Siege Rolls; Confirmation Bias my ARSE!!!

[tbhayward]

Respectfully, I believe it is you who are providing no counter argument. Two separate people have now produced high quality data clearly showing a lack of bias in rolls. You've merely mentioned a couple of sieges that you've experienced without allowing for all of the times lady luck was in your favor.

Edit: I took the liberty of combining my and Lilje's results together. The error bars I show are 1 standard deviation. If the data is behaving as expected, then approximately 2/3 of the error bars should intersect the red line and 1/3 should not. The red line is the expected results of 446/14=31.9. This is precisely what is happening.

 

[Republic of Mercury]

Oh when you play as a nation that has fort defense stacked and pushed your enemies WE through the roof and yet they get from -56% to 7% in a instant such with no cannons no siege general such that you have to lift siege (keep 1k there to keep the progress) that you had full cannon bonus more than three times in a row to kick them out and have to do such a thing for multiple forts in the same war, you no longer see it as confirmation bias. As Ethiopia when fighting Ottomans, it got to the point of me having to get call for peace as soon as I got more than threshold. Why had I not had 100%, well despite not having lost a single battle and in fact have wiped their army the forts just didn't fall so I started to get WE because of that.
Another case of AI turbo siegeing was when I trolled AI France ally by dowing England for Gascogne and calling them in, the siege of Labourd (which I was the leader) took much more time then the 2 sieges that France did and led. Those sieges were done by the same army in sequence, with no siege general or cannons. All sieges were unblockaded ports, I had cannons in my and started it before they started theirs. Look at how outrageous fast they were siegeing Normandy, 2 months of war and they had walls breached and it did not stop there but I forgot to screenshot those. And it was not one siege but 2 in sequence, if there were more forts they might have taken a third one with that one army before Labourd fell.

It happened so many times of the AI turbo siegeing that remembering when it does not happen is much easier.

I don't understand what your screenshot is supposed to illustrate. The AI got a 14 on their first roll. I got some wall-breaches in the first few months of a siege multiple times when I was testing breach mechanics earlier. If you want the AI to be unable to roll a 14 in the first few months of a siege, that would be a biased dice.

 

[highsis]

I do think it is a confirmation bias but there's been a precedence of the countercase; EU3's merchant success chance calculation was wrong and I'm fairly certain of that.

I've seen 99% chance merchant placement failing 5 times in a row not rarely but quite often, like countless times as I played the game. It wasn't rigged towards certain outcome per se(the result could be skewed either way) but there were a certain percentage or situations that would trigger this weird effect; you could load the save game over and over and 99% placement would always fail.

I only believe it's confirmation bias in EU4 because Paradox must have learned from their mistake before and I don't feel it's rigged from my experience. On the other hand, EU3's merchant placement chance was very, very odd to say at least.

 

[pikaemperor]

I don't understand what your screenshot is supposed to illustrate. The AI got a 14 on their first roll. I got some wall-breaches in the first few months of a siege multiple times when I was testing breach mechanics earlier. If you want the AI to be unable to roll a 14 in the first few months of a siege, that would be a biased dice.

They keep getting it again and again on almost every siege. The fact they fully sieged 2 forts with the same army while I was still siegeing the first fort, and I was the one that had cannons on the siege.

 

[Aries666]

Looks like someone skipped statistics class

 

[Republic of Mercury]

They keep getting it again and again on almost every siege. The fact they fully sieged 2 forts with the same army while I was still siegeing the first fort, and I was the one that had cannons on the siege.

...So you want the dice rolls to be biased in your favour so that that's no longer possible?

 

[IndraSunrise]

I've seen 99% chance merchant placement failing 5 times in a row not rarely but quite often, like countless times as I played the game. It wasn't rigged per se but there were a certain percentage or citations that would trigger this weird effect; you could load the save game over and over and 99% placement would always fail.

If the seed is the same, 100% of the time the saved game should be the same results.

