PDA

View Full Version : There has to be some bug with the die rolls


Fahbs
12-30-2011, 02:24 PM
Not just saying this as someone bitter of defeat, but there's no way the odds can be working correctly with this.

I'm talking about rolling double "attacker down" dice when you get 2 dice for a block roll. When you choose to reroll this, it comes with the same result WAY too often for it be coincidence. It seemed like it was happening every other game, so I started keeping track of it. 3 out of 5 times so far.

That is way too high to be coincidence. Has anyone else experienced this?

Spattercat
12-30-2011, 02:34 PM
Are you sure it's not happening with a loner? Loners have a chance to fail their re- roll.

Fahbs
12-30-2011, 07:43 PM
Nope.

Another thing I've noticed is the goblin chainsaw backfire. It's supposed to be 1 in 6 but it's 42% in the five games I've been keeping track.

Spattercat
12-30-2011, 08:33 PM
I dunno, there have been extensive compilations and studies regarding the die rolls over thousands of games. The odds can get ridiculously low, due to modifiers... like 16.7%. Check out BloodBowl Manager for good stats and tables, if you haven't.

Andywilc0
01-17-2012, 11:04 AM
I think I've got to agree with Fahbs, I find it hard to believe some of the results that come back.
In the two weeks since I got this game, my apothecary has returned a better result twice. Even when I suffered a dead player, it still came back dead! That's a double six followed by a double six. The computer has never failed a regeneration roll, despite only having a 50% success rate (4+ on a six sided dice). And I've stopped counting the number of rerolled block results that still manage to come up double attacker down.
I wanted very much to like this game, but I'm struggling to, I just find it's dice results so suspicious.
I'm hoping to have better luck with multiplayer, not like the computer can play favourites eh?

VoodooMike
01-18-2012, 04:21 PM
I think I've got to agree with Fahbs, I find it hard to believe some of the results that come back.
In the two weeks since I got this game, my apothecary has returned a better result twice. Even when I suffered a dead player, it still came back dead! That's a double six followed by a double six. The computer has never failed a regeneration roll, despite only having a 50% success rate (4+ on a six sided dice). And I've stopped counting the number of rerolled block results that still manage to come up double attacker down.
I wanted very much to like this game, but I'm struggling to, I just find it's dice results so suspicious.
I'm hoping to have better luck with multiplayer, not like the computer can play favourites eh?
Reality is not a democracy. It doesn't much matter what you agree or disagree with if you have no real foundation for your belief. In this case, your "gut feeling" tells you the dice aren't rolling right. In the other corner we have:

- Hundreds of games being logged by players (not the programmers) and the dice rolls compiled to find any skew in the values.

- The RNG having been dismantled (by users) and replicated such that the rolls can be accurately predicted before the computer rolls them.

In each case the dice have been shown to work exactly as they are supposed to with a far, far better distribution than physical dice.

Olmoz
01-19-2012, 03:08 AM
There is some sort of bug with the rerolls yes. But strangely enough u can bypass it by waiting for a short while before using the reroll. Maybe 10 secs or so.
Try it, i works for me

VoodooMike
01-20-2012, 12:24 AM
There is some sort of bug with the rerolls yes. But strangely enough u can bypass it by waiting for a short while before using the reroll. Maybe 10 secs or so.
Try it, i works for me
That's total crap and always has been. That little urban legend has been around since the beta stage, and it has been disproven via exactly the same methods used to determine that the dice rolling is done correctly.

The "reroll bug" as people called it was nothing more than people not being aware of Pro and Loner skills applying to a situation - a failed attempt to use a reroll (as in, it rolls to see if it lets you use a reroll) results in the display of the same result, but no reroll is used.

FAButzke
02-07-2012, 03:14 AM
Reality is not a democracy. It doesn't much matter what you agree or disagree with if you have no real foundation for your belief. In this case, your "gut feeling" tells you the dice aren't rolling right. In the other corner we have:

- Hundreds of games being logged by players (not the programmers) and the dice rolls compiled to find any skew in the values.

- The RNG having been dismantled (by users) and replicated such that the rolls can be accurately predicted before the computer rolls them.

In each case the dice have been shown to work exactly as they are supposed to with a far, far better distribution than physical dice.

First, let me tell you that I love this game.
But, I don't f*ing care about the statistics provided by any of these "so called" RNG logs. The CPU cheats. Period.
That, or some of us have EXTREMELY bad luck.
I've seen the CPU roll SIX times in a row a 6 in order to escape unpunished. Me, on the other hand, rolled FOUR times a 1.
And that's just one of MANY examples I've seen in my games.
Is it possible to get those results normally? Of course. But something smells fishy when it happens EVERY SINGLE game.
I'll repeat: The CPU cheats (even on easy). Period.

Cutlery
02-07-2012, 08:58 AM
First, let me tell you that I love this game.
But, I don't f*ing care about the statistics provided by any of these "so called" RNG logs. The CPU cheats. Period.
That, or some of us have EXTREMELY bad luck.
I've seen the CPU roll SIX times in a row a 6 in order to escape unpunished. Me, on the other hand, rolled FOUR times a 1.
And that's just one of MANY examples I've seen in my games.
Is it possible to get those results normally? Of course. But something smells fishy when it happens EVERY SINGLE game.
I'll repeat: The CPU cheats (even on easy). Period.

