An exploration of randomness.
  • 266 posts
  • Page 7 of 18
jonboy1967 wrote:
My only question im asking answered is... Is there a part of that program that takes in account the fact 3 di are against 1. Not just the randomness. There is a naturaly built advantage to rolling 3 di vs 1, and it has nothing to do with randomness. If you want to just say your right, thats fine. But at least try to explain it without telling me...go look, go look at coin toss. Your talking about 1 item flipped a million times. Lets take a fight for instance. If i fight someone who is exactly the same power as me. Yes, we fight a million times, Its probably a 50/50 split. But, if its just me against three/two people exactly as strong as me. My chance of winning naturaly changed. I'm not going to win 1/10th of a million times. It a NATURAL advantage. Thats the part I think is missing. Anyway, I will continue to play. I like the site, and I think you guys do a great job and probably don't get enough credit. Peace
Cireon wrote:
Aha, I understand your question now.

Let me explain to you then the different things that can happen. I have to admit I haver personally looked at the code in detail, but I am fairly sure that all of these cases are properly handled in the code.
Before I start, note that if the attacking country has n troops, you are only attacking with n - 1 (i.e. a territory with 3 troops only counts as 2 while attacking).

Case 1 - Attacking 1 vs Defending 1
Both sides "throw" a die. The person with the higher roll wins (equality in favour of defender).

Case 2 - Attacking 2 or 3 vs Defending 1
Same as case 1, highest attacker roll counts.

Case 3 - Attacking 1 vs Defending 2
A defender can never roll with more dice than an attacker, so this case is equivalent to case 1, as the second troop is ignored.

Case 4 - Attacking 2 vs Defending 2
Both sides "throw" two dice. The highest roll of both players is compared like in case 1 and the same happens for the lower rolls, resulting in a 2-0, 1-1 or 0-2 loss.

Case 5 - Attacking 3 vs Defending 2
Same as case 4, but the lowest attacker dice is ignored.

Other cases do not exist, as the attacker can never attack with more than 3 and the defender can never defend with more than 2.

Well then. The best case is if you are attacking 3 vs 1 and you are saying there should be a "natural advantage". Well, this is not a war simulation game, first of all. That means that even with 20 vs 1, you will not be stronger than 3 vs 1. There is however an advantage: the attacker rolls 3 dice and the strongest is taken, while the defender only has one die. Hence the attacker has more chance (not 3 times as much per se!) rolling a higher number. If the defender throws 6, well bad luck, you won't win as attacker anyhow. If the defender throws 3 however, you need to roll 4 or higher as an attacker to win. Obviously, the chance in doing that in 3 rolls is higher than doing that in 1 roll, thus providing the attacker advantage.

I hope I answered your question. Like I said, I have not checked the code, but as these are the official rules, I am assuming this is how the attacking is implemented.

EDIT
Wikipedia has a nice overview of the changes in each of the above cases. It clearly shows the difference if you understand what it means: http://en.wikipedia.org/wiki/Risk_(game)#Dice_probabilities
“This is how humans are: We question all our beliefs, except for the ones that we really believe in, and those we never think to question.”
- Speaker for the Dead, O.S. Card
jonboy1967 wrote:
That is a wonderful explanation. It helps me understand how my strategy has been flawed. I think i should do better now. Thanks
TomC wrote:
you can say whatever you want, but I just lost a 20-1 fight... and that is just so unreal...

also losing 1-5 happens every single game at least 1 time = unreal

just no words for these dices and I think it's unbelievable how many people keep saying this is normal...
Matty wrote:
It is very much real.

It has happened before in history that one man could beat 20 others (and much more btw), just by being a very good and very smart warrior.

Also, with dice rolls the chance of rolling 20 consecutive 1's with all three dice (though you don't need 1's to lose 20 to 1) is not 0%, but a tiny bit higher, and thus not impossible.
Just improbable.

But randomness means improbable things happen, just not too often.

The problem with ppl is that they don't know what is too often, and what is not.

