Randomness of dice

[bck1234]

I know the die are supposedly, but I am in my third online game in which 2-3 bombers are strategic bombing Berlin and the defender has rolled snack eyes. 7 bombers lost over 3 games. I guess I need to play a 1000 more games to see my luck even out. Just a rant.

[Dbl390]

No, your correct, i just lost another game by a complete one sided blow out. If they don't institute low luck for the Rank/Tournament play, I will post negative feedback everywhere I can.

[oddball]

There is clear evidence (p value 0.006) that the dice favor the defender.

This matches the experience of many commenters here including mine.

Remember it affects all players so it's not unfair exactly - just adjust your strategy accordingly.

[oddball]

Originally posted by Dbl390:

No, your correct, i just lost another game by a complete one sided blow out. If they don't institute low luck for the Rank/Tournament play, I will post negative feedback everywhere I can.

They prefer to lock threads raising the dice issue rather than investigate the bug.

[ibew934dcc]

Predictable patterns are not random. You will always loose one battle badly no matter What and defensive dice hit at much deadlier rate. yes I have played thousands of hours and 10,000 of thousands online and 40 plus years with dice in hand. Other than one friend I had years ago God rest his soul who could snake eye you to death all day long these dice on this platform are the cruelest ever in my 40 plus years of game play....I do not need lectures or test dice by anyone thanks not a true measure of game play , Test dice in the manner it has been done does not represent real time battle conditions. I am sure the dice are 1000% random in code. I still question situations and certain properties being programmed "weighted " and less random....Not accusing anyone of cheating just my observation we all know what we are dealing with by now and change our game play accordingly not a good plan if it cant be altered. The dice give and take no matter how smart you think you are. I used to fight till the end in every game for I find it satisfying myself after long game to roll it out and get the Victory so hate to take that feeling from others when i am loosing i honor them by playing it out. No more if the dice say you cant win this one after three rounds ....I am gone I have noticed others doing this as well many pulling the pin after a bad first round .... No doubt capitals are weighted against the attacker in my opinion not fact opinion but for me I calculate it as a stone cold fact if I do not have a huge advantage or the game is on the line I will not attack knowing 80% kill ratio or better is facing me in defense ....some territories I will not attack period unless I have 10-1 advantage some i just will not attack period. I will Not waste bombers after IPC's unless game is over and you are pinned down. 1/6 my arse take one bmr in might as well take it off the board or enjoy the one IPC you will cost them not worth it unless you are bombing me I can not remember last time an opponent lost a bmr bombing me. Well now I just proved it was random since I loose every bomber I send I guess it does even out that I never shoot one down in return......Rant over but the dice Gods give and take away no matter your ranking, helps knowing the game, but it stings when the dice turn bad. I stop playing till next day when patterns turn south and apologize to all not slow playing, but will not keep watching impossible rolls even when I am winning the contest....I ask what does a screen shot prove ? how can anyone share any possible evidence without saving entire games in the hundreds and look at round by round game play? The old game used to give you a % of hit to miss in each in game stats win loose or draw, I have seen it be as one side as 80% once and a while. Here 80% kill rate is average for Wins not just once and a while allowing for few close games that are not decided from the first round or two or one bad battle down the road,sure seems like I face a 75% of better kill rate against when attacking always especially early on. I can take 50 three against 30 twos and loose every time.....I want to know How all you guys that use fighters and bombers to strafe attacks never loose a plane. I loose half or more every time?

[Julius Borisov]

Originally posted by oddball:

Originally posted by Dbl390:

No, your correct, i just lost another game by a complete one sided blow out. If they don't institute low luck for the Rank/Tournament play, I will post negative feedback everywhere I can.

They prefer to lock threads raising the dice issue rather than investigate the bug.

Stop spreading lies, please. Just stop it right there. Moderation of the Steam forum is separate from the game.

There is no bug with the dice though. We've released multiple pieces of evidence many times, and this argument has been settled. The community has come up with thousands of rolls to confirm: the dice are random. Check the stickied thread please.

[KagawaShinji]

"There is clear evidence (p value 0.006) that the dice favor the defender."
Very Happy to read from you, Oddball. Can u further elaborate this?