You might not understand what the word "Random" means.

If you flip a coin 10 times and it comes up heads all 10 times, what are the chances it will come up heads on the 11th time?

Here's a hint. It has nothing to do what came up the previous 10.

FAButzke
02-07-2012, 11:11 AM
You might not understand what the word "Random" means.

If you flip a coin 10 times and it comes up heads all 10 times, what are the chances it will come up heads on the 11th time?

Here's a hint. It has nothing to do what came up the previous 10.

Here's another hint: You're in denial. ;)
One of us is.
Random is one thing. No-so-random is another.
I'll use the same example as you did:
When YOU flip the coin 10 times and it come up heads 10 times and you need tail. Either the coin is rigged or you must die, because something is wrong with you.
Worse still is when the owner of the coin (the CPU on this case) flip the coin and get tail 10 times in a row...

So, there are just two possibilities here (the two I've provided before):

1) The computer cheats.
2) We are jinxed beyond hope.

Cutlery
02-07-2012, 02:58 PM
Jesus dude, you're just digging deeper and looking dumber with every post. Stop, now.

Ever play Risk? You know, the tabletop game of world domination? Played entirely with dice?

You ever seen a lone defender kill 10 battalions? I have. On a tabletop. With dice. Rolled by people.

It doesn't happen often, but it happens. The hallmark of a true random number generator is that there are streaks of the same results, you don't get those when you're merely "attempting" to be random.

Basically, you don't know what you're talking about and assume that your anecdotal evidence means the game cheats when in actuality, you're just bad.

VoodooMike
02-07-2012, 11:48 PM
But, I don't f*ing care about the statistics provided by any of these "so called" RNG logs. The CPU cheats. Period.
Oh, well if you put it like that then you must be true! In addition to stats run on logs, someone (we'll call him HoodooIke to protect his identity) reverse engineered the dice routines, allowing him (and anyone else) to sit there and see what the next roll would be ahead of time, and miraculously, the computer used those rolls rather than inventing its own.

, or some of us have EXTREMELY bad luck.
That's the thing with luck, on a long enough timeline you have patches of both good and bad, and it all averages out to neither.

One of us is.
It's you.

Random is one thing. No-so-random is another.
I'll use the same example as you did:
When YOU flip the coin 10 times and it come up heads 10 times and you need tail. Either the coin is rigged or you must die, because something is wrong with you.
No, what you're looking for is an even distribution not a random distribution. The latter only approaches the former as your sample size gets large.. and 10 is not large.

It's a common misconception, really, that the two are the same. Take a piece of paper and give anyone 10 pennies, and tell them to place them, one at a time, in random spots on the page. What you'll end up with is very unlikely to be a random distribution - it'll be an even distribution, because people think random immediately results in even, and will place the majority of the pennies equally distant from each other.

So common is this misconception and the resultant expectation, that many online games actually eschew randomness in favour of forced evenness, such as MMOs that guarantee you won't miss three times in a row... That just makes people even less likely to properly understand random.

FAButzke
02-08-2012, 02:00 AM
Jesus dude, you're just digging deeper and looking dumber with every post. Stop, now.

Oh... We are in the "personal attack phase" now. Ok.


Ever play Risk? You know, the tabletop game of world domination? Played entirely with dice?

You ever seen a lone defender kill 10 battalions? I have. On a tabletop. With dice. Rolled by people.

Your example is invalid. I could forget about it if it would happen once in a lifetime in Bloodbowl but, as I have said and you ignored it: It happens in every single game. This is not random, it's crap, you know it.


It doesn't happen often, but it happens. The hallmark of a true random number generator is that there are streaks of the same results, you don't get those when you're merely "attempting" to be random.

Do you know Occams Razor? You know, that concept where the simple explanation is usually the right one?
So... What do you think it is most probable?
That the dice here is actually random or that the CPU cheats in order to increase the challenge for the player because otherwise it wouldn't be a match for him.


Basically, you don't know what you're talking about and assume that your anecdotal evidence means the game cheats when in actuality, you're just bad.

Oh... Here I was thinking that personal attacks where out of the way by now. Silly me...
I'm not bad and I've been playing games since 1988, so please, stop assuming you know me.
I have no trouble beating the CPU. That was not the point. The point is that it cheats and it can lead to frustration.
Perhaps it doesn't cheat all the time. Perhaps it only cheats when he is losing badly. Or perhaps it only cheats in certain situations like when he is surrounded and need to get out of a box. I highly doubt it that the "RNG logs" shows this. My guess is that they just sum all the time all the results (1-6) actually get rolled and do not take into account the situation at all. Am I right? It I am, than the "RNG log" is crap, just like I said. Because when the computer cheats, it would go unnoticed.
Do you get it now?


Oh, well if you put it like that then you must be true! In addition to stats run on logs, someone (we'll call him HoodooIke to protect his identity) reverse engineered the dice routines, allowing him (and anyone else) to sit there and see what the next roll would be ahead of time, and miraculously, the computer used those rolls rather than inventing its own.

See above.


That's the thing with luck, on a long enough timeline you have patches of both good and bad, and it all averages out to neither.

One or two times I would get it, but when it happens every time: No.
Besides, it's not just me.


