To The New Owner of CC

[Dukasaur]

Except we know that the same sample of dice is being re-used over and over.

Incorrect. We don't "know" that. It's a theory that's been passed around, and I believe the theory is incorrect.

... The "haven't been replenished" allegation is pure nonsense. The skew is pretty definitely proven, but it's far more likely to result from a rounding error than from he alleged non-replenishment.

How can you call it nonsense? A mod has already admitted that the "haven't been replenished" allegation is a possibility that would account for the results we're seeing:

We were told that the list is replaced every hour. So one possible failure mode is if, for some reason, the list is no longer updating and got stuck on a list that was particularly non-uniform. But if the lists are still updating once per hour, then your explanation would require the sum total of those lists to be non-uniform.

His second sentence essentially confirms that any other explanation is extremely unlikely. A rounding error does not make any sense because there is nothing to round. The numbers they get from random.org are integers (1-6), and I've already tested numbers directly from random.org and found no bias. Also, the skew is not just to the right, - it skews away from 1's, and towards 2's and 4's. That could not be the result from a simple rounding error.

That is not true.

Let's take your points one at a time.

First, the idea that Mets has some unique knowlege of the process. He's a mod, and a Team Leader, and so am I. We're not given the keys to the inner office. The workings of the core engine are pretty much strictly webmaster territory. Now, Mets works on the technical side of it and knows the stuff a little better than I do, but he's still guessing. And granted, he's a graduate student in astrophysics and a pretty smart cookie, but still, he's guessing, just as you and I are. And anyway, we're dealing with pretty basic high school algebra here, so even a rube like me can probably cope.

Second, the idea that random.org provides integers. Incorrect. If you and I go to random.org and ask for 100 or even 1000 integers, then yes, we will get them. That's because those are small numbers and the cost of crunching the numbers for you is small enough that they can afford to take the hit. But when you buy numbers from random.org in commercial quantities, they come in byte form (See: http://www.random.org/files/) and you have to convert them to integers yourself.

That's pretty easy if you want binary numbers. If we were playing RPGs here and rolling a lot of D4s and D8s, that would be great. But CC plays with D6s, and bytes are not easily converted to D6. A byte gives you a discrete value from 0 to 255, ergo 256 discrete values, and 6 does not go into 256 evenly. If it did, it would be a ludicrously simple calculation to do bitewise modulus division on the byte and come up with nice D6s.

Alas, 6 = 3 x 2 and there is no easy way to divide 3 into any binary root. Oh! the 3, that Buddha, that Janus! Such a godlike blessing in geometry and such a demonic curse in arithmetic. Any process of division by 3 results in a repeating decimal and a guaranteed source of error.

Furthermore, if you do want to divide integers by 3 and start getting into decimal fractions, you're talking a massive hit on the servers. Again, not a big deal when you're processing 100s or 1000s, but when you need to roll millions of virtual dice every day, it's probably enough to force a more powerful processor and boost the server cost from $1500/month to $2000/month. Now, to Exxon that might be pocket money, but if you're a small business with gross annual revenues in the $125,000 range, you have to pay the customary rake-offs to the government (ie taxes) and the banks (ie. processing percentages on online payments, which are not cheap, let me tell you!) pay three staff members, and somehow still find the money to keep your servers running, an extra $500/month is a really big deal, so if I was lack/Jefe/Wham, I would avoid floating-point operations at all costs, and go with strict bitwise math based on the discrete values in the byte.

What are some of the tricks we could pull to squeeze a number divisible by 6 out of 256? Well, there's a whole pile of possibilities. I'm not going to bother analysing all of them, for two reasons. First, I really don't care all that much. Second, I'm only guessing, and there's a limit to how much time one should waste on guesses with no real facts.

You can take one byte, multiply it by 2, use one bit from a different bite to dither the succeeding value so that it contains odd numbers, and do MOD(6) on the result. Or you could do regular (non-modulus) division on the 255 and come up with segments of either 42 or 43 values in length, assigning a value between 1 and 6 to each segment.

Tantalizingly, if I ballpark the figures on your graphs, the difference between rolling a 1 and rolling a 2 is 1/((16.775/16.425)-1) which equals an error of 1/47, pretty damn close to the 1/43 error that I would expect by the segmentation method. That's what originally led me to believe (many months ago, when you first posted the graph) that this was a rounding error, not a randomization error.

