alright you got me. Not only P&P dice. Almost all existing dice. The only dice that are more or less reliable are casino dice. All else, although unintentionally are slightly skewed.



Fact: You are categorically disregarding attempts to educate you about probability and statistics of which you are repeatedly demonstrating your ignorance.

I get the idea that statistically 1,1,1,1 and 4,7,9,2 have the same chance of occurring.
What I didn't get yet, is a way to explain observations (apparently by other people than me as well) that show distinct deviations which cannot be as easily dismissible as you claim.

BurningDownTheHouse
01-06-2011, 08:53 AM
In over a decade of playing MMO's I have seen this complaint come up multiple times relating to every game I've played. Generally from people who have no understanding of statistics what so ever. No offence.

There is one simple concept people need to understand. Using an example of 5 rolls of a d20, you are just as likely to get 1/1/1/1/1 as you are 2/19/16/17/5, or 10/2/5/11/2, or 1/2/3/4/5, or any other combination of 5 specific numbers. Statistically, there is nothing different between getting five 1's or five different numbers. Every roll of the dice has a 1/20 chance of being a 1, 2, 3, etc... It's the human mind that give the importantce to a patern.

That said, if you find you are getting a specific roll/patern at specific places/times, it may be more than a statistical anomoly. Proving this would requrie a statistical test to be completed within a reasonable margin of error. It's credibility would then need validated by calculating the power of the statistical test. Basically statistics are a mess. It's not hard math like algebra, trig, or geometry. There's no proff the can illustrate the basis for statistics as it's mostly rules man has made up in order to quantify probabilities and predictions.

There was a point in time where there were legitimate arguements about randomization in computer algorythms. Those days have passed, it's time to move on to something else.

Very well, I'll take what you say at face value.
Sadly all I get is that even if something is wrong, there is no way that I personally can prove it.

Thanks for all the replies guys, gotta sign off for a bit.

BDog77
01-06-2011, 08:55 AM
The fact is that no one in this forum and quite possibly none of the Devs could even tell you whether or not the RNG used is truly random, in the same way that, as you pointed out, dice may be weighted. The best anyone can do is say, in theory, this particular algorithm "should" produce a fairly random sampling and then by testing, ensure that it meets certain base criteria, one of which is that there must be a certain minimum number of clusters. In any set of random numbers, looking at it in a "meta" sense, clustering is a mathematical certainty.

When you look at at a series of rolls and say something like the chance of 4 ones appearing is 1 in 8000 (or whatever), you are essentially applying that criteria to a static set of rolls (that is, rolls that have already happened), and then making a statement about how probability applies to that static set of rolls. The fact is that when those rolls were made, the probability of a 1 on any roll was (on a d20) exactly 5 percent, no more, no less. In order to have a truly random system, any previous results must be ignored.

Based on some of your earlier comments, I am curious as to what you consider a truly random system? For example, you mention something that samples the atmosphere somehow and creates a random number based on this. Do you consider atmospheric phenomenon to be random? I don't. What is that old wives tale about lightning, that it never strikes the same place twice? Not at all random.... Even if you wanted to use something like that, I can't imagine how you would test it any better than any of the RNGs we have today for so called "randomness", and even if you could, I'm not sure it would pass the minimum requirements for randomness. I think there are far too many factors that "weight" atmospheric phenomenon for it to be truly random.

If by "random", you mean a system where you do not understand the forces weighting it, well then, rejoice! You absolutely do not understand all the factors weighting the DDO RNG, making it the perfect system for you.....

Cam_Neely
01-06-2011, 08:56 AM
Having said that, this is ridiculous. The chances for 3 1's in a row on a d20 is 1 in 8000, it happens way more than that. I've had the same number rolled 10 times in a row on a d100 roll as well.

Just to put it out there, if what you said is true (sounds about right, but did not check) there is also a 1 in 8000 chance in rolling a 1, 10 and 20 in a row which is a large distribution. If you had a decent sample, I think you would find that it happens about that often, so your more random random number generator is in place.