It's you.


When you say it like that, then it must be true! Ha! =D


No, what you're looking for is an even distribution not a random distribution. The latter only approaches the former as your sample size gets large.. and 10 is not large.

It's a common misconception, really, that the two are the same. Take a piece of paper and give anyone 10 pennies, and tell them to place them, one at a time, in random spots on the page. What you'll end up with is very unlikely to be a random distribution - it'll be an even distribution, because people think random immediately results in even, and will place the majority of the pennies equally distant from each other.

So common is this misconception and the resultant expectation, that many online games actually eschew randomness in favour of forced evenness, such as MMOs that guarantee you won't miss three times in a row... That just makes people even less likely to properly understand random.

I know what you're talking about man, I really do. But the problem with examples like that is that they really don't translate well to what I'm talking about.
Do you know a game called Baldur's Gate?
One of my favorite games ever. Do you know that the dice used on that game is a 20 sided dice? And the odds of getting a 1 is even lower than on a 6 sided dice? Well... I managed to roll 1 four times in a row on a dire situation. Tell me the odds of that and tell me that it was just bad luck. I even had a screenshot to prove it.

So you see? I'm not telling that it CAN'T happen I just saying that if a improbable situation happens way too often... then it's not that improbable.

Let's just drop it because this will lead nowhere. I'll continue to assume that the programers made the CPU harder by cheating and you continue to assume that it doesn't cheat based on the logs and the program to predict CPU moves. Everyone will be happy. Trust me.

VoodooMike
02-08-2012, 03:01 AM
Let's just drop it because this will lead nowhere. I'll continue to assume that the programers made the CPU harder by cheating and you continue to assume that it doesn't cheat based on the logs and the program to predict CPU moves. Everyone will be happy. Trust me.
True, there is little point in discussing the matter with people like you (http://www.urbandictionary.com/define.php?term=pigeon%20chess) - you imagine you're rational, and that you use logic, except that you plug your ears when it doesn't support what you decided, intuitively, to be true.

The program predicted dice rolls, not moves. If the computer were altering the dice rolls or discarding rolls that did not suit it, that would have been instantly obvious since the CPU's rolls would not match what the program predicted they would be, but it did.

Since you feel there's no point in the topic, and aren't satisfied by the use of logic or evidence outside your own paltry personal experience, then I'm sure we won't hear from you again. Knowing your kind, however, I won't get my hopes up.

Matheau
02-08-2012, 03:47 AM
Do you know Occams Razor? You know, that concept where the simple explanation is usually the right one?
So... What do you think it is most probable?
That the dice here is actually random or that the CPU cheats in order to increase the challenge for the player because otherwise it wouldn't be a match for him.

Occam's Razor would say the most likely solution is the dice are random. In order for the CPU to cheat at Blood Bowl, the developers would have to create a very complicated and elaborate AI that would involve manipulating the random number algorithm before it generates numbers while knowing exactly how many times and what actions the player will take before they take them.

Someone made a program that correctly predicted every random number Blood Bowl generated in advance, so it is known for a fact the AI never alters a dice roll.

Keep in mind that the difficulty of creating a "cheating" AI varies substantially between different games. If reaction time or accuracy are the sources of difficulty, it is very easy to make the AI cheat by giving it reflexes and abilities beyond humans. In a game like Blood Bowl, it is pretty much impossible due to the heavy use of random numbers and how the random number algorithm works. The only way the AI can effectively "cheat" would require psychic powers. It can, however, very quickly, and effectively predict chances of success very quickly (yes, a computer can do basic multiplication and division). So, against incompetent players, it seems like a genius. Against competent players that also have a good idea of the odds for or against each action, the AI is a complete push over because anyone that thinks the entire game is totally based on random numbers doesn't know what they are doing. Yes, random numbers play a major role and can win or lose games, but games are mostly won through player selection, skill choices, and positioning.

At best, the AI could look ahead and see if their action is going to succeed or fail. However, to invoke Occam's Razor here. The simpler solution is luck rather than the malevolent developers programing the AI with an elaborate system that also fails even at the simplest of tasks on occasion to try to "trick" the players into thinking it isn't secretly cheating.

It must be especially devious because a couple games ago, a Wood Elf with dodge carrying the ball ran through one square next to my rookie Ogre (by the way there was no reason for him to do this because it was impossible for them to score before the end of the game) with no skills, attribute boosts, or even SPP. The Wood Elf failed both his rolls, the Wood Elf failed the armor roll, the Wood Elf was then injured, he dropped the ball, the ball scattered towards my Ogre, and the Ogre (AGI 2 and in a tackle zone) caught the ball without any use of a reroll. On my next turn, the Ogre made his bonehead check with no reroll, dodged out of a tackle zone (still AGI 2) with no reroll, and made two GFI checks with no reroll. If the AI is actually cheating, it is very devious to do something that seemingly incompetent and unlikely.

By the way, that sequence of events would occur 0.0000009653% of the time or 1 time in every 10360 attempts for that exact sequence of events.

Also, it is very odd that the two complaints about the AI are "OMG! THE AI CHEATS" and "The AI is complete garbage that is completely incompetent." If the AI actually did cheat, all the players should notice it, but there is a clear dividing line between the perceptions of experienced players, especially ones that regularly play against other experienced players, and brand new players that probably haven't even read the rules.