Now that I've shown you a direction to look in, you'll probably be able to hunt around and find some formulas much better than the ones I've shown. I personally don't care. Why do I not care? First, because as you admit yourself, the bias is the same for everyone, so nobody loses or gains from this. There would only be a concern if the bias was somehow targetted against some subset of our members, which it isn't. Second, 95% of the dice bitch threads are made by people who complain about the streakiness of the dice, not the bias against ones, and streakiness is a perfectly normal feature of truly random numbers. People who think that just because they rolled twenty 5s in a row, they are now due to roll something other than a 5, are engaging in the gambler's fallacy.

You've really done a disservice to the people engaging in the dice bitching. By showing that there is an error in terms of skewness, you've reinforced their paranoid delusions about streakiness, even though by your own admission the one has nothing to do with the other.

[Gilligan]

Some kid just rolled 15v32 against me and won.. Because that ever happens.

It happens 1.8% of the time. Clearly you were part of that 1.8%.

[RiffArt]

This is an interesting thread. From what I understand:

Section 1: There is a bias in the dice, specifically against rolling 1's. The bias is small. Unfortunately, most people seem fixed upon this bias and are missing the important point - how does it affect the outcome of combat? The dice themselves aren't important, only whether the chance of W/D/L 3v2 accurately reflects that chance if we were rolling actual dice. This does matter despite the odds being the same for everyone; since the "intensity cubes" are represented as dice, specifically D6's, - rather than, say, numbers picked randomly from the ether - a player coming across the site is entitled to expect the probabilities for the outcome of a battle to match those for rolling actual dice.

Using the information provided by degaston earlier in the thread I have analytically estimated some of the battle outcomes (i.e. not simulated):
1 Attacker vs. 1 Defender
Completely unbiased dice give the attacker a 41.666...% chance of winning, CC dice give the attacker ~41.6662%
2 Attackers vs. 1 Defender
Completely unbiased dice give the attacker a 57.870% chance of winning (125/216), CC dice give the attacker ~57.8558%
3 Attackers vs. 1 Defender
Completely unbiased dice give the attacker a 65.972% chance of winning (855/1296), CC dice give the attacker ~65.9571%
1 Attacker vs. 2 Defenders
Completely unbiased dice give the attacker a 25.463% chance of winning (55/216), CC dice give the attacker ~25.4485%

I did not estimate 3v2 because it takes longer and they are just estimates since I don't have direct access to the data degaston provided, I have estimated the probabilities to roll each number from the graph provided then normalised so that the probabilities sum to 1. If it becomes important I'd be willing to provide more accurate estimates and for all battles if given the data. As it stands, it looks like the dice cause the attacker to lose in the order of 1 battle in 10,000 more than he ought. From my own dice stats I can estimate a player has 60,000 battles in 100 games, so this may affect one battle in every 15 games or so (as attacker, so it costs you troops, or as defender so you benefit from the anomaly).

Those are very rough estimates atm since by far the most battles are 3v2. In fact I'll try and improve on these numbers.

If these estimates are correct, would such an error be noticeable to a human playing, i.e. will it change our tactics, or reduce our enjoyment? No.

Section 2: As for how the "random numbers" are generated, everything posted so far is just supposition: if the 50,000 dice are not refreshed it would explain the small bias. The quote provided by Leehar does support this theory

This is how the intensity cubes now work:

• We have a series of 50,000 high quality random numbers from random.org
• Each time the game engine generates a random intensity cube, the next number is read in sequence from the series (e.g. in a 3v1 attack 4 numbers are read sequentially)
• When the last number in the series is read, we "rewind" and continue with the first number in the series

Emphasis mine

If this is the case (non-refreshing), the bias is unavoidable, unless the number of occurrences of each 1-6 is equal in the sample which is unlikely without the numbers being manipulated in some way. (Impossible with 50,000 dice unless some are discarded)

Tantalizingly, if I ballpark the figures on your graphs, the difference between rolling a 1 and rolling a 2 is 1/((16.775/16.425)-1) which equals an error of 1/47, pretty damn close to the 1/43 error that I would expect by the segmentation method. That's what originally led me to believe (many months ago, when you first posted the graph) that this was a rounding error, not a randomization error.

