I demand to know the parameters of the Mersenne Twister! And I need to know how you're seeding it, too!
/conspiracy
/conspiracy
20 combat die rolls vs computer in iron man game.
- Thread starter unmerged(785632)
- Start date
-
We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.
- Status
If you actually read my post, I thought I explained it. It someone were to (and I'm sorry if this is incorrect, I don't know whether or not Paradox uses a seed#, I merely assume they do) to take a subset of the seed numbers used to create dice rolls, and there was a bias in how that subset was selected, it is possible that the dice rolls would not be representative of the entire range of seed numbers, thus making them an unrepresentative sample. This clearly doesn't apply in this case, but was more of an observation.
If you actually wanted balanced battles, you would have to get away from a 0-10 RNG perspective, you would have to use chance/probability based on the standard-deviation for it to be truly lucky to deliver a massive 10-0 blow =)
This would remove quite a bit of the randomness of the dice, and have battles mainly controlled by all the external factors =)
Just saying that with this system you will get confirmation bias, due to widely varying results =)
This would remove quite a bit of the randomness of the dice, and have battles mainly controlled by all the external factors =)
Just saying that with this system you will get confirmation bias, due to widely varying results =)
There isn't any conformation bias.
People have been playing this game for YEARS and have always had this issue.
It's clearly confirmation bias.
As I said earlier, if you record examples where you feel that the AI is rolling better than you, then BY DEFINITION, your data will show that the AI is rolling better than you.
Sorry, I guess I missed that part of your post. I don't know much about seeding and the computer science of RNGs like the Mersenne twister. I was more talking about a logical point of view; implementing a d10 is not the as the idea of a d10.
Might be more of survivor bias, since he probably didn't start out thinking "the AI rolls better than me" but rather reached that conclusion AFTER the AI did roll better than him.
If he now continues to record only examples where the AI outrolls him, then that is confirmation bias. It's only confirmation bias if you already have the hypothesis.
True, that is the more correct terminology. "Confirmation Bias" is a more commonly used phrase but you are right. Thanks for the correction.
100% of the Titanic survivors survived the sinking of the Titanic!
War Memoirs are written by people who weren't KIA.
Because people have, for years, had confirmation bias.
I've been playing SRPGs for years. Going back to 2002, every other thread I see on message boards for the game series Fire Emblem is about how the RNG is clearly broken because they AI should never hit on a 3% hit chance.
One reason for the bias is because of how the human plays differently from the AI. In a typical war I may fight dozens of battles. In the majority of those I'll have an advantage in numbers, simply because I'm better than the AI at deploying my troops. Unless the AI has a huge advantage I'll probably be able to take apart his army one bit at a time, even if his army is larger overall.
As a result human players see many, many more battles where they outnumber the enemy than vice versa. And as a result the human player will see many, many more battles where the AI wins despite the odds rather than the other way around. There simply won't be many cases where the player has to rely on luck to win.
As a result human players see many, many more battles where they outnumber the enemy than vice versa. And as a result the human player will see many, many more battles where the AI wins despite the odds rather than the other way around. There simply won't be many cases where the player has to rely on luck to win.
1) confirmation bias is as old as humanity, so I don't see how the age of the game can disprove that this can be contributed to confirmation bias.
2) EU4 is only a couple of months old, so how can people have playing this for YEARS? If you mean the whole EU franchise, then you must claim that for years a bug in the RNG of these very different games (they have been implemented in completely different engines, f.e.) has never been found?
The whole issue is clearly psychological. Beside confirmation bias, there is this tendency to "blame" the dice if things don't go the way you want, as anecdotes in the thread have refered to. There are probably other biases at work, too, such as gamblers fallacy.
You only think that because of your confirmation bias. In reality, people have been playing this game for "YEARS" and have always shown that this issue is not in fact real.
This cracked me up, awesome post!
You can infer you have a problem, but it doesn't prove there is one.
.png){kind=avatar}
RNG bias is the computer game development equivalent of 9/11 trutherism. No matter how much you explain the science, they just won't believe that an exploding jet fuel tank can melt steel.
This again. It's impossible to set metal on fire. Might aswell claim you can melt rocks.
This again. It's impossible to set metal on fire. Might aswell claim you can melt rocks.
Wrong. Iron is a flammable metal, especially in powder form. Alternatively, you can say that the enthalpy change for the oxidation of iron is negative.
EDIT: I should say that iron releases heat when heated in flames, and can burn in powder form. But trolling others is so much fun
grisamentum said:
I'll go Bayesian on you. If you create a new dice, that has never been rolled before, and your first 20 rolls all roll 1, it's quite likely (but still not not certain) you have a problem. The hypothesis that the dice is loaded would be indeed the most probable one, if you have any reasonable chance that the dice _could_ be loaded to start with.
If you have a dice that has been used for years, and generated a sequence of millions of rolls, and (all) you're shown a sub-set of the rolls that starts with 20 ones (and then say you're shown next two rolls only), it doesn't mean much. Indeed, if you didn't have such a sequence in millions of rolls, you'd better start being suspicious! Indeed, I'd expect a few sequences of (say) 1,2,3,4,5,6,6,5,4,3,2,1 as well and many other "funny" ones to be in a sequence of millions of 1D6 rolls (but I'd be never guaranteed to have one, of course).
- Status