FAButzke
02-08-2012, 04:22 AM
True, there is little point in discussing the matter with people like you (http://www.urbandictionary.com/define.php?term=pigeon%20chess) - you imagine you're rational, and that you use logic, except that you plug your ears when it doesn't support what you decided, intuitively, to be true.

The program predicted dice rolls, not moves. If the computer were altering the dice rolls or discarding rolls that did not suit it, that would have been instantly obvious since the CPU's rolls would not match what the program predicted they would be, but it did.

Since you feel there's no point in the topic, and aren't satisfied by the use of logic or evidence outside your own paltry personal experience, then I'm sure we won't hear from you again. Knowing your kind, however, I won't get my hopes up.

Oh no! You've hurt my feelings. =(
For the sake of argument, I'll explain to you (http://en.wikipedia.org/wiki/Attention_deficit_hyperactivity_disorder) why this discussion is pointless.

I've stated that the computer cheats. That OR I have bad luck (back there in the first post, go check it, you might have missed. Very common with people with your condition)
Did you see what I did there? There was two options right off the start:
OR the computer cheats (And I'm right) OR I (and others) have bad luck (And I'm wrong).
Have you never felt frustrated by something to the point of blaming other things besides fate itself?
This discussion could've ended right there. You could just have said: "I have no problems with the rolls. I've tested them. Must be something on your end. The computer behave fairly for me."
But, as you took another path, so did I. You've tried to prove that the computer didn't cheat so I did the opposite. Why? Because I wanted to hear a better explanation to why the rolls behave like that against me instead of: "It's random, man. I can prove it. You just suck!"

Now, as you have said it, knowing your kind, you'll just twist everything I just said, ignore the rest and come up with another way of attacking me or trying to prove your point.

Let's just leave it at that: The computer does not cheat and we are in denial, losers and have the luck to prove it.

Oh wait, I just proved you right! =(

FAButzke
02-08-2012, 04:36 AM
Occam's Razor would say the most likely solution is the dice are random. In order for the CPU to cheat at Blood Bowl, the developers would have to create a very complicated and elaborate AI that would involve manipulating the random number algorithm before it generates numbers while knowing exactly how many times and what actions the player will take before they take them.

Someone made a program that correctly predicted every random number Blood Bowl generated in advance, so it is known for a fact the AI never alters a dice roll.

Keep in mind that the difficulty of creating a "cheating" AI varies substantially between different games. If reaction time or accuracy are the sources of difficulty, it is very easy to make the AI cheat by giving it reflexes and abilities beyond humans. In a game like Blood Bowl, it is pretty much impossible due to the heavy use of random numbers and how the random number algorithm works. The only way the AI can effectively "cheat" would require psychic powers. It can, however, very quickly, and effectively predict chances of success very quickly (yes, a computer can do basic multiplication and division). So, against incompetent players, it seems like a genius. Against competent players that also have a good idea of the odds for or against each action, the AI is a complete push over because anyone that thinks the entire game is totally based on random numbers doesn't know what they are doing. Yes, random numbers play a major role and can win or lose games, but games are mostly won through player selection, skill choices, and positioning.

At best, the AI could look ahead and see if their action is going to succeed or fail. However, to invoke Occam's Razor here. The simpler solution is luck rather than the malevolent developers programing the AI with an elaborate system that also fails even at the simplest of tasks on occasion to try to "trick" the players into thinking it isn't secretly cheating.

It must be especially devious because a couple games ago, a Wood Elf with dodge carrying the ball ran through one square next to my rookie Ogre (by the way there was no reason for him to do this because it was impossible for them to score before the end of the game) with no skills, attribute boosts, or even SPP. The Wood Elf failed both his rolls, the Wood Elf failed the armor roll, the Wood Elf was then injured, he dropped the ball, the ball scattered towards my Ogre, and the Ogre (AGI 2 and in a tackle zone) caught the ball without any use of a reroll. On my next turn, the Ogre made his bonehead check with no reroll, dodged out of a tackle zone (still AGI 2) with no reroll, and made two GFI checks with no reroll. If the AI is actually cheating, it is very devious to do something that seemingly incompetent and unlikely.

By the way, that sequence of events would occur 0.0000009653% of the time or 1 time in every 10360 attempts for that exact sequence of events.

Also, it is very odd that the two complaints about the AI are "OMG! THE AI CHEATS" and "The AI is complete garbage that is completely incompetent." If the AI actually did cheat, all the players should notice it, but there is a clear dividing line between the perceptions of experienced players, especially ones that regularly play against other experienced players, and brand new players that probably haven't even read the rules.

Now that was some explanation. Thank you.
Just one consideration. There's a simple method to make a "cheating AI" here.
Just roll the dice. If the number is less than the desired one, just give the CPU the number desired to pass the roll. No complicated system needed. I'm not saying that's how it works, because clearly it was checked by our friend back there. All I'm saying is that it's not complicated to make the CPU cheat.

Also, as I've stated before, perhaps the computer doesn't cheat all the time. I've seen it fail (horribly) too. Its just it succeeds more often than fails (and with a 6 most of the time).