No offense, I don't realize it myself sometimes.
"Strength doesn't lie in numbers, strength doesn't lie in wealth. Strength lies in nights of peaceful slumbers." ~Maria
Cireon wrote:
Here we go again...
“This is how humans are: We question all our beliefs, except for the ones that we really believe in, and those we never think to question.”
- Speaker for the Dead, O.S. Card
Thorpe wrote:
Give them the link to this toptic
95.5% of the time you kill a players cap before your 2nd turn in... you fail or die next
elysium5 wrote:
Just a reminder as I've read a few comments in the game chats recently; The dice are the same for premium and non-premium members alike.
"I can picture in my mind a world without war, a world without hate. And I can picture us attacking that world, because they'd never expect it."
Dferguson wrote:
Read up and still think this generator is flawed... random does give light to the possibility of this happening but and yes the average outcome for kills and loses may add to that magic number of 53.??% but The amount of 1-0 wins skew the percentage to get it there (still realizing even if you win 1-0 you still lose one guy. I realize "pseudo-random" implies that this can happen but it happens, FAR, FAR to often in games:

all in the same game moments apart. 
3 Oct, 22:10 Goose Green (whitmirb) attacks Great Island (Neutral) killing 0 troops, losing 5.
3 Oct, 22:18 San Carlos (dalfire) attacks Port San Carlos (Neutral) conquering it, killing 3 troops, losing 11.
3 Oct, 22:32 Goose Green (whitmirb) attacks Darwin (Boub666) killing 4 troops, losing 10.

also alot of kill 3 lose 5 which is more likely to happen; no there were a few good outcome kill 20- loss 15. but their is a huge gap in losses were there is not in wins... this is only one game and more would change the overall results but again i believe the 1-0 wins are teh only reason the numbers balance. so why are there so many horrible losses but no so many fantastic wins? Personally think the dice on this site are not suited to fit a standard deviation. the losses would have far more outliers then the kills. 

I dont mind a loss but some are hard to swallow and do happen far to often.
elysium5 wrote:
Play real life craps and you will see that you have a succession of really bad rolls followed by a succession of really good ones and you will never have a contant 'close' average. You will have an equal amount of ups, downs and close rolls in no particular order that will average out in the long run.
"I can picture in my mind a world without war, a world without hate. And I can picture us attacking that world, because they'd never expect it."
Dferguson wrote:
well how about this my very next game:
4 Oct, 17:03 dragon007: 4 Oct, 16:55 Burgundy (Dferguson) attacks Rhône-Alpes (westchester) killing 2 troops, losing 5.
4 Oct, 17:05 Dferguson: 4 Oct, 16:40 Burgundy (Dferguson) attacks Rhône-Alpes (westchester) killing 0 troops, losing 6.
4 Oct, 17:05 Dferguson: 4 Oct, 16:47 Burgundy (Dferguson) attacks Rhône-Alpes (westchester) killing 0 troops, losing 4.

kill 2 lose 15... my game total kill 20 loss 35. I must be the most unlucky player on the site or there is a fault with the dice generator. Ye those losses should happen but not ever damn game.
lifeinpixels wrote:
Everyone gets those all the time. Only a few complain.
Additionally, our brain tends to remember the bad over the good. The good rolls will balance out the bad ones, but we don't usually take note of them.
urgul wrote:
its human nature. thats why if you aced one test and failed one the same day, you will think much more about the failed than the aced. its weird.
aeronautic wrote:
Here is a question to the programmers of the dice program.
Are the attack dice random generater program and the defence dice generater program run inependantly and not associated to one another until the results are read by the program and compared to give the end result?
Or are they all generated and compared in the same code area.
The reason I ask, is I see patterns in the dice in every game and my belief is that although written correct and true to randomness, it is what a computer sees as random that concerns me. I strongly believe it is influenced by it's own results within a calculation (generation of 2 sets of random results) and this is why you see 654 / 65 & 211 / 22 regularly where every high roll and low roll is synchronised and is where the extremely bad rolls (from a human perspective) are experienced.

It is just a theory, but have you tried running the 2 dice rolls (attack / defence) in sepearate generaters and store the results seperately so that the program does not know the result until it is called for analysis?
In addition to this, can the results be taken to a different storage area and deleted from the programs memory?

If this can be done, I believe you will have the truest form of randomness from a natural perspective.
I believe it will then allow itself to only deal with 3 dice for 1 roll and 2 dice for 1 roll independantly and not compare its own results and not try and create more randomness, which could result in the program trying to give an even spread of results, which as explained, can cause a synchronisation of high and low rolls between the Attack and Defence dice.
Hyd yn oed er fy mod Cymraeg , dim ond yn siarad Saesneg, felly yr wyf yn gobeithio y bydd y cyfieithu yn gywir.