I wish the dmg rolls were not loaded, that bugs me -.-

Lord_WC
01-06-2011, 09:01 AM
Don't blame M$ now, pls :D Actually generating high-quality random numbers with computers is not easy, and they will never be random strictly speaking. Sure, it would be great to know what function turbine did call for the rng, but for the time being lets just accept that it's truly random.

Really, someone neg repping me for this?

Cam_Neely
01-06-2011, 09:08 AM
No I won't because I understand mathematics and randomness. Frequent clustering is to be expected in any random distribution.

Here's are two example of randomness
1) How many people do you need in any random grouping to have a 50% chance that two of them share the same birthday?
2) How many people do you need in any random grouping to have a 99% chance that two of them share the same birthday?

Answers below
.

23 and 57
Actually, this proves that you dont understand randomness. A persons birthday is not purely random. ie if you went up and asked a person their birthday there is not a 1/365 chance that there birthday is a set day. Its been shown in various places that there are more babies conceived at specific times of the year, so there are peak birthdays. Ie there are alot of birthdays around mid November, due to it being 9 months after valentines day. This also does not take into account people born on Feb 29th (leap year)

Not to be picky, but your example is not a random distribution.

BDog77
01-06-2011, 09:23 AM
This also does not take into account people born on Feb 29th (leap year)

Actually, as i understand it, this does take into account Feb 29th. But you are right, not really a random sample....

Dendrix
01-06-2011, 09:25 AM
how'd you work those out?

sincere curiosity since it's been a while since i played with statistics.


http://en.wikipedia.org/wiki/Birthday_problem

It's computed as follows
the chance that the 1st person shares a birthday with any of the other 22 +
the chance that the 2nd person shares a birthday with the remaining 21 +
the chance that the 3rd person shares a birthday with the remaining 20 +
the chance that the 4th person shares a birthday with the remaining 19 +
etcetera down to
the chance that the 22nd person shares a birthday with the last person.

SiliconShadow
01-06-2011, 09:26 AM
Corrected a bit. I don't think you made extensive testing or wrote the code of the game. While you probably notice 4 1s or 20s in a row, may I ask you what was the frequency of numbers 11,12,13,14 appearing in your rolls in this order?

You probably don't know it, because they are absolutely insignificant for the game. You will notice a 1, but a 11 will be hidden.

Until a dev confirms it or someone gets me numbers I say you two post some rare occurances while 10.000 other people don't post about rolling 9 on their saves 4 times in a row.

Maths of random and programming random are different, in plain and simple terms you are wrong, this is a PRNG that does very obviusly throw clustering on some numbers and 3, 9 and 16 are other numbers that cluster often.

This is obviously from a small seed and perhaps a threaded function which has a table for the seeding for performance.

This in lamens terms means that number clusters will repeat and repeat for an individual in a single instance if observed for long enough, and this is already evident in damage rolls, to hit rolls etc.

There is no true random function at work on DDO, it is impossible for the technology to cope with that burden in the first place, the fact is the only way is it uses a optimised PRNG and not TRNG fact and it cannot be denied the servers just could not cope with anything else with the amount of rolls made.

bobbryan2
01-06-2011, 09:28 AM
Even distributions != random.

Now, to humans, streaks appear that something isn't random. But any truly random sequence will include clumping.

auximenes
01-06-2011, 09:32 AM
Having said that, this is ridiculous. The chances for 3 1's in a row on a d20 is 1 in 8000, it happens way more than that. I've had the same number rolled 10 times in a row on a d100 roll as well. The chance for this happening is lower than the chance of winning the grand prize in the lottery two consecutive times in a row!!!
A bit of a late reply, but, have you ever played P&P D&D? Streaks of bad rolls happen all the time. It has cost many an adventurer while swimming or after a critical fumble. :cool:

Cam_Neely
01-06-2011, 09:33 AM
Actually, as i understand it, this does take into account Feb 29th. But you are right, not really a random sample....