If the numbers are provided as bytes (i.e. 2^8=256 possibilities), then the error from simply calculating mod6 is at most 0.26% for each number i.e. |Prob(n)-(1/6)|<0.2605% for n=1,2,..6. note: degaston's data makes this possibility unlikely - only 1 shows significant bias whereas there should be 2 numbers that fall below what's expected. However it could be that this error combines with errors from not refreshing the numbers. I'm not a programmer myself but the simplest solution that comes to mind would be to discard 4 of the possibilities (e.g. bytes 00000000-00000011) and then calculate modulo 6.

Finally, to anyone quoting a bad roll or streak as evidence: please understand that no lucky roll or streak is evidence of a fault with the dice.

To anyone using an argument from authority (including the OP) without supporting their point of view with actual evidence, please don't:
i. this is the internet, people make things up so we automatically assume you're lying;
ii. even if you're not, it's unlikely that you are the most intelligent or qualified person on the site;
iii. arguments are supported by facts and evidence. Anything else is superfluous.

[-Maximus-]

Roller coaster dice

[iAmCaffeine]

Some kid just rolled 15v32 against me and won.. Because that ever happens.

It happens 1.8% of the time. Clearly you were part of that 1.8%.

Stop throwing numbers at me!

[TeeGee]

Some kid just rolled 15v32 against me and won.. Because that ever happens.

It happens 1.8% of the time. Clearly you were part of that 1.8%.

Stop throwing numbers at me!

What can we throw at you then? Looks like a friendly statistician has done the maths for us and come up with - there is no advantage or disadvantage with the CC dice and numbers generated. Maybe the atmospheric noise at random.org is biased against 1's. How can we fix that? maybe start purchasing strings of numbers 2-7 and subtract 1?

[iAmCaffeine]

Some kid just rolled 15v32 against me and won.. Because that ever happens.

It happens 1.8% of the time. Clearly you were part of that 1.8%.

Stop throwing numbers at me!

What can we throw at you then? Looks like a friendly statistician has done the maths for us and come up with - there is no advantage or disadvantage with the CC dice and numbers generated. Maybe the atmospheric noise at random.org is biased against 1's. How can we fix that? maybe start purchasing strings of numbers 2-7 and subtract 1?

Throw coffee!

Or maybe fix the recurring sample of dice?

[degaston]

Except we know that the same sample of dice is being re-used over and over.

Incorrect. We don't "know" that. It's a theory that's been passed around, and I believe the theory is incorrect.

Though we do know that the code is written to re-use the same set of numbers over and over until they are replaced.

... A rounding error does not make any sense because there is nothing to round. ... Also, the skew is not just to the right, - it skews away from 1's, and towards 2's and 4's. That could not be the result from a simple rounding error.

That is not true.

Let's take your points one at a time.

First, the idea that Mets has some unique knowlege of the process. He's a mod, and a Team Leader, and so am I. We're not given the keys to the inner office. The workings of the core engine are pretty much strictly webmaster territory. Now, Mets works on the technical side of it and knows the stuff a little better than I do, but he's still guessing. And granted, he's a graduate student in astrophysics and a pretty smart cookie, but still, he's guessing, just as you and I are. And anyway, we're dealing with pretty basic high school algebra here, so even a rube like me can probably cope.

I was not saying that Mets agreed with my theory that the numbers aren't being replaced, or even that I'm sure I'm right. Merely that a mod (who happens to be titled the "Head Thinker") has admitted that it is a possibility, and therefore it is not "pure nonsense" as you claimed. I will say that I have not seen anyone provide evidence to refute my claim, or provide any more logical explanation for the results that we're seeing.

Second, the idea that random.org provides integers. Incorrect. If you and I go to random.org and ask for 100 or even 1000 integers, then yes, we will get them. That's because those are small numbers and the cost of crunching the numbers for you is small enough that they can afford to take the hit. But when you buy numbers from random.org in commercial quantities, they come in byte form (See: http://www.random.org/files/) and you have to convert them to integers yourself.

Actually, it can provide integers. The generateIntegers method (https://api.random.org/json-rpc/1/basic) gives you up to 10,000 integers at a time in whatever range you want. Calling that 5 times and merging the results into one 50k digit dice file would be trivial. If you purchase bits in bulk, the cost drops to 4 million bits for $1, which would provide more than 1 million rolls. There is really no need for CC to be attempting the conversion from bits to dice rolls, when random.org will do it for you. If they are, then they've wasted a lot of effort writing the code, and a lot of processor time doing the conversion.

