Dice Complaints and Various Suggestions to Fix Them

[Evan Thomas]

new dice generator- are you sure its random? ive conquered a couple of terits out of about 10 turns since the new generator came.
Are you sure They are random?

[RecklessB]

I'm pretty sure there is no such thing as a 'truly' machine generated random number. However, there are some pretty good solutions out there that are pretty damn close. For example this site offers next to perfect random numbers using atmospheric noise.

http://www.random.org

There is also the "Mersenne Twister" which can be incorporated into any dice rolling algorithm quite easily.

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

I certainly hope that CC is using something more robust than the standard "rand" function that comes with most languages.

[RecklessB]

Um guys, I looked up this thread because I just lost 19-9 and 18-5 in back to back games. However, I'm now 100% certain CC has a damn good dice generator after reading through the thread posted above by spiesr. Anyone who doubts the dice randomness on CC should at least take the time to read through that thread.

[e_i_pi]

ok, i just played 16 on 6. lost 15 and he lost only 1 so at the end i was left with 1 and he survived with 5. tell me, what stream of luck would it have to be in real life fr something like THAT to happen?

About 1/20 of a percent. It should happen once every 16000 rolls. That sucks...

I think it's 1 in 2200, which isn't that rare.

[Gunner1980]

yeah, but when i one day u loose a game cause of this, and then another same shitty pattern on a tourney game, and in another game u watch a man attack 17 with 9 (just to be clear - attacker had 9) and manages to take defender down to 6 (6-6 was the outcome)... and in the same time i am afraid to go 9 on 3 lol

[hecter]

I'm pretty sure there is no such thing as a 'truly' machine generated random number. However, there are some pretty good solutions out there that are pretty damn close. For example this site offers next to perfect random numbers using atmospheric noise.

http://www.random.org

*cough cough* That's what CC uses *cough cough*

[waynef100]

lets be real... a dice file is not random dice.... especially when the roll (and the rolls before and after) are repeated several times per day. get a real random dice generator it will take about 1 line of code... and dont kid yourself... 1M rolls per day is NOT CPU intensive. if you want ill donate my commodore 64.

[hecter]

lets be real... a dice file is not random dice.... especially when the roll (and the rolls before and after) are repeated several times per day. get a real random dice generator it will take about 1 line of code... and dont kid yourself... 1M rolls per day is NOT CPU intensive. if you want ill donate my commodore 64.

Are you high? The dice comes from random.org, have been looked at by a trained statistician, and the data has been backed up by numerous people recording thousands of dice rolls with a 3rd party greasemonkey userscript, all stating that the dice are, infact, random. Do you have any hard data to back up your outlandish claims?

[waynef100]

Please explain how conquer club implements random.org. I think you'll find my "outlandish" claims inside your response... assuming you know how conquer club ACTUALLY uses random.org. I assume you do not.

[hecter]

Please explain how conquer club implements random.org. I think you'll find my "outlandish" claims inside your response... assuming you know how conquer club ACTUALLY uses random.org. I assume you do not.

Actually, I do. Random.org gives them a "dice file". This dice file contains hundreds of thousands of rolls in it, and as people roll, CC takes the numbers out of the dice file. I'm not sure how often it is, but after a set amount of time (a day or couple days, I'm not sure) random.org gives CC a new dice file and the old one is discarded. Pretty simple stuff. If you do a search you can find out the exact details, as they have been posted publicly.

[waynef100]

Very nice you have a small level of knowledge. Now... get the facts before you say my claims are "outlandish" or that i am "high."

How big is this file?
How is the file read (sequentially or randomly)?
How many times per day is it read?
How often is this file replaced?

I have set you on the path to knowledge. You are right about one thing... its all publicly available. You just are still lacking the knowledge of the answers but its always refreshing to come across someone who thinks they know answers but then admits they don't. Why don't you answer those questions above... any intelligent person can deduce where I am going with this.

[hecter]

See, but the thing is that I know the dice are random. I've seen it in the past, the dice file is very large (somewhere around 500'000 lines), it's read about 2 a day, (last figure I saw was 800'000 lines a day), and it's replenished daily. But, I'm not positive, which is why I told you to not be so lazy and look for yourself. How else do I know the dice are random? Well, I've looked at the random.org website, seen the graphs that show CC's figures compared to "true randomness" (if there is such a thing) and they are pretty much identical, and I've seen many people posting their findings with the greasemonkey (called dice analyzer, btw, you can find it in the plug in and addons forum), as well as collected data with it myself. I've also yet to see you, or anybody else who claims the dice are not random, post one iota of data to support their claims.