Ya, was an Assumption on my part ( you know what they say about assumptions http://socialmediaanswers.com/wp-content/uploads/2008/11/motivational_assumptions.jpg :) ), but the wiki link mention above leaves it out. The OP would have to not have rounded his number to be definitive. But thinking more about it, it might not matter. I think the actual answer is 22 and change, but you have to round up to 23, since you cant have half a person to answer the question correctly. The 1/1461 or so chance of Leap Year birthday wont cause you to go to 23.0000000000001, so you would still only be rounding up to 23 people.

BDog77
01-06-2011, 09:37 AM
this is already evident in damage rolls


These are admittedly weighted......

Lord_WC
01-06-2011, 09:43 AM
Maths of random and programming random are different, in plain and simple terms you are wrong, this is a PRNG that does very obviusly throw clustering on some numbers and 3, 9 and 16 are other numbers that cluster often.

This is obviously from a small seed and perhaps a threaded function which has a table for the seeding for performance.

This in lamens terms means that number clusters will repeat and repeat for an individual in a single instance if observed for long enough, and this is already evident in damage rolls, to hit rolls etc.

There is no true random function at work on DDO, it is impossible for the technology to cope with that burden in the first place, the fact is the only way is it uses a optimised PRNG and not TRNG fact and it cannot be denied the servers just could not cope with anything else with the amount of rolls made.

The difference between the two is so small that you probably will not pull the shorter straw because of it in a lifetime.

Otherwise, what you are saying is that the program repeats 'small' seeds. As far as i know smallest seeds range in the 2^32 order of magnitude. Even if the game would use the worst prng out there, you cannot possibly see it in 213 results, nor in 14 months of playing:)

Just for the numbers, in the above example you would have to make 118 rolls every second for 14 months 24/7 to have a full sample. And erm, recording it, not FEELING about it:)

I repeat, 118 rolls/s for your 14 months of continous gameplay.

Sorry, I don't see your reasoning either obvious, or true. Quoted samples are just too small for that.

brlftz
01-06-2011, 09:44 AM
In over 500 UMD rolls recorded, I am four times as likely to roll a 1 as I am a 20.

My theory is that Turbine uses skewed dice rolls to try to balance the game that they completely unbalanced with ridiculous loot and exploitable monster AI.

Were you involved in that testing that was done a couple of years or so ago? There was a long thread about umd rolls, and my recollection is that there's a very convincing case to be made re a low bias on umd rolls, at least. your theory about why is interesting.

regarding clusters, it seems to happen to me a LOT, and i've always assumed it had something to do with a poor connection rather than the rng. when data is flying fast, it seems like the die display stops refreshing.

xTethx
01-06-2011, 09:59 AM
This thread still doesnt explain why I roll a 1 on soundburst 9 out of 10 times a mob throws it on me. :)

Dirac
01-06-2011, 10:25 AM
Wheeee! Here we go again! :) Seriously, though, it is always good to get this argument out of everyone's system. For those that really care, relevant previous discussions of the RNG are here (http://forums.ddo.com/showthread.php?t=228373) and here (http://forums.ddo.com/showthread.php?t=277484).

First, the obvious: You can never tell if an RNG is purely random anecdotally. The human brain is hardwired to do this incorrectly.

Second, the boilerplate to dispel common misconceptions:
1. Computer RNG's should be completely random for us. Any standard RNG will produce results indistinguishable from pure randomness.
2. Randomness can be measured and quantified without any knowledge of the inner workings of the system. We can determine if the RNG is sufficiently random.

Next, the current state of affairs: After a lot of testing there is no confirmed deviation from pure randomness (seen in the previous threads). I've also tested clustering on my data set. Also nothing.