" I can take 50 three against 30 twos and loose every time..."
100% true. Cant tell u how many times i attacked moscow with like 15-20 tanks + fighters and bombers and get destroyed by pure INF+ART. Every. Time.

[oddball]

Originally posted by Julius Borisov:

Originally posted by oddball:

They prefer to lock threads raising the dice issue rather than investigate the bug.

Stop spreading lies, please. Just stop it right there. Moderation of the Steam forum is separate from the game.

There is no bug with the dice though. We've released multiple pieces of evidence many times, and this argument has been settled. The community has come up with thousands of rolls to confirm: the dice are random. Check the stickied thread please.

I have checked the thread - and your testing is not sufficient at all.

You have simply rolled large numbers of dice in large battles and discovered that the average number of hits is correct.

The problem with this is that the most likely form of bug, given the RNG you are using, is that you are 'seeding' the RNG with the same seed at the same point in the battle code, and that the results are therefore skewed for the first few dozen rolls because the sequence is heavy with - for example - 2s in the first fifty numbers - and this sequence is repeated every battle.

In a battle where you roll hundreds or thousands of dice the 2s would be balanced out by a smaller number of 2s later on in the RNG sequence (that's how the RNG you are using works) - but that is not how the game is played. In practice we actually roll just a few dice in the vast majority of battles - and in my experience the effect is strongest in tiny battles.

And you don't need to test with thousands of dice. You only need to roll a few hundred to get statistically reliable results if the dice are badly off. Even KagawaShinjis sample of 150 rolls produces useful information.

[Julius Borisov]

There is a spreadsheet with hundreds and thousands of rolls, you ignored that. There are thousands of hours I spent in this game (up to 6k considering the pre-Steam version), and my experience is the dice are not off.

Whatever we provide, in whatever form, will never be sufficient for those who don't want to see past their initial argument. Btw, the OP of another thread you necroraised, they switched their opinion to the opposite eventually, just never updated the thread.

[oddball]

Originally posted by Julius Borisov:

There is a spreadsheet with hundreds and thousands of rolls, you ignored that. There are thousands of hours I spent in this game (up to 6k considering the pre-Steam version), and my experience is the dice are not off.

Whatever we provide, in whatever form, will never be sufficient for those who don't want to see past their initial argument. Btw, the OP of another thread you necroraised, they switched their opinion to the opposite eventually, just never updated the thread.

1. No. That's not true - at least not in my case. I just want to see a statistically significant test of small battles - the type of battles fought in actual games (preferably with all the game/ dice animation settings set to the default - just in case they somehow affect the timing of interactions with the server).

That's all I've ever wanted to see - and as far as I can see nothing has been provided that shows this. For example the sticky thread provides results where 100 bombers attack. I don't care what happens when 100 bombers attack in one giant bombing run. I care what happens when one single bomber attacks 100 times.

If I have somewhere missed the testing that I would like to see - and frankly the only testing that matters because testing giant battles that never happen in games, while it is not entirely worthless, is not even close to being sufficient testing - please point me at the spreadsheet that shows a thousand 1ftr + 2inf vs 1 inf battles - or something similar.


2. I don't believe I've necro-raised any threads. I don't scroll more than a page or two into comments - and I comment on threads I find there. Perhaps someone else necro-raised it and I commented thereafter given that the thread is still relevant because you guys seem to have failed to perform basic valid realistic game scenario testing even after all these years.

3. The only dataset I have seen for the types of scenario I would like to see tested is the one posted the other day. And it clearly showed a problem. Now there is a chance (6 in 1000 to be precise) that the player obtained his results by chance. I've played a lot of hours myself though, even if not as many as you, and I have definitely observed results which are skewed towards defenders. And I've seen this help me just as often as it hurt me.

[Juggernaut]

Originally posted by oddball:

please point me at the spreadsheet that shows a thousand 1ftr + 2inf vs 1 inf battles - or something similar.

I don't believe I've necro-raised any threads.

You necro raised a thread literally yesterday when you were replying to 4 year old comments cus you were tilted that I slapped your ragepost so hard.