And I confess, I'm not an experienced BB player. I'm not that new to the game either. But, I've read the rules, I've played many games and I understand all the concepts. The problem (for me) lies in the luck factor. When you see a 6, you know that's an instant lucky success. When you see six 6 in row (in every single game)... Well... It's frustrating.

Cutlery
02-08-2012, 08:55 AM
When you see six 6 in row (in every single game)... Well... It's frustrating.

You don't see that. You just think you do and ignore all the times when you don't. It's called human perception, and humans are notoriously awful at it. I would assume you're a human, so you fall into that category.

It's why things like reverse engineering the program and looking at massive logs are far more important than your anecdotal evidence. We tried to tell you that, but you got all salty about it.

FAButzke
02-08-2012, 10:43 AM
You don't see that. You just think you do and ignore all the times when you don't. It's called human perception, and humans are notoriously awful at it. I would assume you're a human, so you fall into that category.

It's why things like reverse engineering the program and looking at massive logs are far more important than your anecdotal evidence. We tried to tell you that, but you got all salty about it.

First, I got "all salty" because you were arrogant and directly insulted me. With that out of the way:

Now you are generalizing. I am human, but not all humans are the same. Right? So, you're either:

1) Calling me a liar.
2) Calling me an idiot.

For you, I'm either seeing one thing and reporting another or thinking I'm seeing one thing when actually I'm not. Based on your post, I'll assume I'm number 2.

Let me tell you again. I don't care about what happens with a gamer in the other side of the world playing against the CPU. Because that's not my reality. What I care about is what happens when I'm that gamer and I'm the one analysing MY logs and I do see a lot of 6s in the CPU side and a lot of 1s in my side.

What are going to tell me now? That I'm seeing 6s but they aren't 6s?

I'll be clear once more: For me (in MY BB games) the CPU cheats, ok? Because that's what I see in my logs. And no amount of discussion will change that. Why? Well, because you can't change what I see every time I play (and yes, I check the logs every time). Perhaps I'm jinxed, but for me it's a simple CPU cheat. It's easier to explain my unluckiness that way.

You're welcome to call me a stubborn idiot and a loser now. I won't even respond.

Before I go, I wish to thank you (and the others) all the same. Seriously. No hard feelings on my part.

Genosidoh
02-08-2012, 03:05 PM
Let me tell you again. I don't care about what happens with a gamer in the other side of the world playing against the CPU. Because that's not my reality. What I care about is what happens when I'm that gamer and I'm the one analysing MY logs and I do see a lot of 6s in the CPU side and a lot of 1s in my side.

I'll be clear once more: For me (in MY BB games) the CPU cheats, ok? Because that's what I see in my logs. And no amount of discussion will change that. Why? Well, because you can't change what I see every time I play (and yes, I check the logs every time). Perhaps I'm jinxed, but for me it's a simple CPU cheat. It's easier to explain my unluckiness that way.


Wow, I'm going to have to give Cyanide more credit! Who knew they made each individuals copy of Blood Bowl, specifically for that individual. You must be incredibly lucky to have the only copy of BB that contains a cheating AI!:rolleyes:

Obviously this is infinitely more likely than being unlucky or unfamiliar with the rules.

Cutlery
02-09-2012, 10:40 AM
Obviously this is infinitely more likely than being unlucky or unfamiliar with the rules.

Also hilarious because it was brought up by the guy who also brought up Occam's Razor.

aepurniet
02-10-2012, 08:52 AM
for people who think the AI cheats.

http://en.wikipedia.org/wiki/Selective_perception
http://en.wikipedia.org/wiki/Confirmation_bias
http://en.wikipedia.org/wiki/Belief_bias

i know it *seems* like it does, but it doesnt. i play a fair share of asoBrain eXplorers. and people talk about this all the time there too, its easier to blame the dice (or some unknown AI) than accept credit for poor risk management.

Snorkel
02-10-2012, 02:13 PM
It's funny how there are threads like this in the forums for every single video game conversion of a board game. People just forget that they get the crazy dice rolls in real life too.

I liked this guy's solution: http://gamesbyemail.com/News/DiceOMatic

Hint4U
02-19-2012, 04:28 AM
I dunno, there have been extensive compilations and studies regarding the die rolls over thousands of games. The odds can get ridiculously low, due to modifiers... like 16.7%. Check out BloodBowl Manager for good stats and tables, if you haven't.

this is cited for many RNGs, but the problem is that most RNGs tend to handle numbers in blocks.

NWN is notorious for this, and having the rolls visible allows you to actually see blocks of 20 rolls being +/- 3 almost 100% of the time.

On a large enough sample, you get a statistically expected result, but it completely ignores these persisting patterns.

VoodooMike
02-19-2012, 05:33 AM
this is cited for many RNGs, but the problem is that most RNGs tend to handle numbers in blocks.
What isn't cited, however, is a legitimate source that supports what you're saying. The RNG in this game is MT19937.. if you'd like to find something that says it is insufficiently random for the purposes of simulating dice rolls, I'll be all eyes.

NWN is notorious for this, and having the rolls visible allows you to actually see blocks of 20 rolls being +/- 3 almost 100% of the time.
Notorious for people feeling it, you mean... or has it be proven beyond people's gut feelings on the matter? Streaks are an inherent part of random distributions... because they're random. People keep looking for evenness in random distributions, and it only happens when you have a large number of iterations.