Finally, how you can help: Take a large data set of random numbers anywhere: UMD, attack dummy, anything. Post the number of 1s, 2s, etc. I, and others, can test for deviaitions from pure randomness.

p.s. I do not have time right now to look at the numbers posted in this thread, but will try to get to them later.

cdemeritt
01-06-2011, 10:40 AM
Alright, I admit that I am not a mathematician and definitely not a statistician.
But...
I was just beating on the dummy till its death, 213 hits, and the results are:
No clusters of 13,16,8
2 clusters of 3 2's
1 cluster of 3 5's
2 clusters of 3 13's
1 cluster of 3 20's
and...
1 cluster of 4 3's

OK, so no large clusters of 1's this time (there were 2 in a row, but I didn't record those)... :rolleyes:

Now tell me, is this just stupid old me, or does look just a tad weird?

Now do that 100 more times and then tell us there is a problem. And at some point somewhere a cluster of 20 20's will occur.

The statistics of random numbers is that a lot of the time random doesn't look random... Also humans are wired to look for and see patterns where patterns don't exist...

Look at how many generations and man hours have gone into researching "The Bible Code" (over 700 years, and countless number of people spending their lives to find a few relevant clusters of words, to them)

People want to find patterns and hidden messages... typically they find something they want to see.

Cam_Neely
01-06-2011, 10:43 AM
Number Theory vs. Numerology
Best video clip ever

http://www.youtube.com/watch?v=d1IzNKIHhp0&feature=fvst

KingOfCheese
01-06-2011, 10:55 AM
Following up on Dirac's post--anecdotal evidence is particularly dangerous in assessing statistics in a venue like this because there is a reporting bias. Let's assume 20 people in game have bothered to wonder about the dice and conducted a test. Of those, one, by random chance, will get a fairly dramatic anomalous result. The 19 that got the "normal" result will tell themselves, "oh well, I guess things aren't biased" and will quietly go on with the game play. The one person with the extremely lucky or unlucky result (a random effect), will come to the forums and post their unusual result. Over time, the forums become cluttered with unusual results, making unusual results look common--when in fact, it is a reporting bias (the people with normal results don't come to the forums and talk about it).

I have yet to see any post that has convinced me that the game dice are anything but random. I have seen some extreme results posted. But where appropriately conducted follow-up tests have been carried out, they have all been debunked.

Yaga_Nub
01-06-2011, 11:13 AM
Look at the combat log in the image below:


Does anything about the rolls there strike you as being less than random?

Now, I'll be the first to admit that the kind of number clustering you see in the image works both ways, and sometimes you get 3 vorpals in a row as well.

Having said that, this is ridiculous. The chances for 3 1's in a row on a d20 is 1 in 8000, it happens way more than that. I've had the same number rolled 10 times in a row on a d100 roll as well. The chance for this happening is lower than the chance of winning the grand prize in the lottery two consecutive times in a row!!!

I know that logical random number generators tend to cluster over short instances, but games like this is the exact place where this kind of clustering is detrimental.

There is a way to have a real random number generator used (the one using atmospheric readings comes to mind). The problem is, that it costs money.
I think that it would be a worthwhile investment none the less.

Your thoughts now.

I've had over 20 straight vorpals in a Shroud part 1 more than once. True the confirmation rolls weren't 20s (and didn't need to be), but that number of 20s shouldn't happen but it did and does. That's the great thing about randomness ... it's random. Sometimes you're lucky and sometimes your not.

cwfergtx
01-06-2011, 12:00 PM
I have been playing pnp a long time and there are night that you cannot hit anything and every save you fail. One night I think I had about 15 failed saves on a 1. Then there are other nights where nothing can touch you and you are criting on all rolls.

RJBsComputer
01-06-2011, 01:39 PM
When running a fix set of random numbers like 1 thru 20, the number of number clusters you well get is going to be larger then if you used the fix set of 1 thru 100. If you were to eliminate the number clusters, you would get 1 thru 20 in twenty rolls and then start all over again. The RNG creates a random number, that random number then is made to fit into the D20 dice set. In a D20 dice set there is a 5% chance of getting the number you want on each roll.

This is not a lotto drawing where once the number is drawn, it can't be drawn again. On the next roll, the whole number set is up for grabs again.

Cyr
01-06-2011, 02:08 PM
OP number clustering is a way to indicate weighted distrubitions.