[waynef100]

why did lackattack stop posting the dice file publicly?

brilliant! your answers begin to show conquer club's flaw. knowing that the file is of FINITE size and re-read MULTIPLE times daily AND sequentially... can you think of any way to exploit this?

if your answer is NO then you do not have sufficient statistics knowledge to continue our conversation and you probably think that card counters are wasting their time just because they cannot predict the next card.

if your answer is YES then my point is made and i ask you to detail how you can exploit such a fact.

ps. i am not lazy. i have done all the research and then some. you continue to show that you have not.

[hecter]

He never did post the file publicly Yet you're using me as a source for all of your information, yet I have clearly stated that I don't know the answers for certain. So, get up off your lazy ass and do some research for yourself. If all your basing these assumptions on is "that guy over there who I just said doesn't know what he's talking about, and he agrees with that statement" then you have a really really strange way of going about things... Not to mention the fact that you STILL, and this is the third or fourth time I've said it, haven't posted any evidence to support your claims.

[waynef100]

yes he did post that file publicly... so you cant answer my questions and your lack of knowledge is even more apparent. i do have the answers to my questions but i am trying to teach you to fish... not give you a fish. do some searches... youll find the same answers i did. if you have any math or statistics skills an obvious flaw in the design is apparent. i do have the answers to the questions i posed... you do not and you refuse to get them on your own. im sorry for you.

on the bright side the solution to conquer club's flaw is extremely simple:

either 1) randomize each dice roll on its own (no dice file)

or 2) randomize each read from the dice file (non-sequential reads)

but will lackattack implement such a solution?

[cspare]

why did lackattack stop posting the dice file publicly?

brilliant! your answers begin to show conquer club's flaw. knowing that the file is of FINITE size and re-read MULTIPLE times daily AND sequentially... can you think of any way to exploit this?

if your answer is NO then you do not have sufficient statistics knowledge to continue our conversation and you probably think that card counters are wasting their time just because they cannot predict the next card.

if your answer is YES then my point is made and i ask you to detail how you can exploit such a fact.

ps. i am not lazy. i have done all the research and then some. you continue to show that you have not.

I'm assuming the dice file doesn't actually contain pre-calculated dice results, but something like a seed for the randomizer.
If you don't know the contents of the dice file and have no knowledge about how CC uses the file then it would be incredibly difficult, if not impossible, to exploit it. Making the file public would be an open invite to exploit, but I don't see a way now. Do you? One exception would be data mining a huge set of dice results, but as you only have half a day before the seed changes thats no option.
Also you emphasize on that CC is reading the file sequentially, what's wrong with that? If you wouldn't read sequentially, that would leave random as the only option, right? But at that point there is no real random, that was the problem wasn't it?

[waynef100]

I'm assuming the dice file doesn't actually contain pre-calculated dice results, but something like a seed for the randomizer.

unfortunately you assume incorrectly. as you will see in previous forum threads the file contains the dice rolls themselves. it is a 5 column file. each row contains a number 1-6 that was randomized by random.org.
ie
1 2 5 6 2
2 3 4 1 4
etc

youre definitely seeing the light. "data mining" tells me youre on the right track

[hecter]

yes he did post that file publicly... so you cant answer my questions and your lack of knowledge is even more apparent. i do have the answers to my questions but i am trying to teach you to fish... not give you a fish. do some searches... youll find the same answers i did. if you have any math or statistics skills an obvious flaw in the design is apparent. i do have the answers to the questions i posed... you do not and you refuse to get them on your own. im sorry for you.

And you're inability to argue a point is glaringly apparent. Don't you remember having to write all those stupid essays for English class? Where you had to argue your thesis statement? Situations like this is why you had to do that. So far, I believe (along with many, many others) that the dice are random. You say they aren't. You're trying to sell us a product, so to speak. Convince us that your point is the right one. And so, you have to tell us why. It's not that hard to do. But you gotta do it. Otherwise you just look like an idiot running his mouth off. Provide us with the proof. Don't tell us to "go look for ourselves" because it's not going to work, we know (in our minds, at least) that the dice are random. Why should we look? We already have looked and came to that conclusion. So, either provide us with some actual proof, or get out because you're wasting every bodies time.

