Arcane Wonders Forum
Mage Wars => General Discussion => Topic started by: Sailor Vulcan on February 01, 2014, 09:56:42 PM
-
So I just played against a Druid on OCTGN. During the game, I made two perfectly identical 5 dice attack rolls against the Vine Tree at critical moments: 4 blanks + 1 normal and a 6 on the effect die. I feel like this is astronomically unlikely to happen. While it is possible, I think the chances are probably higher for something going wrong with the program.
Here is the calculation of the probability:
(2/6)^8*(1/6)^2*(1/12)^2=
2.94011941 * 10^-8
In other words, the chances of this happening through pure chance is about 3 out of 100 million. While it is possible, I think it's more likely to be a bug. I think it's worth looking into.
-
http://octgn.gamersjudgement.com/wordpress/magewars/2013/06/22/randomness-in-mage-wars/
-
Ok. That doesn't change the fact that the chances of these rolls were 3 out of 100 million and therefore probably lower than the chances of a bug in the program. I'm not denying that it's possible to happen from pure chance, but that's astronomically unlikely. I suppose you don't have to, but I still think this really warrants at least looking into. As far as I can see, the chances of this being a bug likely dwarf the chances of it happening purely by chance by a VERY large margin. I mean, 3 out of 100 million...I don't think it's something that can be successfully handwaved without even looking at it and confirming that it actually was pure chance.
"As soon as someone implements dice rolling in a computerized version of a game, someone else is bound to question the validity of the randomness of the rolls. "
I find the fact that you didn't say anything but post a link to this to be very dismissive and kind of insulting. If you had actually read my OP, you would know that I actually have EVIDENCE to justify questioning the reliability of the randomness of the rolls. That evidence being that the probability of the rolls in question are so astronomically slim that they are likely dwarfed by the chances of a bug. Unless you are claiming to know that the chances of there being a bug in the program are somehow lower than 3 out of 100 million...but honestly I think you'd have to have expensive state of the art equipment to be that confident.
And notice that I am questioning the reliability, rather than making claims about it. I have asked the question, and not only are you not giving a real answer, you're all but flat out ignoring me. That's just plain rude.
And for the record, the fact that the dice rolls are computerized have nothing to do with this. If this happened with real life dice, I would want to find out whether the dice were weighted.
-
I should probably just stay out of this, but I feel I must point out that that is not evidence of a bug. Showing something to be unlikely to happen does not show that something messed with the results. That's like a man saying his wife cheated because it is unlikely for his kid to be albino. No evidence what so ever for the accusation. The fact that the event is possible, but unlikely to occur, does not prove anything one way or another.
I think sIKE's response was very on point. You are questioning the results of the RNG so he pointed you to where the system was explained. I personally have doubts to the accurate distribution of his model as well, but I do not have evidence to call it into question. Just a general distrust of electronic RNGs. If you want to call it into question, you should collect more data than just two rolls.
-
Zuberi, you're confusing evidence and proof, and that invalidates your entire argument. This IS HUGE evidence of a bug. You do not need proof to have good reason to investigate something, you just need strong evidence. The proof one way or the other is what you get AFTER you've actually LOOKED.
-
I should not have used the word prove as I do understand the difference. It does not point one way or another. Is that better? How exactly do these two dice rolls which you admit could be produced by the system working as intended supply evidence that there is a bug? You need more data.
-
Friday night was at my local gaming guild on campus when two of my best friends were playing. I think it was Warlord VS Warlock. My Warlock buddy has a Ring of Fire on and decides to Fire Ball the Akiro Hammer. He literally gets 8 blanks! They're both stunned beyond belief and so am I. Then next turn it happened again......there is a world of difference between IMPROBABLE and IMPOSSIBLE.
I have no idea how to get the dice on an online game to favor you. But for my personal dice I like to routine purge those who under perform, and make the others watch as an object lesson.
-
You are not paying attention to what I'm actually saying. I explicitly said that it is not impossible. What I said is that the chances of such a thing happening out of pure randomness are so astronomically small that they are LAREGLY dwarfed by the chance of a bug in the program. You cannot assume this is just coincidence without actually checking to see if it is, since it is FAR more likely that it is not coincidence.
The chances of the rolls in your real life game are 0.000000023230573, or about 2 out of 100 million. While it's possible to happen through pure chance, it's WAY more likely to be due to lazy dice rolling.
Aylin, could you please help me out here? Thanks!
