dice rolls....

[moshpitkosh]

are so ♥♥♥♥♥♥. ive been keeping records. the AI gets a HUUGE advantage. my record (in the past 3 days) so far is rolling (and this counts for the whole 3 tick cycle) 0 0 0 1 0 1 1 0 2
meanwhile the AI rolled 9 7 6 7 5 8 7 7 6. thus 50k men defending a mountain pass were defeated by 20k.... that crosssed a rivver and assaulted a mountain pass... a -3 modifier.... the generals didnt factor much. i had a +1 bonus on fire, they had a +1 bonus on charge. this is with level 16 military tech.

things happen. you get ♥♥♥♥ rolls. no plan survives encounter with the enemy blah,blah,blah. but this happens WAAAAAAAY too often. in my current war. my coalition is 50k men stronger (200k vs 150k). yet we cant win a single battle due to dice roles and of course the AI ♥♥♥♥♥♥♥ around on the other side of the map FAAAAR away from where the enemy is (northern germany vs northern italy. when the enemy fights their forces all come running. when i fight no one shows up. but small little armies WILL charge in against 60k strong armies AFTER the battle ended (they dont have the little locks so they can turn around, they just choose not to)

so my question is does anyone know if paradox is going to do anything about this or are they just going to say "thats life, AI needs bonuses to be competitive.

[lukas]

You don't seem to understand the concept of randomness. Your records prove nothing.

[ancistrus]

Not taking sides, but there is a point when just saying random is not enough. Its been a while since I had ...any class, but iirc there are ways to determine if a random sample follows uniform distribution. So OP you might want to take things a step further.

http://www.xkcd.com/221/

[Exarch_Alpha]

This was discussed a billion times already, dice rolls are not one of the ways the AI has to cheat. If you still have doubt about it post in the official forum so that for the 1000th time a developer explains it to you.

[moshpitkosh]

ya ya figured that was the response id get from the normies. so if i posted 100 battles and they showed the AI rolled 436 points above (cumulative, which would show about a 19% preferance for teh AI) my rolls that wouldnt mean anything to you smarties? you cant have random rolls with a program. in case you didnt know that. its based on an algorithm, and if its flawed (which they all are, to varying degrees) the results will be skewed. but this can easily be countered with - if player X rolls 3 0s in a row add a plus 2 and re-roll any zeros for 2 rolls or w/e along those lines.

also thanks for not answering the question and contributing nothing.

and thank you ancistrus ill check into that. i appreciate your constructive response.

[lukas]

Originally posted by moshpitkosh:

ya ya figured that was the response id get from the normies. so if i posted 100 battles and they showed the AI rolled 436 points above (cumulative, which would show about a 19% preferance for teh AI) my rolls that wouldnt mean anything to you smarties? you cant have random rolls with a program. in case you didnt know that. its based on an algorithm, and if its flawed (which they all are, to varying degrees) the results will be skewed. but this can easily be countered with - if player X rolls 3 0s in a row add a plus 2 and re-roll any zeros for 2 rolls or w/e along those lines.

also thanks for not answering the question and contributing nothing.

and thank you ancistrus ill check into that. i appreciate your constructive response.

Yes, the results of 100 battles don't prove anything. It's as simple as that.

Yes, those algorithms have flaws. The change you propose probably will worsen the situation though. Since we don't have access to the code we can't know for sure.

I recommend you to open a thread over at Paradoxplaza too, since you have a better chance of catching the eyes of a developer there. You might also get more well-grounded responses.

That's what you get for starting a thread based on a hunch. Handling criticism doesn't seem to be your strongest suit either. A perfect combination for disaster.

[kaiyl_kariashi]

100 battles is nothing and is a joke of sample size to even make a remote claim of something being wrong.


I can roll a physical dice 100 times and still get ♥♥♥♥ results, and there's no fault there but pure chance. No algorithrims to blame, no cheating Ai to be there. Just pure chance.


And the thing is...the algorithim used is the same for BOTH the ai and the player, so it doesn't matter if the algorithim is flawed it's still being applied equally and you're just having bad luck.