 

[Aries666]

Oh when you play as a nation that has fort defense stacked and pushed your enemies WE through the roof and yet they get from -56% to 7% in a instant such with no cannons no siege general such that you have to lift siege (keep 1k there to keep the progress) that you had full cannon bonus more than three times in a row to kick them out and have to do such a thing for multiple forts in the same war, you no longer see it as confirmation bias. As Ethiopia when fighting Ottomans, it got to the point of me having to get call for peace as soon as I got more than threshold. Why had I not had 100%, well despite not having lost a single battle and in fact have wiped their army the forts just didn't fall so I started to get WE because of that.
Another case of AI turbo siegeing was when I trolled AI France ally by dowing England for Gascogne and calling them in, the siege of Labourd (which I was the leader) took much more time then the 2 sieges that France did and led. Those sieges were done by the same army in sequence, with no siege general or cannons. All sieges were unblockaded ports, I had cannons in my and started it before they started theirs. Look at how outrageous fast they were siegeing Normandy, 2 months of war and they had walls breached and it did not stop there but I forgot to screenshot those. And it was not one siege but 2 in sequence, if there were more forts they might have taken a third one with that one army before Labourd fell.

It happened so many times of the AI turbo siegeing that remembering when it does not happen is much easier.

Everyone else brings stats you bring anecdotes

 

[grommile]

Look at how outrageous fast they were siegeing Normandy, 2 months of war and they had walls breached

The requirement for a breach is very simple: Roll 14 on the siege die. Indeed, roll the very 14 that you have captured in your screenshot!

I've rolled breaches in the first month or two myself.

 

[Aries666]

The requirement for a breach is very simple: Roll 14 on the siege die. Indeed, roll the very 14 that you have captured in your screenshot!

I've rolled breaches in the first month or two myself.

Indeed this isn't even a rare thing seeing as it is a 1/14 chance!

 

[Ziggy187]

I just ran another data set last night and I may have discovered another anomaly. I started a new Iron Man run as Brandenburg and to save time I would rival Pomerania and declare war on Dec 11, 1444. I got up to about 80 dice rolls or so and the data seemed to be within reason and I was beginning to think maybe my hypothesis was bogus. Small sample set, I get it...
So I tried another run but decided to play more realistically like an actual player would and take the time to get a Claim and go to War for conquest. Wow, the results were night and day difference. Coincidence? Maybe but maybe not. Too small of data set to conclude as of yet. I got (6) 1's and (3) 14's with Humiliate CB but a conquest CB yielded (12) 1's and (2) 14's on about 80 dice rolls. This is quite a HUGE spread in the data.
HERE IS MY POINT. All the data I have seen posted here or even my own data sets show the 1 rolls always outnumbering the 14 rolls. If this were truly random, wouldn't we be seeing some data sets where the 14's exceed the 1's? It doesn't happen, it comes close to being tied but the times that 1's far exceed the 14's is more like the rule than the exception. I will continue to compile this data. Thank you for all the comments and feedback on this post. Please try to get data sets on Iron man using Conquest CB's like you would in a real actual game.

 

[Chaingun]

I didn't trust the source RNG implementation when I came to EU4, but I did test it empirically, and it converged to the mean one would expect. IIRC there's a "testrng" console command remaining from that, though I'm unsure if output means anything without knowing context.

The source RNG generates a 31 bit positive random integer or alternatively a floating point number in the range [0, 1[. However, there's a possibility for the programmer to use generated numbers in an erroneous way due to lack of high level random variable shaping functions in Clausewitz. I've seen nothing of that in dice roll code though since that is utterly simple. e.g. die_roll = random31uint % 6 + 1; (lazy almost linear approximation of one die).