-
Imaginator you're right. I didn't mean to miss the point of your post. I apologize. To me it sound like you just had some rough luck, but then I don't play online so can't honestly discuss the likeliness of a bug in the system.
-
Well, I kind of doubt that likeliness is less than 3 out of 100 million.
Thank you.
-
You are saying it is statistically unlikely as your evidence, and I am saying you don't have the data to show that. Out of the many players on octgn for the many months it's been running, we probably have several million dice rolls accumulated and you are the only one to mention this occurrence. Assuming the system is working as intended, it had to happen to someone.
You are asking sike to put in quite a bit of work without showing any reason why he should. Remember he is doing this voluntarily in his free time. There's nothing wrong with asking, but don't get upset when he gives a legitimate response.
-
Zuberi, first of all I wasn't demanding him to do anything. I even explicitly said as much in the OP. But what he did do was very dismissive. It would have been far kinder for him to give a plausible explanation like the one you just gave, or even say something like "I appreciate your concern, but I must respectfully decline from commenting now since I'm busy."
Instead he did the equivalent of telling me to "talk to the hand".
-
My friends perhaps we should table this for a bit? We shall solve no problems with this current thread and merely creature hard feelings I believe.
-
Fair enough. His response was kind of cookie cutter. I doubt he meant it to be unkind but I can understand it leaving you with a bad taste. At this point I have said far more than I intended in a thread I don't really have business in, so I shall step aside. I am sorry to hear about your misfortune, whatever the cause may have been.
-
@silverclaw Nah. Zuberi's already proved me wrong about the chances of a bug with the reasonable argument he made in his last post. I'm just upset that most of the responses I've gotten in this thread up to this point were so dismissive without really backing up what they were saying. Considering that all it took was for someone to say "there have already been millions of dice rolls" for me to realize that my concerns about the randomness of the dice are baseless, then I don't see why or how this had become a conflict.
-
I have no idea how to get the dice on an online game to favor you. But for my personal dice I like to routine purge those who under perform, and make the others watch as an object lesson.
ROFLMAO
-
@silverclaw Nah. Zuberi's already proved me wrong about the chances of a bug with the reasonable argument he made in his last post. I'm just upset that most of the responses I've gotten in this thread up to this point were so dismissive without really backing up what they were saying. Considering that all it took was for someone to say "there have already been millions of dice rolls" for me to realize that my concerns about the randomness of the dice are baseless, then I don't see why or how this had become a conflict.
@Imaginator
My response was not meant to be an insult, nor cookie cutter.
Let me take things back in time to my first games on OTCGN and before I got involved with it on the dev side. I had a series of game where I had totally awful rolls and I threw a fit at COSWORTH about the impossible nature of my rolls. Think what you experienced and cube it. His response was to implement what he wrote there. I have looked at the code since and (I took over in mid-December) what he coded looks to be properly implemented. However, since the dice roller reaches out to the Inet to generate the numbers from rnd generator there in the cloud. There is also a fallback piece in case of a time out. I wonder if that is what happened here. None of that is logged so I can't review and give you an informed feedback.
I will look into having some code added to trap this info (at least for us devs) so we can see if things are working as expected.
Hope this helps and once again the link I was pointing you to, was so you would have an understanding of how that part of the game was coded. Not excuse.
-
Aylin, could you please help me out here? Thanks!
Without a larger sample size it is impossible to say whether or not the program might not give truly random outputs. In a truly random system, previous results will not influence present or future rolls.
I think the main issue is due to the human brain's knack for finding patterns where none actually exist.
Consider say, a 5-dice roll with effect die that gives:
1x Blank, 1x 1 damage, 1x 2 damage, and 2x 1 critical damage, with any number showing on the effect die.
Probability is: 1/3 * (1/6)^4 * 1/12 = 1/46656 ~ 2.143 E -5
Later in the same game a roll comes up with:
2x Blank, 1x 2 damage, and 2x 1 critical damage, with any number showing on the effect die.
Probability is: (1/3)^2 * (1/6)^3 * 1/12 = 1/23328 ~ 4.287 E -5
But when multiplying those two together we get 1/1088391168 ~ 9.188 E -10. The probability is much lower, yet no one would find it odd that they both occurred in the same game.
I hope that makes sense; I'm trying this just as I'm going to bed.
-
I think I see what you mean. You're saying the chances of any two particular 5 dice rolls occurring in the same game are that astronomically small.
You're right. I just calculated that the chances of any particular set of results with no blanks for a 10 dice roll are about 0.000000016538172, or nearly 2 out of 100 million, regardless of whether they are the same or different. And that's ignoring the effect die.