It's just as basic RNG script that simulates a d10. Nothing more, and nothing affects it anywhere in the game files aside for a line that allows you to change the size of the dice used in the calculation.

You're just having a run of bad-luck and too small of a sample size for the statistics to bare out.

[Rooter]

The game cheats like a mother ♥♥♥♥♥♥ or Im as unlucky as you can get.
Anyone noticed the "increased" amount of army retreats, instantly after you are move locked for support?
Anyone noticed the "increased" amount of enemy AI sieges that win at 14 or 28% just before you arrive?
Anyone noticed the sudden appearance of rebels in foreign countries that pop up directly in your armies travel route when sending them long distances?
What is the go with fort zone control at the moment? AI seems to ignore it more often than not?

[Kijin Eira]

good rolls in this game? nah.

[Blue Knight™]

Originally posted by moshpitkosh:

ya ya figured that was the response id get from the normies. so if i posted 100 battles and they showed the AI rolled 436 points above (cumulative, which would show about a 19% preferance for teh AI) my rolls that wouldnt mean anything to you smarties? you cant have random rolls with a program. in case you didnt know that. its based on an algorithm, and if its flawed (which they all are, to varying degrees) the results will be skewed. but this can easily be countered with - if player X rolls 3 0s in a row add a plus 2 and re-roll any zeros for 2 rolls or w/e along those lines.

also thanks for not answering the question and contributing nothing.

and thank you ancistrus ill check into that. i appreciate your constructive response.

Babe ur a normie to me and the rolls are random gtfo noob

[ancistrus]

The point I was trying to make is that 100 rolls actually might be a decent sample, provided that the observations are sufficiently different from what you would expect. If I tossed a coin 100 times and didnt get a single heads, would you call that sample a joke? I am not a statistician, I was never that interested in it, but I recall using even fewer than 100 observations to make a statement like "based on the data and using 5% significance level we reject the hypothesis that the random variable follows normal distribution bla bla"

[Blue Knight™]

Originally posted by ancistrus:

The point I was trying to make is that 100 rolls actually might be a decent sample, provided that the observations are sufficiently different from what you would expect. If I tossed a coin 100 times and didnt get a single heads, would you call that sample a joke? I am not a statistician, I was never that interested in it, but I recall using even fewer than 100 observations to make a statement like "based on the data and using 5% significance level we reject the hypothesis that the random variable follows normal distribution bla bla"

well what if your 100 rolls you got 1-4 rolls most but my 100 rolls i got 4-7 and someone elses 100 rolls he got 7-above then what, having a statistic of 50 people is still called fact sometimes by people who want to make a point, so people can easily rig statistics, they are all random

[moshpitkosh]

Originally posted by Rooter:

The game cheats like a mother ♥♥♥♥♥♥ or Im as unlucky as you can get.
Anyone noticed the "increased" amount of army retreats, instantly after you are move locked for support?
Anyone noticed the "increased" amount of enemy AI sieges that win at 14 or 28% just before you arrive?
Anyone noticed the sudden appearance of rebels in foreign countries that pop up directly in your armies travel route when sending them long distances?
What is the go with fort zone control at the moment? AI seems to ignore it more often than not?

i have noticed some of those things. the AI seems to be able to retreat after 1 day (which is historicaly accurate) yet i cant retreat after 3 days. i have not noticed the AIs sieges being being better/faster... but i have noticed the revolts that quickly happen when i occupy a provence.

Originally posted by ancistrus:

The point I was trying to make is that 100 rolls actually might be a decent sample, provided that the observations are sufficiently different from what you would expect. If I tossed a coin 100 times and didnt get a single heads, would you call that sample a joke? I am not a statistician, I was never that interested in it, but I recall using even fewer than 100 observations to make a statement like "based on the data and using 5% significance level we reject the hypothesis that the random variable follows normal distribution bla bla"

i feel that 100 rolls is a decent sample even though statisticly 1,000 is the base sample size for credability. my point earlier (which i think you caught) was that out of 100 battles the AI had a +400 point advantage (cumulitive) on the rolls. this coupled with the negative modifiers that i usually exploit indicates a flaw in the RNG logic. which has always existed (but ♥♥♥♥ does happen so i havent been too upset about it) and it seems to have gotten worse after this latest patch.

also....