On a large enough sample, you get a statistically expected result, but it completely ignores these persisting patterns.
In order to display the expected even distribution across a large number of iterations, while still having non-random streaks, it would need to somehow compensate for those streaks by filling in the distribution gaps. Parsimony suggests you're just seeing randomness in action - that streaks happen in randomness (which they do) and that you're reading a mathematical conspiracy into it when none exists.

iviv
03-29-2012, 04:05 AM
http://i.imgur.com/5AfIS.jpg

I'll just leave this here. In short, someone in this thread doesn't understand what random is.

Don_
03-29-2012, 09:01 AM
For the guy talking about waiting a period of time before re rolling that is a plausible approach to the method. I'm not claiming to be an expert to how random number generating works these days but at a basic programming stand point RNG used your system's clock as a way to accomplish this.

VoodooMike
03-29-2012, 10:51 AM
For the guy talking about waiting a period of time before re rolling that is a plausible approach to the method. I'm not claiming to be an expert to how random number generating works these days but at a basic programming stand point RNG used your system's clock as a way to accomplish this.
Yes, you're really not an expert on the subject. When you use the clock in association with an RNG it is to give it an initial seed value. That means you often use the system's seconds from wheneverthehell (last boot, 1980, whatever) when you initialize the RNG, but not afterward... re-seeding it for every roll would be moronic on an epic scale.

mrnoo
04-28-2012, 08:58 PM
Basically, you don't know what you're talking about and assume that your anecdotal evidence means the game cheats when in actuality, you're just bad.

Err if its completely random as you say, how can he be bad?

Nizar
04-29-2012, 12:47 PM
If a program can successfully predict all the rolls doesn't that suggest that it is scripted? So neither random, nor changed to suit the cpu on the moment.

VoodooMike
04-30-2012, 04:36 AM
If a program can successfully predict all the rolls doesn't that suggest that it is scripted? So neither random, nor changed to suit the cpu on the moment.
Only if you also believe that every card game is "scripted" since the order in which the cards will be drawn are technically set, but absolutely nobody knows what that order will be.... and if you believe that, you've probably got an IQ under 100 - half the population has to, afterall.

Nizar
04-30-2012, 09:20 AM
Only if you also believe that every card game is "scripted" since the order in which the cards will be drawn are technically set, but absolutely nobody knows what that order will be.... and if you believe that, you've probably got an IQ under 100 - half the population has to, afterall.

The program predicted the order, if a program correctly predicted every card turned in an online card game I would be suspicious of the game. I am just interested in how a program correctly predicting dice rolls proves the rolls are random, at face value that sounds the opposite to me.

Do you have a link for the program/experiment?

I haven't played blood bowl in years but am thinking about getting it when the Chaos edition is out. Personally I hope you are right as I would play a lot of single player if I bought it, I have read alot of complaints about AI bias which is why I was drawn to this thread.

VoodooMike
05-08-2012, 10:24 PM
The program predicted the order, if a program correctly predicted every card turned in an online card game I would be suspicious of the game. I am just interested in how a program correctly predicting dice rolls proves the rolls are random, at face value that sounds the opposite to me.
Right, and again, it's because you aren't fully wrapping your head around randomness as a practical concept, only a theoretical one. Theoretical randomness is a result that cannot be predicted because it is inherently unpredictable. Such a beast simply does not exist in a deterministic universe. What exists are expressions of practical randomness in which results cannot be predicted by you though they can, in theory, be predicted if one knew all the variables that went into the final outcome.

Consider the rolling of a physical dice. You're throwing a cube onto a surface and waiting for a result. Is the result genuinely random? No, it's the result of a myriad of variables that the player can't keep track of, such as the composition, weight, and skew of the dice, the friction of the table, the angle and force of the throw, and so on. It's not a genuinely random process, but it is a sufficient engine for randomness because you have no hope in hell of keeping track of all the variables. So long as, across a large number of throws, each of the possible values comes up an even number of times, you can say that the system is an effective system for generating random results.

Random number generators in computers are similar - they are calculation engines which, when fed a number, give a result that, unless you know what went in, and what the exact formula is, you're unlikely to ever guess the NEXT number that will pop out when the engine is called again. This doesn't mean it isn't possible if you DO know that information, just that, like the dice, it is typically outside the realm of possibility for players.

The issue with Blood Bowl that allows the dice prediction to occur is that everything for matches is held on the client side, even when playing an online match... so in memory you can get the needed information from the RNG necessary to clone its state... once you do that you effectively have all the variables, and then you can ask your separate program to tell you what subsequent results will be. In most (read: intelligent) games that have an online aspect, all of this information is held on the server side machines so you only ever have access to the output, making it near impossible to duplicate the RNG's state.

Do you have a link for the program/experiment?
The only public version of the program was for the original Legendary Edition version of the game, so unless you want to re-install the first version in order to run the tests, you're out of luck on that.

I haven't played blood bowl in years but am thinking about getting it when the Chaos edition is out. Personally I hope you are right as I would play a lot of single player if I bought it, I have read alot of complaints about AI bias which is why I was drawn to this thread.
There is no AI bias... the AI is wretchedly bad and you will destroy AI controlled teams in your sleep if you know how to play Blood Bowl... the problem is that most new players don't understand the game and think they'll just figure it out while they play. Those people get creamed. Additionally, the AI takes ridiculous risks which, given the design of the game, sometimes work out for them and when they do, people scream foul.