[spiesr]

Actually there is no one dice file. A new file is generated when the old one is almost used up.

[waynef100]

Actually there is no one dice file. A new file is generated when the old one is almost used up.

this would solve the problem as i stated earlier. can you please provide evidence or a link to such information?

hecter... why would they stop using a dice file if it weren't *gasp* random! have fun writing your essays. let me know when you've finished statistics class. lol

[hecter]

Ya, you're high...

[waynef100]

Ya, you're in junior high...

[waynef100]

lol. http://www.conquerclub.com/forum/viewtopic.php?f=4&t=628.... why doesnt dice.zip work anymore lol

[e_i_pi]

How has this thread gone under my radar? I shall attempt to answer your questions as best I can waynef100. I did the analysis on the dice file, so I am very familiar with it, the way it is accessed, and so on.

How big is this file?

The file I did the analysis on was 500000 lines, which ended up being 600Mb of analysis files. So the file is at least 500k lines long, depending on whether I was afforded a snapshot of the file, or the entire file.

How is the file read (sequentially or randomly)?

Sequentially

How many times per day is it read?

That's not information I have access to, I could probably work it out, but I'd have to request site stats

How often is this file replaced?

That's not information I have access to, but I assume it is not replaced

why did lackattack stop posting the dice file publicly?

As far as I know, it has never been posted publicly, though I have only been here about 8 months.

Knowing that the file is of FINITE size and re-read MULTIPLE times daily AND sequentially... can you think of any way to exploit this?

I have access to the file itself, and even I cannot exploit it. Assuming perfectly even distribution of all combinations, and an already made 3v2 attack (showing all 5 dice) that would place the current position of the dice file in one of 64 places. The file itself is accessed by everyone, so my next roll will not necessarily be the next line, it may in fact be 100 or 1000 lines down the file. Predicting where I am in the file, and what the next roll is likely to be (2W, 1W1L, 2L) is not possible, as the file position is directly affected by real-world 'random tampering' (ie - an unknown number of people making an unknown number of rolls in any given period of time).

if your answer is NO then you do not have sufficient statistics knowledge to continue our conversation and you probably think that card counters are wasting their time just because they cannot predict the next card.

For starters, card counting doesn't work in this game, because every time a card is generated, it is given a random colour. There may be 'booking out' cases occurring on province names, but not on colours.
Also, my answer is 'NO', and I do have sufficient statistics knowledge, at least that's what the Dean of UoW thought when he handed me my BMath degree.

on the bright side the solution to conquer club's flaw is extremely simple:
either 1) randomize each dice roll on its own (no dice file)
or 2) randomize each read from the dice file (non-sequential reads)

I did an analysis on the streakiness of the dice, the distribution of number of pips for both A and D dice, and success of attacks x vs y, where x ε {1,2,3}, y ε {1,2}. There was no significant deviation from the norm on any particular analysis. Ergo, there is no need to "fix" the process.

[waynef100]

thank you for your thoughtful answer!

so assuming you have the dice file and can predict roughly where in the dice file conquer cub is currently (very easy to do).... it *is* possible for someone to perform say... a 60 army attack at a fairly "quiet" time on conquer club and receive the EXACT SAME dice rolls for another 60 army attack later during another "quiet" time.

i ask you what the odds of that (2 60-army attacks generating the exact same sequence of rolls) happening in real life are ... i propose that number in real game play is very very very low.... but almost 1 on conquer club if you are aware of the current position in the dice file. on the second attempt you may not get all 60 rolls the same but you will get a large number the same including back-to-back-to-back-etc similar rolls.

since this behavior does not really exist in real life but can be replicated on conquer cub i conclude the dice do not exhibit real-world behavior.

i understand you did analysis on the dice file but this file is replaced at unknown intervals. did you analyze the new one? clustering is a statistical fact and streaks will exist. maybe not in the file you examined but maybe in another it exists a lot.

just like counting cards (in blackjack)... youre not predicting the next card... youre predicting the probability of good cards.

why is it so unreasonable to just read from the dice file randomly (not sequentially)? you can have a much smaller dice file... about 7K lines total.

why is it so unreasonable to generate random dice on-the-fly?

the current design lends itself to be exploitable. im just trying to help