One interesting thing I noticed from the guidelines for writing automated clients is:

3. Use a long timeout value for your requests. Unless you have used up your quota, the RANDOM.ORG server actually tries to satisfy all requests, so if you use a short timeout value, your request will be abandoned halfways and the numbers discarded. This increases load on the server. Allow at least a couple of minutes for the server to complete your request.

So they may have set a timeout that worked when it was written, but if random.org response times have slowed since then because of higher demand, it could cause their request to never be fulfilled.

The link you provided was for a file that "contains daily true random numbers generated by RANDOM.ORG in the past." These files are generated once per day, and contain 2^23 bits. That's enough to generate 2 million dice rolls, and would last them for about 2 days. If they were using these, then why would they generate dice files containing only 50,000 rolls? And why would they talk about getting new data from random.org every hour? From everything I've seen in the discussions, it makes a lot more sense that they're trying (but failing) to get fresh sets of integer dice rolls every hour, and there is no need for them to worry about converting bit values to 1-6 dice rolls, so rounding error should not be a factor.

... 95% of the dice bitch threads are made by people who complain about the streakiness of the dice, not the bias against ones, and streakiness is a perfectly normal feature of truly random numbers. People who think that just because they rolled twenty 5s in a row, they are now due to roll something other than a 5, are engaging in the gambler's fallacy.

You've really done a disservice to the people engaging in the dice bitching. By showing that there is an error in terms of skewness, you've reinforced their paranoid delusions about streakiness, even though by your own admission the one has nothing to do with the other.

Yes, shame on me for pointing out a problem with skewed dice generation, because some people use that to claim that there are other problems as well. The dice bitch threads were around long before I got here, and I'm sure they'll continue long after I'm gone, so I don't think you can blame them on me. Prior to the Which number do you roll the most? thread, the only reason I looked at them occasionally was to laugh at the stupidity, and I had no reason to assume that the dice were anything but random.

But now that it is apparent that there is one problem, who's to say that there are not others? I do not have access to the dice data file, so there is no way that I can test it, but if the dice file is not being replaced, as I suspect, then that could lead to more problems than just skewed rolls.

Suppose there is an unusually long streak of the same digit within that file? A streak that normally might occur only once a month could now happen 20 times a day. Or perhaps there is a sequence of three low values followed by two higher ones (or vice versa) that repeats several times. Whenever the program gets to that section, there is a 1 in 5 chance of someone going on a very hot or cold streak.

I find it somewhat disturbing that no one at CC is even interested in looking into this, because where there is one identifiable problem, there could easily be others that are less obvious, but more significant. Once I have seen that someone has explained and corrected the situation, I will be happy to go back to ignoring the dice bitch threads.

[Dukasaur]

Second, the idea that random.org provides integers. Incorrect. If you and I go to random.org and ask for 100 or even 1000 integers, then yes, we will get them. That's because those are small numbers and the cost of crunching the numbers for you is small enough that they can afford to take the hit. But when you buy numbers from random.org in commercial quantities, they come in byte form (See: http://www.random.org/files/) and you have to convert them to integers yourself.

Actually, it can provide integers. The generateIntegers method (https://api.random.org/json-rpc/1/basic) gives you up to 10,000 integers at a time in whatever range you want.

You can only do that a finite amount of times before they will block you and insist that you open a commercial account with them. They don't specify how many times, so I suspect it's flexible and negotiable, and they're probably lenient with non-profits and educational uses, but I'm sure a commercial enterprise that uses 1,200,000 numbers a day would be required to start paying.

Calling that 5 times and merging the results into one 50k digit dice file would be trivial. If you purchase bits in bulk, the cost drops to 4 million bits for $1, which would provide more than 1 million rolls. There is really no need for CC to be attempting the conversion from bits to dice rolls, when random.org will do it for you. If they are, then they've wasted a lot of effort writing the code, and a lot of processor time doing the conversion.

You're contradicting yourself. You say that in order to get the bulk price you have to buy you numbers in bits (by which I assume you mean bytes, or at least some form of binary file). Then you say that they will convert the number to integers for you, but one can safely assume that there would be a significant extra charge for that.

No binary number will ever be evenly divisible by three. Base two and base three just don't conveniently mix. You will need a conversion of some kind, whether you do it yourself or whether you pay random.com to do it for you. You just can't squeeze D6's out of bits without a conversion of some kind, and whether you pay the cost at your own end by processing it yourself, or whether you pay the cost at the acquisition end by paying random.com to produce integer files instead of binary files, you will pay it.