I think the bigger issue is the fact pure RNG is used which will sometimes be unfair by its very nature. It would perhaps be cool to try something that adjusts the random variable's expected value to cancel out mean unbalanced series after some number of outcomes. Not my concern any more though, but I once read about some version of Civilization taking a similar intervention approach in battle outcomes to prevent players from feeling unlucky or feeling that the game cheats even though it doesn't. (Perhaps they simply reduced the probability of the really bad events for humans, can't recall.)

 

[tbhayward]

Boo! Tricking the game into making me feel like it isn't cheating me would be cheating me.

 

[iquabakaner]

I didn't trust the source RNG implementation when I came to EU4, but I did test it empirically, and it converged to the mean one would expect. IIRC there's a "testrng" console command remaining from that, though I'm unsure if output means anything without knowing context.

The source RNG generates a 31 bit positive random integer or alternatively a floating point number in the range [0, 1[. However, there's a possibility for the programmer to use generated numbers in an erroneous way due to lack of high level random variable shaping functions in Clausewitz. I've seen nothing of that in dice roll code though since that is utterly simple. e.g. die_roll = random31uint % 6 + 1; (lazy almost linear approximation of one die).

I think the bigger issue is the fact pure RNG is used which will sometimes be unfair by its very nature. It would perhaps be cool to try something that adjusts the random variable's expected value to cancel out mean unbalanced series after some number of outcomes. Not my concern any more though, but I once read about some version of Civilization taking a similar intervention approach in battle outcomes to prevent players from feeling unlucky or feeling that the game cheats even though it doesn't. (Perhaps they simply reduced the probability of the really bad events for humans, can't recall.)

In fact a lot of competitive RPG games would use a less random roll so that the game results are less affected by randomness. It is very easy to create a pseudorandom number generator with a low period (i.e. the number of rolls after the rolls starts to repeat itself).

Looks like someone skipped statistics class

I can get him some notes for the statistics courses I used to teach in the university as a tutor.

 

[Fishman786]

I didn't trust the source RNG implementation when I came to EU4, but I did test it empirically, and it converged to the mean one would expect. IIRC there's a "testrng" console command remaining from that, though I'm unsure if output means anything without knowing context.

The source RNG generates a 31 bit positive random integer or alternatively a floating point number in the range [0, 1[. However, there's a possibility for the programmer to use generated numbers in an erroneous way due to lack of high level random variable shaping functions in Clausewitz. I've seen nothing of that in dice roll code though since that is utterly simple. e.g. die_roll = random31uint % 6 + 1; (lazy almost linear approximation of one die).

I think the bigger issue is the fact pure RNG is used which will sometimes be unfair by its very nature. It would perhaps be cool to try something that adjusts the random variable's expected value to cancel out mean unbalanced series after some number of outcomes. Not my concern any more though, but I once read about some version of Civilization taking a similar intervention approach in battle outcomes to prevent players from feeling unlucky or feeling that the game cheats even though it doesn't. (Perhaps they simply reduced the probability of the really bad events for humans, can't recall.)

Lies! Yet another cover-up to hide the great Paradox conspiracy to irritate the fanbase and gradually destroy their perception of reality for nefarious purposes. Sooner or later you will be selling cosmetic DLC packs for £30 each, and everyone will be brainwashed to think the supposed price increase was just another example of cognitive bias at work! You can't hide the truth forever!

 

[Cymsdale]

I think the bigger issue is the fact pure RNG is used which will sometimes be unfair by its very nature. It would perhaps be cool to try something that adjusts the random variable's expected value to cancel out mean unbalanced series after some number of outcomes.

I hate when game programmers do this.