Originally posted by kaiyl_kariashi:

100 battles is nothing and is a joke of sample size to even make a remote claim of something being wrong.


I can roll a physical dice 100 times and still get ♥♥♥♥ results, and there's no fault there but pure chance. No algorithrims to blame, no cheating Ai to be there. Just pure chance.


And the thing is...the algorithim used is the same for BOTH the ai and the player, so it doesn't matter if the algorithim is flawed it's still being applied equally and you're just having bad luck.

It's just as basic RNG script that simulates a d10. Nothing more, and nothing affects it anywhere in the game files aside for a line that allows you to change the size of the dice used in the calculation.

You're just having a run of bad-luck and too small of a sample size for the statistics to bare out.

100 battles easily accounts for 1,000 rolls. which IS A DECENT SAMPLE SIZE. and again with a negative modifier placed upon the enemy (crossing a river to attack an entrenched army in the mountains) they shouldnt win every (read: majority of the) time. ALSO we both know that the RNG isnt a dice roll. and after 1,000 hours in game i would think anyone would notice that it is a bit off.

regarding my original question... does any know if pardox is addressing this issue or are they just saying "life sucks"

also. do the math. what are the cahnes that you will get a +400 point bonus out of rolling 2 dice 1,000 times? its pretty small

[moshpitkosh]

Originally posted by KoA Blue Knight™:

Originally posted by ancistrus:

The point I was trying to make is that 100 rolls actually might be a decent sample, provided that the observations are sufficiently different from what you would expect. If I tossed a coin 100 times and didnt get a single heads, would you call that sample a joke? I am not a statistician, I was never that interested in it, but I recall using even fewer than 100 observations to make a statement like "based on the data and using 5% significance level we reject the hypothesis that the random variable follows normal distribution bla bla"

well what if your 100 rolls you got 1-4 rolls most but my 100 rolls i got 4-7 and someone elses 100 rolls he got 7-above then what, having a statistic of 50 people is still called fact sometimes by people who want to make a point, so people can easily rig statistics, they are all random

100 rolls at 2.5 (1-4) is 250 pts.100 rolls at 4-7(5.5) is 550 pts thats still only a differance of 300 points. which is well outside the range of a 440 pts advantage by a margin of over 30%.

[Blue Knight™]

Originally posted by moshpitkosh:

Originally posted by KoA Blue Knight™:

well what if your 100 rolls you got 1-4 rolls most but my 100 rolls i got 4-7 and someone elses 100 rolls he got 7-above then what, having a statistic of 50 people is still called fact sometimes by people who want to make a point, so people can easily rig statistics, they are all random

100 rolls at 2.5 (1-4) is 250 pts.100 rolls at 4-7(5.5) is 550 pts thats still only a differance of 300 points. which is well outside the range of a 440 pts advantage by a margin of over 30%.

im not making a point about average im telling you that no matter how many tests you do itll always be random, your ai could get 100 10 rolls ffs and so could i and so could u and THEN THAT EVEN WOULDNT PROVE ANYTHING

[ancistrus]

Originally posted by KoA Blue Knight™:

Originally posted by moshpitkosh:

100 rolls at 2.5 (1-4) is 250 pts.100 rolls at 4-7(5.5) is 550 pts thats still only a differance of 300 points. which is well outside the range of a 440 pts advantage by a margin of over 30%.

im not making a point about average im telling you that no matter how many tests you do itll always be random, your ai could get 100 10 rolls ffs and so could i and so could u and THEN THAT EVEN WOULDNT PROVE ANYTHING

Yes, technically true. But I guess you would be pretty horrified if you found out just how many decisions that influence your life were based on tests that "dont prove anything". Like, 90% of medicine for example.