[degaston]

Actually, it can provide integers. The generateIntegers method (https://api.random.org/json-rpc/1/basic) gives you up to 10,000 integers at a time in whatever range you want.

You can only do that a finite amount of times before they will block you and insist that you open a commercial account with them. They don't specify how many times, so I suspect it's flexible and negotiable, and they're probably lenient with non-profits and educational uses, but I'm sure a commercial enterprise that uses 1,200,000 numbers a day would be required to start paying.

Do you have any evidence to back up your claims here? Other than not being able to get 1,200,000 rolls per day for free, everything you've said seems to be either pure speculation, or is contradicted by the random.org website. http://www.random.org/quota/

Every IP starts with a quota of 1 million bits. If you use some, you get up to 200,000 bits added to your quota each day, until you're back to 1 million. I've seen nothing on the website that says you couldn't get 200,000 bits (~75,000 dice rolls) every day for free. This obviously would not be enough for CC, so a little later in the paragraph I mentioned that you can pay to increase your quota.

Calling that 5 times and merging the results into one 50k digit dice file would be trivial. If you purchase bits in bulk, the cost drops to 4 million bits for $1, which would provide more than 1 million rolls. There is really no need for CC to be attempting the conversion from bits to dice rolls, when random.org will do it for you. If they are, then they've wasted a lot of effort writing the code, and a lot of processor time doing the conversion.

You're contradicting yourself. You say that in order to get the bulk price you have to buy you numbers in bits (by which I assume you mean bytes, or at least some form of binary file). Then you say that they will convert the number to integers for you, but one can safely assume that there would be a significant extra charge for that.

I wasn't contradicting myself, I was using the terminology that random.org uses. Bits are their basic unit for data, and it seems to be their preferred term when talking about file sizes, pricing, quotas, etc. For example, when talking about the pre-generated files you linked to:

Each file on this page contains 8,388,608 bits (one mebibyte) of raw random data.

and when talking about pricing:

Buy once-off top-up worth: 600,000,000 bits for US $150 (approx. €110.4) (4 million bits per dollar)

This does not mean that you can only receive data from them in a binary format. For example, if you use the generateInteger method to request 6 dice rolls, your response format will look like:

Code: Select all

{
    "result": {
       "random": {
            "data": [
                1, 5, 4, 6, 6, 4
            ],
            "completionTime": "2011-10-10 13:19:12Z"
        },
        "bitsUsed": 16,
        "bitsLeft": 199984,
        "requestsLeft": 9999,
        "advisoryDelay": 0
    },
    "error": null,
    "id": 42
}

They do not mention any extra charge for converting the bits to an integer range, so what are you basing your assumption on that they do? If you're interested in how the conversion is done without introducing a bias, you can read http://mathforum.org/library/drmath/view/65653.html. In the example above, they used 16 bits to generate 6 rolls (2.667 bits per roll), which is pretty close to the theoretical minimum of 2.585 bits per roll.

No binary number will ever be evenly divisible by three. Base two and base three just don't conveniently mix. You will need a conversion of some kind, whether you do it yourself or whether you pay random.com to do it for you. You just can't squeeze D6's out of bits without a conversion of some kind, and whether you pay the cost at your own end by processing it yourself, or whether you pay the cost at the acquisition end by paying random.com to produce integer files instead of binary files, you will pay it.

... unless random.org will do it for free, in which case there is no need for you pay any extra, or to do it yourself. CC might be able to squeeze out a little more efficiency by optimizing the algorithm for large quantities of dice rolls, but why would they do that just to save a few cents each day? If they really want to save money, it would be much easier for them to use their daily quota with a randomness extractor (http://en.wikipedia.org/wiki/Randomness_extractor) to generate all the rolls they need for free.

It would be nice if someone who knows (not that they read dice bitch threads) could confirm whether CC is trying to do its own conversion from binary data, or if it is trying to get dice roll integers directly from random.org.

[Dukasaur]

They do not mention any extra charge for converting the bits to an integer range, so what are you basing your assumption on that they do?

Simply on what you posted:

If you purchase bits in bulk, the cost drops to 4 million bits for $1,

If you buy your bits in bulk, the cost drops. The logical inference is if you buy it in some other form, the price was higher.

If I buy washer fluid in 1000 litre totes, I pay the bulk price. If I buy it in convenient 4-litre pour jugs, I pay a lot more.

Anyway, like I said, I'm just guessing, you're just guessing, and we don't know what contractual arrangement exists between them.