Genosidoh
05-25-2012, 02:09 AM
The ridiculous risks the AI takes are the best part of the PC game IMO.

Cutlery
06-06-2012, 08:16 AM
Err if its completely random as you say, how can he be bad?

People whine about games when they lose. It's safe to say he feels he's losing because the AI cheats and rigs rolls. That is not the case, the dice rolls are completely random. It's impossible to completely shield yourself from the effects of "bad luck" in this game, but it's not hard to mitigate your risk. Quite simply, if you're losing as a result of bad rolls, you are just making bad moves. You should be playing to minimize your exposure to the dice, and therefore "bad luck."

That's kind of the whole strategy behind the game...make as few dice rolls as possible while forcing your opponent to make as many as possible. That's pretty much the best and most consistent way to win. Once you figure that out, the single player is absolutely no challenge at all. I have a Chaos team in SP campaign who is 162-0-1, and the tie was the first game I played when I had all unskilled players and picking up the ball was a challenge.

CaptainSkeptic
06-12-2012, 02:35 PM
Just a note here: Occam's Razor does NOT mean "The simplest idea is usually right". This is a common misunderstanding. Occam's razor is the principle that we should never unnecessarily multiply entities for explanation. If you get a flat tire while driving down the road, "A genie cursed me" may well be the simplest explanation but it is not the correct one.

VoodooMike is correct on the points he makes. I think most of you are not familiar with behavioralism and psychology and so do not know the sorts of errors we humans are prone to. Confirmation bias is what keeps casinos in business. People tend to remember the 'hits' and dismiss or downplay the 'misses'. What this means for Blood Bowl is that if you are convinced that the AI is cheating then you will notice every bad result you get(because it confirms what you already believe to be true) but you won't recall all the times when you got away with passes into coverage, one dice blocks to permanently injure the opposing minotaur etc.

The problem with Blood Bowl is not that the AI cheats. it is that the game itself is (IMO) poorly designed. Nothing about this game is intuitive or plays out the way one would expect and extremely bad and good results are almost as common as results that are middle of the road/average. There is not a wide range of numerical possibilities with the block dice. They are in essence 3 and 1/2 sided dice so double skulls and such come up way more often than say snake eyes(when rolling two regular six sided dice). People who expect the game to play out anything at all like American football are in for a shock because BB is about as close to American football as MMA fighting is is.

Cutlery
06-14-2012, 07:33 AM
There is not a wide range of numerical possibilities with the block dice. They are in essence 3 and 1/2 sided dice so double skulls and such come up way more often than say snake eyes(when rolling two regular six sided dice).

Wrong.

1 - Skull
2 - Both Down
3 - Push
4 - Push
5 - Defender Stumbles
6 - Defender Down.

http://www.blood-bowl-miniatures.de/bb/bb_dices.html

The game is designed fine. You are just apparently not aware of all the rules.

Ixnatifual
06-25-2012, 09:52 PM
The best part of dice-based computer games is searching their respective forums for RNG conspiracy theorists. I simply find it hilarious reading. I can't stop loving it! :D

paralipsis
07-03-2012, 04:32 AM
VoodooMike, if it were possible for me to add rep to every single one of your posts in this thread, then I would do so.

Your solid reasoning, clear explanations, and sound analogies are all to be admired.

Ribo
07-03-2012, 07:19 AM
VoodooMike, if it were possible for me to add rep to every single one of your posts in this thread, then I would do so.

Your solid reasoning, clear explanations, and sound analogies are all to be admired.

Hah, I agree. I think its funny the way he sugar coats nothing and tells you exactly where you are wrong.


As for dice it is easy to visualize as numbers on a string. When the game starts the string is randomly populated. The first thing you do requiring a die roll uses the 1st number on the string. The 2nd thing requiring a die roll uses the 2nd number on the string and so forth.

VoodooMike
07-03-2012, 08:06 AM
As for dice it is easy to visualize as numbers on a string. When the game starts the string is randomly populated. The first thing you do requiring a die roll uses the 1st number on the string. The 2nd thing requiring a die roll uses the 2nd number on the string and so forth.
Saying that the string is populated at the start of the game is a bit misleading, and tends to make people think that the rolls are in some way pregenerated - they're not. You could say that they are deterministic in so much as the next value can only be what it's going to be based on the value before it.. but they are no more or less deterministic than physical dice rolls in a deterministic universe.

If we take the formula X = X+1 and start with 1, we get a sequence that looks like {1,2,3,4,5....} to infinity... we don't actually have the numbers written out somewhere to pull from, we just have a reliable way of generating the next number because, well, we know the process. In that way the numbers past 5 are ungenerated but deterministic, much as the dice rolls are in our game.

It is interesting to note, however, that because the numeric D6 rolls and the block dice rolls are not created using the same secondary algorithm (that is, the one that turns the output of the mersenne twister into the resulting dice value) the actual dice rolls themselves are dependent on your decisions, and so you (and your opponent) become direct sources of entropy for the aggregate RNG. What might have been a roll of 1 on your Go For It roll instead becomes a Defender Down result, using the same RNG output, because you chose to blitz instead of run.