 

[Aries666]

I just ran another data set last night and I may have discovered another anomaly. I started a new Iron Man run as Brandenburg and to save time I would rival Pomerania and declare war on Dec 11, 1444. I got up to about 80 dice rolls or so and the data seemed to be within reason and I was beginning to think maybe my hypothesis was bogus. Small sample set, I get it...
So I tried another run but decided to play more realistically like an actual player would and take the time to get a Claim and go to War for conquest. Wow, the results were night and day difference. Coincidence? Maybe but maybe not. Too small of data set to conclude as of yet. I got (6) 1's and (3) 14's with Humiliate CB but a conquest CB yielded (12) 1's and (2) 14's on about 80 dice rolls. This is quite a HUGE spread in the data.
HERE IS MY POINT. All the data I have seen posted here or even my own data sets show the 1 rolls always outnumbering the 14 rolls. If this were truly random, wouldn't we be seeing some data sets where the 14's exceed the 1's? It doesn't happen, it comes close to being tied but the times that 1's far exceed the 14's is more like the rule than the exception. I will continue to compile this data. Thank you for all the comments and feedback on this post. Please try to get data sets on Iron man using Conquest CB's like you would in a real actual game.

Please stop digging. As people have already pointed out to determine whether or not RNG is working you need to account for the output of all possibilities i.e. 1,2....13,14. You are also using increasingly smaller samples despite people telling you you need more not less.

 

[Ziggy187]

Given that I only have 2 data sets to go off I ran a confidence interval to see if the theoretical odds we are told will fall within line of the actual experimental data. Since I got (18) 1's on the first 100 rolls and then another (18) 1's on the next 157 rolls, that totals (36) 1's out of 257 rolls. This gives a mean value of 14.0% chance of rolling a "1" per the data set. The sample standard deviation is 2.17%. In order to be within the 95% confidence interval, we need to be within 1.96 standard deviations of the mean. In order to have 95% confidence that my data set is indicative of the actual odds of rolling a "1" the data set should yield odds of 9.75% to 18.25%. Since the theoretical odds of a 1:14 chance is 7.14%, it does not fall within the 95% confidence range. I am 95% confident that the siege rolls do NOT have "fair" odds as we have been told.
I realize this may be a tough pill for some of you to swallow. I suggest nothing nefarious is taking place. This could be a simple program coding to double the odds of a disease outbreak on siege rolls in order to artificially make the game a little bit harder. There is certainly something at play here. I would appreciate if folks would stop with the judgment and preconceived notions and take a look at the hard data sets compiled in actual game play and not take someone's word for it because they said so.
Go on, click the "respectfully disagree" without leaving a post as to why, that's always helpful. There is always that 5% chance this is wrong and the data set is the anomaly. I'll let you guys decide...What does your data set tell you?

 

[tbhayward]

My data set tells me that the siege rolls are random.

 

[Aries666]

Given that I only have 2 data sets to go off I ran a confidence interval to see if the theoretical odds we are told will fall within line of the actual experimental data. Since I got (18) 1's on the first 100 rolls and then another (18) 1's on the next 157 rolls, that totals (36) 1's out of 257 rolls. This gives a mean value of 14.0% chance of rolling a "1" per the data set. The sample standard deviation is 2.17%. In order to be within the 95% confidence interval, we need to be within 1.96 standard deviations of the mean. In order to have 95% confidence that my data set is indicative of the actual odds of rolling a "1" the data set should yield odds of 9.75% to 18.25%. Since the theoretical odds of a 1:14 chance is 7.14%, it does not fall within the 95% confidence range. I am 95% confident that the siege rolls do NOT have "fair" odds as we have been told.
I realize this may be a tough pill for some of you to swallow. I suggest nothing nefarious is taking place. This could be a simple program coding to double the odds of a disease outbreak on siege rolls in order to artificially make the game a little bit harder. There is certainly something at play here. I would appreciate if folks would stop with the judgment and preconceived notions and take a look at the hard data sets compiled in actual game play and not take someone's word for it because they said so.
Go on, click the "respectfully disagree" without leaving a post as to why, that's always helpful. There is always that 5% chance this is wrong and the data set is the anomaly. I'll let you guys decide...What does your data set tell you?

Ugh, it doesn't matter what your descriptive stats say about a sample if your sample size is too small. If I flip heads four time that is not enough for me to claim that the coin is weighted, same principle here.