[degaston]

They do not mention any extra charge for converting the bits to an integer range, so what are you basing your assumption on that they do?

Simply on what you posted:

If you purchase bits in bulk, the cost drops to 4 million bits for $1,

If you buy your bits in bulk, the cost drops. The logical inference is if you buy it in some other form, the price was higher.

If I buy washer fluid in 1000 litre totes, I pay the bulk price. If I buy it in convenient 4-litre pour jugs, I pay a lot more.

Anyway, like I said, I'm just guessing, you're just guessing, and we don't know what contractual arrangement exists between them.

The difference is, they spell out their various volume discounts, but do not mention any fees for receiving your data in one format vs another, so there's no reason to assume that those charges exist.

So given that random.org can supply non-biased integer dice rolls from 1-6, is there any reason to think that CC would prefer to receive the data in a binary format and try to convert it themselves? I've seen plenty of bad programming decisions in my life, so I'm not saying it's impossible, but which do you think is more likely?

And if they are receiving integer dice rolls, where is the bias coming from? They deny any attempt to "tweak" the data, and I have no reason to think that random.org data is biased, so where does that leave us?

Any single set of random data is almost certain to have a bias of some sort (It would be pretty suspicious if it didn't), and it may also contain some rare patterns. Normally, these would be offset by counter-biases in subsequent data, but if you use the same data set over and over, the bias will be reinforced over time, exactly as we are seeing.

[Dukasaur]

Any single set of random data is almost certain to have a bias of some sort (It would be pretty suspicious if it didn't), and it may also contain some rare patterns. Normally, these would be offset by counter-biases in subsequent data, but if you use the same data set over and over, the bias will be reinforced over time, exactly as we are seeing.

.
Maybe. The new administration has been making radical changes, so maybe this is something else that will soon change. We shall see.

In the meantime, the part I would like to emphasize is this:

I did some analysis earlier. If the dice were extremely biased so that they never rolled a 1, then the attacker's win rate (3v2) would drop from 53.96% (expected for normal dice) down to 51.616%. The CC dice are only slightly biased against 1's, and when I simulated 1 million battles with CC dice, the change in win rate was undetectable.

So I don't think it's valid for anyone to say that the dice are affecting their win rate - unless they want to claim that the emotional stress of knowing that the dice are biased is causing them to make poor strategic decisions. In which case, it could be argued that the people who don't know about this may actually be at an advantage because they aren't freaking out about it.

[Armandolas]

This is pretty simple.
Ok, dice is equal to everyone, imk not complaining . It is what we get, but dice results do not reflect on what could have been real life, so it is not random, not at all, it is programmed to achieve some dice results and it is programed to balance those results in a regular basis.
What is random about that?

Best ex is 1vs1 battles:
2480 3460 attacking
2166 1588 defending

So , how is this possible?Does it mean it is programed for defense have better results if u have an attack 1vs1?

You guys complaining about this thread cant just try to believe that intention of the OP is to change for the better and not just ranting about bad dice?
I dont complain much about dice, but if i could have a more "real" one i would be happy

But that made me think. Isnt it better for strategic purposes if i have a biased dice like the one we get now?
If we had a tless bieased dice would we have a noob fest with players practicing the auto assault frenzy and winning games without even knowing how?

[degaston]

This is pretty simple.
Ok, dice is equal to everyone, imk not complaining . It is what we get, but dice results do not reflect on what could have been real life, so it is not random, not at all, it is programmed to achieve some dice results and it is programed to balance those results in a regular basis.
What is random about that?

From what the mods have said, I don't believe there is any intentional manipulation of the dice. I think what is going on is the unintentional result of a programming error.

Best ex is 1vs1 battles:
2480 3460 attacking
2166 1588 defending

So , how is this possible?Does it mean it is programed for defense have better results if u have an attack 1vs1?

You have attacked 1v1 5940 times (2480 + 3460) and won 2480 of those attacks for a win rate of 41.75%
Your opponents have attacked you 1v1 3754 times (2166 + 1588) and won 1588 of those battles (your loss is their gain) for a win rate of 42.30%
The expected win rate for 1v1 attacks is 41.67%, so you're both doing slightly better than expected, but nothing out of line, given the small sample size.