Or hey, if you need the less math and science version for, say, a republican... just say that because Jesus knows everything, he knows what your next dice roll will be, and because your next dice roll is known by ANYONE, it is predetermined! Praise the lord!

Kjelstad
07-31-2012, 07:04 PM
I dunno, there have been extensive compilations and studies regarding the die rolls over thousands of games. The odds can get ridiculously low, due to modifiers... like 16.7%. Check out BloodBowl Manager for good stats and tables, if you haven't.

That^^

BBman shows when you have bad dice and when you have good dice in bad spots.

wyrdfate
08-14-2012, 09:19 AM
it seam to me that the better i do, the worst my dice roles are.

the better the CPU is fooign the more balance my game is.

i've learned that scoting point serve me nothing till after the second half and then i had better not gove the CPU a chance ot have more thne 3 turns.

the game cheats to make it more difficalt.

it's an old soft wear trick.

the programers make the cpu cheat to make the game more chalanging.

also ther is not usch thign as rendom in programing.

normally programers use the CPU clock and or other thing like a CPU instruction number, in stend of a random number genreator, to make roles apper random.

they are nto actlay random normaly.

this game amungst other are less blanced them most i play due ot the nature and dependancey of dice roles needed.

also due to the limided number of verable.

as such prdictable number turnign up often is common.

my pc may be oen of the few that make me unlucky but that just life.

simple fact.

i've had to stop playing the game.

i'm sick of alwsy losing due to the fact i can makes toughdown in the first half.

typcaly it workd this way.

fair random dice till i scora point.

then i get lower and lower results the more poitns i get.

whiel the more points i get the better roles the cpu team gets.

to the point that soem teams seam to get massive disavantages.

also the higher my rating is the worst my roles start off as.

soem game i jsut give up half way though due to the steam rolelri'm getting due ot the bad dice.

it's so much so that i'm just drained a relived just ot win.
relived i never lsot to many of my team and forget abotu getting a tonimant win the better i do the hard it gets.

i'm thinkign of jsut delbratly lossinga few game ot get the player better, skilled so i cna havea slight advantaged. but thne to me what thep oint of that?

so as i said i've feedup playing.

Cutlery
08-15-2012, 09:37 PM
I feel dumber having read that.

Matheau
08-16-2012, 07:06 AM
The AI isn't cheating. Blood Bowl is a game where the difference between an incompetent player, decent player, and good player are gigantic. The AI is basically a decent player, but one that makes some really stupid errors every now and then. It is more than good enough to destroy an incompetent player. A decent player might struggle a little bit on occasion, but usually will win because of the errors the AI makes. A good player can wreck the AI and beat it 4-0 using Khemri with brand new players or outbash a developed Chaos team using Halflings.

If you are losing because of bad rolls, you aren't good at Blood Bowl. That's just the simple fact of the matter. Unless you are extremely out classed by your opponent, a bad roll will virtually never turn the tide of a game. A good player makes every single move assuming the absolute worst thing possible will happen and plans accordingly. If you aren't making it as difficult as possible for your opponent to capitalize on a bad roll, you will lose very consistently at Blood Bowl against the AI because it is just competent enough to score points due to that.

Also, if you think trying to score as many touchdowns as possible is a good strategy outside a few specific teams, you really don't understand basic strategy. Most teams need to win by control, not putting points on the scoreboard. Trying to score touchdowns is generally more risky than just trying to control the ball, so if you are putting the ball unnecessarily at risk for unnecessary points, you are just increasing the odds of something bad happening.

the game cheats to make it more difficalt.

it's an old soft wear trick.

the programers make the cpu cheat to make the game more chalanging.

It's an old trick, yes, but that's because in most games, the AI is incapable of analyzing certain situations. If the computer didn't cheat, it would be easily beaten just by doing things it doesn't know how to respond to.

Blood Bowl is one of the few games the computer can't really cheat on. Technically starting with an established team and having simulated matches constitutes as "cheating," but it is such a minor advantage that it is pretty negligible. It doesn't even work in their favor all of the time since it forces their team values higher than normal.

also ther is not usch thign as rendom in programing.

normally programers use the CPU clock and or other thing like a CPU instruction number, in stend of a random number genreator, to make roles apper random.

they are nto actlay random normaly.

Yes, there is technically no such thing as "random" number generation in programming. You know what else is technically not random? Shuffling cards and rolling dice. If you know all the variables, you can correctly predict the order cards get shuffled in and what side of a dice will come up. Just because something isn't technically "random" doesn't mean it doesn't work well enough to be used to generate random numbers.

By the way, computer generated random numbers gravitate more towards the actual probabilities than shuffling cards or rolling dice. There has also been testing done specifically on Blood Bowl rolls and they do go to the actual probabilities they should better than actual dice rolls.

Also, I'm not sure where you heard that about how programmers generate random numbers, but either you misunderstood something or you were given bad information. Anyone generating a random number on a computer uses a random number generator. It is literally impossible not to do so because anything on a computer that generates a random number is a random number generator.

Also, no competent program would use the CPU clock in place of a random number generator. I think what you misunderstood is that programmers regularly use the CPU clock to seed a random number generator. It isn't the only way to seed it, but it is an easy way to seed it without using something predictable.