However, the easiest way to show a high degree of confidence that a certain dice roll is less likely then another (and not just that a streak of a certain number ie cluster) is to have a as large a sample size as possible and add up the total rolls of each number.

Draccus (if I remember correctly) did this with UMD checks to demonstrate the odds of rolling a 1 are weighted (at least as far as UMD checks go).

I would be very interested to see this same test being done on saving throws made by players as well as attack roles. My feeling has always been from what limited testing I have done that the distribution is not flat and is actually weighted on the high and low sides for these rolls.

KraahgDaAxe
01-06-2011, 02:26 PM
I don't think that only one instance of a random number generator will be able to support the games needs.
I would guess that there is one per character, mob, or any other relevant game object. Makes more sense to me.

Actually multiple die rollers would be far more detrimental to the server. Plus it would effectively limit the number of "users" that can roll dice altogether. It has to be from a base source per server.

Kraahg

Xilraazz
01-06-2011, 03:25 PM
just a couple nights ago playing PnP i rolled 4 9's in a row on a d20. using the exact same die.
how's that for random...

elraido
01-06-2011, 03:33 PM
If it is so random....why do I only roll ones when it matters....and 20's on vorpals when the creature was going to die on that swing anyway? :D Cursed I tell you, cursed!!!

cedad
01-06-2011, 03:57 PM
As for the number of roll you would have to test to have the beginning of an idea about the randomness of the clusters.
You are looking for series of 3 numbers each of which has 1/20*20*20 = 1/8.000 chance of happenning.
In order to has a signinficant result you would need something with about a thousand of each occurence therefore you would need somewhere in the 8.000.000 results (considering that a 6 number sequence inhalt 4 series of 3 numbers).
strict minimum would be around 40.000 for 5 opportunities of each part (minimum needed for Khi Square application) if everything is perfectly randomised on your sample...

Oh, and one last thing random does not mean everything has the same probability...

stockwizard5
01-06-2011, 04:20 PM
I would point out that I there was discussion/instances back in the day about something (lag, etc.) causing rolls not to update and there were some rather long strings recorded. I'll look around and see if I can find some of that data as I remember running some math that showed there was in fact an issue.

That said - I have no doubt the generator is fine.

Quarterling
01-06-2011, 04:28 PM
There are two ways of doing random numbers.

One is using a uniform random distribution, and the other is using a Halton sequence.

What you just witnessed is proof that Turbine is using a uniform random distribution, because the Halton sequence always makes sure that all random values are distributed fairly evenly to give the illusion of randomness. This is not true in the real world however, because it IS possible to roll let's say, a one on a d100 over fifty times in a row. This would never happen in the Halton sequence.

Siskel
01-06-2011, 04:29 PM
The roles are random because you cannot predict the outcome of the next role. The distribution appears to be uniform given the tests others have run indicated earlier in the thread. The only thing I did not see was testing the distribution in multiple dimensions. Not entirely sure if it is even necessary to test in multiple dimensions given how the RNG is being used in this case. There might be some game mechanics that would require the distribution to be tested in multiple dimensions. It has been over a decade since I worked on anything related to random numbers, so it is entirely possible I am wrong.

SiliconShadow
01-07-2011, 05:08 AM
The difference between the two is so small that you probably will not pull the shorter straw because of it in a lifetime.

Otherwise, what you are saying is that the program repeats 'small' seeds. As far as i know smallest seeds range in the 2^32 order of magnitude. Even if the game would use the worst prng out there, you cannot possibly see it in 213 results, nor in 14 months of playing:)

Just for the numbers, in the above example you would have to make 118 rolls every second for 14 months 24/7 to have a full sample. And erm, recording it, not FEELING about it:)

I repeat, 118 rolls/s for your 14 months of continous gameplay.

Sorry, I don't see your reasoning either obvious, or true. Quoted samples are just too small for that.

That isn't what I am saying at all you are quoting maths I am talking about system restraints your reasoning is incorrect for the environment.