You guys complaining about this thread cant just try to believe that intention of the OP is to change for the better and not just ranting about bad dice?
I dont complain much about dice, but if i could have a more "real" one i would be happy

But that made me think. Isnt it better for strategic purposes if i have a biased dice like the one we get now?
If we had a tless bieased dice would we have a noob fest with players practicing the auto assault frenzy and winning games without even knowing how?

The bias is too small to have any affect on what strategy you should use, but I would also prefer to have truly random dice.

[AslanTheKing]

But that made me think. Isnt it better for strategic purposes if i have a biased dice like the one we get now?
If we had a tless bieased dice would we have a noob fest with players practicing the auto assault frenzy and winning games without even knowing how?

wow, i havent had much chance of having a great laugh ( except laughting at my own posts)
but this one made me laugh big time

do you really think u won your games just because of your strategy?
forget it, your having a illusion
its only the dice, but only the dice, NOT YOUR STRATEGY

any kid can play doodle ( or know the basic of gameplay of doodle)
the dice make the game

and hold on, lets change doodle now to hive
anybody who has played hive with all settings in 100 games
will know the basics of that map too ( it will become with experience as easy as doodle, just a bit more complicated,
but driving a car was complicated the first day too)

or any map u can study,
and understand fully, and have a certain strategy
but if your dice dont want u to win, u lose, that simple

u are not a genius ( u have a higher rank, yes)
but your strategy is not the gameplay
its different
dont play 1 vs 1, dont play terminator, dont play bots
dont play against those u lost before ( foe them) .................... the list is long, u can continue

its the dice, some get shitty ones, like i do
( its statistically proven, just read through the posts, youll find it )

and sometimes u get brilliant ones, but i cant enjoy my wins
since i know i had great dice

and i want to try as much as possible, there is so much to explore which im not afraid of,
since i want to have some fun, so i skipp the seriosity about the dice, and try not to get frustrated
like in the early days when i joined, i just ignore the dice ( still they make me upset) but not furious anymore
i try to have fun ( with the bots)

[Armandolas]

Well i dont play doodle or luxembiourg because of that, it involves a lot of luck.
I never foed anyone and im not a genious.I play terminator i play 1vs1..i have 70 medals u have 30...not sure what are u talking about
Im smart enought to tell you that strategy wins games.Patience wins games.You dont seem to be smart enough to understand that.Its not about ranking,Ranking means nothing to evaluate the quality of a player.

IIve seen all your bitch threads about dice, and u dont seem to learn.
Ill leave you a simple question so u can understand that strategy wins games:
If u look at clans like TOFU or KORT they probably win 70% of their games all the time. Can you explain why that happens?

[AslanTheKing]

Ill leave you a simple question so u can understand that strategy wins games:
If u look at clans like TOFU or KORT they probably win 70% of their games all the time. Can you explain why that happens?

yes i scan
first these clans have a very good communication and are very active
those players are all very experienced and know the basics and the maps they have to play
and they know this shitty dice,
they know that attacking dice are better than defending dice
they know that the dice can screw up a game and they have enough brain to skip a round or two
(or killing their own clanmate...)
to get the advantage
in some maps with no spoils, it need only one advantage in round 3 to win the game
i know
the point i am trying to make is that the dice are still the winner, dice wins
in a shortime view
in longtime view
out of 10 games to have a strategy is better to have none
out of 10 games ( if u are experienced with settings and the map) with a decent strategy u will win 5
(unless u play complicated maps like bamboo jack or even that other chinese map, and your opp doesnt know, u win 10)
but in games where u know u win and get screwed by the dice
i get upset
if u dont, fine

[AslanTheKing]

i have 70 medals u have 30...

and so what
u joined in 2008
i joined 2011
u made your 70 medals in 6 years
i made mine in 3 years

and if i look at the games u played in your first days,
it seems to me u have been around here before
and started with a new name again
(maybe u have been bitching about the dice too much)

its not a normal pattern than u see of newbies

so u have been here before 2008

[degaston]

... the point i am trying to make is that the dice are still the winner, dice wins

So, the more successful players win because:

• They use communication to plan their strategy and cooperate with teammates.
• They form alliances with opponents when appropriate.
• They influence the decisions of other players through chat.
• They've learned the proper strategy to use on the maps they play.
• They understand that the attacker has a statistical advantage, and use that to help them win.
• They know advanced strategies, like killing a teammate to get cards and continue an attack.
• They use a small advantage properly to prevent an opponent from having a chance to make a comeback.
• They play logically, and don't let emotion cloud their judgement.
• They don't give up just because they're down.