Blanks have twice the chance of occurring per die as any other outcome, and the chances of rolling ten blanks are significantly higher than rolling one particular combination of ten specific non blanks: 0.000016935087808, or nearly 2 out of 100 thousand.
Even greater than both of these probabilities are the chances of not rolling any blanks at all, which is 0.017341529915833 or about 17 out of 1000.
I stand thoroughly corrected.
-
stop focusing so much on this particular case.
Consider this:
Roll 5 dice. Note down the result.
Roll 5 dice. Note down the result.
Post it here, and have someone good at math calculate the chance of this (these) particular outcome(s).
You will see that the chance is small no matter how close to average the rolls were.
Sure.. it wont be as small as 1/100000000000000000000000 but it will still be small enough to go "woooot what is the chance of this".
-
my worst roll in Mage Wars was when i was attacking i think by Necropian Vampiress by 7 dices and roll 7 blanks :P she suppose to kill and heal, instead she died.. end of story.
-
This is what Bayesian statistics is for, isn't it? (I am not a statistician.) We had one extremely unlikely event, but what are the odds of that single event occurring out of a population of other events? We'd need to estimate priors and such.
-
Basic internet gaming common sense:
1) No "pseudo random number generator" (PRNG) is truly random. With sufficient information about the programming of any given PRNG code, you could predict the pattern of random numbers rolled. This principle has been used repeatedly to steal money from online poker sites. (Therefore, in the modern day they take great efforts to use good PRNG algorithms and difficult to predict seeding)
2) That said, the probability of encountering a "strange" roll is extremely high. Anyone who's played enough tabletop games with real dice knows that you see really weird/frustrating rolls all the time. Randomness is random, and that means that you see "uncommon" results quite often. For example, the probability of getting one specific "strange" rolls may be 1:100 million against, but if there are a million different rolls that the human brain interprets as "strange", then you'll see a strange roll every 100 rolls.
3) Given #1 and #2, when you see a weird online dice roll it is impossible to tell whether it is a bug in the code, or simply a "strange" outcome (that would have happened with truly random dice). Most of the time it is probably #2, but this can never be 100% proven.
-
I think the probability should be closer to 1 in a milion for the two consecutive rolls. There are 5 ways of getting 4 blanks and 1 normal damage so:
p = [ (5 * (2/3)^4) * 1/6 *1/12 ]^2 = 7.4e-7
but that is if you roll two times and only two times.
If you make several rolls over the course of a game (or several games), you have to consider that any of them can be the above. The probability for two of this roll to occur at least twice over, say thirty rolls (435 ways of arranging 2 rolls over 30 rolls total) is:
435 * p^2 * (1-p)^28 = 0.0003
Still unlikely, but not astronomical. And as Aylin suggests, this is only when looking at this particular result. The probability for any 2 results (no matter what result) to be identical over the course of 30 rolls, is higher of course.
There can still be a bug in the random generator I guess.
-
I have made rolls like that using actual dice before, it does happen sometimes. Just like the in the very first game of Mage Wars I ever played I rolled 6 crit 2s on one roll.
-
In one of my very early games (third I ever played), my opponent cast a Group Heal on 4 Foxes and Redclaw. The spell rolled all blanks on Redclaw and two of the Foxes.
-
My friend rolls like a god.. with force master and dancing scimitar he rolled all crits against my priest 8!/7/6 and next turn he hurled boulder for 12 damage (10 crit) turn 5 I was at 1 life so to get your dice to work better give them to a friend to roll
-
The friend of my grand mother once rolled a 3 on a die!... ;D
-
The friend of my grand mother once rolled a 3 on a die!... ;D
LIES! With a D6 you can only roll 0s, 1s, and 2s.
Wait, does your grandmother's friend play Mage Wars? :o
I have this idea that's been put in my head that by the time I'm elderly my tastes will have changed so drastically that I won't enjoy customizable strategy games anymore, and will prefer "tamer" games, like Go, Chess, and Bridge.
I know it's ridiculous. In spite of all of the cultural and psychological evidence to the contrary, nothing that I know of can completely dislodge this fear from my mind short of real life examples of people who are 65 years old or older who play the kinds of games I like.
However, considering how young Mage Wars is, I don't think it's very likely that there will be anyone on here who's that old.
Thank you for all of your well reasoned arguments and real life examples. I was completely wrong about the abnormality of the Dice rolls I got in that one game. I feel much better about it now, thank you! :)