As to your request, did we FINALLY get you to actually make an assertion? Are we going to go in and bet for 100 bucks that 2 inf 1 ftr vs 1 inf favours the defender over say, 300 battles?

I will set up a live game with you and take your money any time. I'll even double it back your way. 200 if you win. What I wont do is go waste 3 hours testing some random ultraspecific thing that you will just weasel your way out of to whine about some other random test the next day. I want you to be involved and clear about what you want tested.

[montrealsteak]

What I don't understand is why complain when they said they won't change the RNG or whatever you call it. So what if the dice favors the defender? We all get to defend and attack so it's a level playing field.

The only thing the dice complaining does is it gives the game a stigma and potentially hurts sales. Therefore, the logical conclusion is that the dice complaining is intended to kill the game. Fine.

AAO is awesome. It brings to life a great board game that is inaccessible to 99.99% of the population. I'm on old man that has few friends and a lot of work, so this is the only way for me to play A&A. If the dice isn't perfect, then that's ok because the game is still the most entertaining, exciting and cost effective way to game online. In fact, all the dice complaining shows how this game can get emotional and that is one of the things that makes this game great. Passion.

[Juggernaut]

Amen^

[4touchdowns1game]

Oddball, you referenced KagawaShinji‘s post regarding dice rolls. He had 152 samples and showed 43% hitbacks. You state it should have been 33% hitbacks.

Yet in his original post he showed all seven samples included an attack on west russia. All the battles in his post were battles that typically take place round 1. His screenshot he supplied showed the board after russia’s round 1 turn.

There’s a German tank that starts in west russia. That German tank dies last. So it gets multiple rolls/battle. Those rolls are at 50%, not 33%.

So the premise that the attacks should have hit at 33% just isnt accurate.

True, he did state all the results were from artillery and infantry hitbacks, but then he also included a bunch of sea zone battles in his results.

Two more problems. 152 rolls simply isn’t enough to state anything definitively. And we have no idea if his evidence is cherry picked. The only evidence he provided certainly implies that it is.

There’s an easier way to gather data. If you really believe the dice a skewed to the defender. Simply add up the Russian casualties in all round 1 Russian 12/9 attacks.
On average:
West Russia should see 3 infantry killed
Ukraine should see 2 tanks surviving

Simply add up the numbers and see for yourself.

My experience with the game is the same as you see in real life. The dice are wild. You get crazy results in all dice and card games.

The idea the dice favour the defender is emotional. It’s because you only see the dice when you’re attacking.

Anyway. Cheers and happy gaming.

[4touchdowns1game]

For fun I went through my game history. I only have six in my history. All had 12/9 attacks.

Based on surviving units in west Russia (expected 3) and surviving tanks in Ukraine (expected 2) I assigned a + or minus to the battle. + in favour of the attacker. - for defender

Game 1: 3 inf wr. (0 to expected)
3 tanks ukr (+1 to attacker)

Game 2: 1 inf wr (+2)
1 tank ukr (-1)

Game 3: 3 inf wr (0)
Ukr -German fighter survives, 2 Russian fighters retreat. Based on expected future hits I assigned at (-3)

Game 4: 4 inf wr (-1)
1 tank ukr (-1)

Game 5: 4 inf wr (-1)
Retreat in ukr with 2f, 3t. 1 fighter survives. I assigned (+0)

Game 6: 3 inf wr (0)
3T 1 Art ukr (+2)

So overall that’s (-2). 2 extra kills for the defenders over 12 battles (conservatively around 300 dice rolls).

Anyway, take a few minutes, check your games and do the math for yourself.

[Julius Borisov]

Originally posted by oddball:

I don't care what happens when 100 bombers attack in one giant bombing run. I care what happens when one single bomber attacks 100 times.

But that is what that spreadsheet provides exactly. It started initially with Quintin and other players just putting in their results from combat one-by-one. Nobody did a 100-bomber raid or anything similar. Those are results of many, many raids in multiple combats. Same of other combat. There are a few comments by myself on Discord where I counted bomber raids in my games, all seem to be even over the course of the game - or multiple games.