BUT... It's not their strategy that causes them to win, it's just lucky dice!

[BigBallinStalin]

do you have proof the dice results are not supported by reality?

your dice stats suggest they are spot on.

No, his dice stats (and yours, and everyone else's) suggest that there is a bias against rolling 1's...




The reality is that there's a problem with the dice generation, but no one seems to care because everyone has the same problem, so it's "fair". But when an obvious problem like this will not even be acknowledged and corrected, people may assume much more serious problems with the dice, and denials lack any credibility because truly random dice are not a priority for the site. I know it's just a game, but it's also a business, and you couldn't get away with dice like this on any gambling site, or any other business where true random simulation is important.

Since I roll more 5s and 6s and less 1s and 2s on average, I would like CC to not change the dice system.


That is all.

[BigBallinStalin]

Blabla I have a major in Philosophy, blabla my son has a math major blabla.

With all those majors you should know that when you state something as a fact, you should have evidence to back it up.

Frankie, you were already in the large graph from the "what number do you roll the most" thread, but since then your percentage of 1's has only gotten worse. At 160k rolls, you're still on the low end for sample size.

Plenty of evidence has been given to show that the dice are not random, and I haven't seen anyone show any evidence to refute that.

I actively posted in that thread, and to the question:''are the CC dice random?'' I would answer a 100% no, the dice are everything but random.
Howhever, I do not care about it. The community proved that 1's are thrown less then other numbers, and that this ''pattern'' exist for every player on the website. IMO this has no priority to be fixed, I have been dice fucked a lot, but I am also fair enough to acknowledge that I have been lucky a lot of times.

The other dice thread: ''which number do you throw the most'' was a good thread in which the community investigated the dice and came up with a minor problem that should be fixed but doessn't influence the game alot.
This dice thread is a stupid no brainer dice bitch thread made by somebody who has no clue.

In short:
Are the dice random: No
Are the dice fair: Yes
Is this topic a whole lot of bullshit: Yes

Are the dice fair?

Look at degaston's graph of the 10 or so players. The outcomes differ, in that some people are rolling more 6s than others. Surely, there is some discrepancy in outcomes among the players, so there seems to be some room for some players to have it better than others, right?

[Armandolas]

and so what
u joined in 2008
i joined 2011
u made your 70 medals in 6 years
i made mine in 3 years

and if i look at the games u played in your first days,
it seems to me u have been around here before
and started with a new name again
(maybe u have been bitching about the dice too much)

its not a normal pattern than u see of newbies

so u have been here before 2008

I talked about medals not trying to show "how great i am". What i tried to say is that those medals reflect that i played a lot of different styles, not only those styles that makes me win more points.
Apart from that , even if i joined 3 years earlier we have the same amount of games played.
Ive never played CC before i open my only account, but yes ive played risk online before and played the board as well.

And coming to strategy again, in your previous post you said it all. Strategy wins games(u mentioned a lot of winning strategies) but u can never take out the important of dices because...guess what!?..you are playing a game that involves rolling them!!


And yes, i do get angry and winne about dice(many times with the "note to self" checked) and yes sometimes my dice is beautifull.

Ohhh...and im not sure if someone thinks that attacking dice is better(i doubt it)...what i know is that attacking gives u 3vs2 dices... thats the advantage..you should know that by now.

[betiko]

A long long time ago in a CC far far away I reached out to the creator of Random.org, Dr Mads Haahr. I explained to him the way the dice work at CC, and he was nice enough to respond back. He confirmed the way CC was picking numbers wasn't a truly random method of picking numbers, only pseudo-random. I tried to explain it to people here, but they thought they were smarter than Dr Mads Haar. I stopped fighting the morons as they kept dragging me down to their level and beating me with experience.

so dr mads haahr didn't find a better method than that BW?

[degaston]

A long long time ago in a CC far far away I reached out to the creator of Random.org, Dr Mads Haahr. I explained to him the way the dice work at CC, and he was nice enough to respond back. He confirmed the way CC was picking numbers wasn't a truly random method of picking numbers, only pseudo-random. I tried to explain it to people here, but they thought they were smarter than Dr Mads Haar. I stopped fighting the morons as they kept dragging me down to their level and beating me with experience.

so dr mads haahr didn't find a better method than that BW?

Nice necro bump.
But yes, getting new dice data from Random.org would be a better method. CC has not used Random.org for a long time. (AFAIK)