Start a new topic

New Dice Algorithm / True dice rolls

Game does not use true dice roll probabilities. Many, many times I've had 97-99% chance to win and dont.

HOW TO VOTE FOR THIS FEATURE? Tap the 'Do you like this idea?' below

91 people like this idea

Eh... If your rank actually falls back down to novice you should be playing novices and beginners. The point of ranks is to find others of similar skill. Otherwise, it's a huge waste of time and no fun. And why would a grandmaster want to play games where coming in 2nd place will knock away a lot of points and winning give next to nothing?

Kevin, I have to agree with Peter.  What you said doesn't make any sense.

"Have your ever noticed how in your dice roll statistics that the percentages are flat across the board??? This proves that the dice rolling is flawed."

If anything, your observation is evidence for the exact opposite conclusion.  If you played games with real dice and kept stats on your rolls, you should expect to see that each individual roll (1,2,3,4,5 or 6) would happen about as much as any other.  As you play more games, you'd expect to see those percentages getting closer to each other.  This is because rolling a 1 is no more or less probable than rolling any other number, that is, every individual roll is equally probable.

If after many games you noticed you rolled ones (or any other roll) a few percentage points more or less than any other roll, that would be evidence that the dice rolling scheme isn't true to real dice.

If you don't believe me get out some real dice, roll them about 600 times (or get 600 dice and roll them all together one time, ha!) and keep track of how many times you rolled each number.  You'll see that you roll each number roughly the same number of times as any other number.  If you then calculate what percentage of rolls come from each number, you'll see that the percentages are very close to one another, and these %s get closer to each other as the number of rolls grows large.

I agree. I have a new account that in 5 games has a 23% for one and 15% for everything else. My main account which has maybe 500 games has developed a 17% on the one and 16% for the others. I suspect there is a bug in the algorithm for selecting the dice roll results.

 The dice sequence is dependent on what SMG refers to as "transition matrix" (TM). The random number sequence if conducted in a "serial mode" requires many more random numbers (one for each state) than the TM approach. In Discrete Event Simulation, to go from State A to State B given a probability P, throw 1 random number in a uniform probability distribution between [0,1] and assign State A if the random number is < P, or State B if the random number is > P. Thats for a simple transition of one state to one other. Now if you want to go from State A1, A2, A3...Aa  to State B1, B2, B3 ..Bb  then you would need an axb size matrix, where the matrix elements correspond to the individual specific transitions. I would think SMG ran say 10 million dice simulations with all types of combinations to populate their overall BlitzMode TM. They have to simulate enough transition states to get whats called The Law of Large Numbers (normal distribution) mean and variance..   The issue is that their matrix element values may be juiced or skewed so when the one dice is thrown Blitz Mode, it might not seem to be in accordance with the probability distribution of the TM.

The Random Number Generator (Merseinne Twister) is a very robust algorithm. However the TM values are not subject to the same statistics as the generator.

If SMG gets the message their TM elements are hosed, then they will correct it.

There is a small probability that 5 armies will prevail against 15.

Good news guys. I have been in contact with support about the heavily biased 1's issue and they have acknowledged that this was a glitch in the 1.9.36 release. They say they have remedied it as of the latest release that came out today, so hopefully this should no longer be a concern.

Justin, I completely agree with you about the dice rolling ones! As you can see in the picture attached to my post right above yours, I have a 21% chance of rolling a one and only 15% chance of rolling every other #! That is no where close to normal or fair. A full 6% difference between the chance of rolling a one compared to rolling any other number is NOT statistically possible if the dice rolls algorithm is truly random. I have played over 120 games so this isn't the result of a small sample size.... I agree this is ridiculously unfair and something they need to figure out a way to fix!

It may also be worth noting that they said they now have a system in place that records highly unlikely outcomes so they can keep track of how often they are occurring in actual gameplay. This means they are at least not blind to the results their new algorithm is producing and should hopefully be able to asses things and adjust accordingly.

FYI, the new algorithm has been programmed to explicitly give Novice users an unfair advantage by heavily biasing their rolls away from 1’s. I’ve investigated this thoroughly and it’s undeniable—create a new account, play a game, then check your stats, you can easily observe this yourself. It’s obviously a highly unethical attempt at keeping new players happy and encouraging them to buy the full game. I suggest you ban all Novice users from any games you create if you want a fair playing field. Please open a new ticket and complain to game creators. We need to hammer them until they decide to be truthful with their paying customers and undo this cheap marketing trick that ruins the game for everyone else!
SMG has clearly stated that they do not alter dice rolls or weight them.

I started playing SMG Risk Global Domination a few weeks ago. I saw "balanced" dice rolls and wondered why they would have this option. This thread explains a lot. A professional game should have only true dice rolls and not this silly balanced.

I kept looking for a good explanation for balanced dice rolls (which should be called "skewed", or "flawed"). With balanced a 5 vs 1 attack wins 100% (say what!?); with true it wins 97% (as it does in real life).

When you know the difference between balanced and true you have an advantage over your opponents. A good explanation can be found at in the section "Dice, Blitz and Slow Roll".

By the way, I first played Risk in 1967 with those nice wooden cubicle armies and sometimes we played with the original French rules.




Have you been able to advance a player to beginner rank without the novice making all but one roll so that you can test dice rolls at the beginner level with a clean slate?

What happens when a novice rolls again a novice? The attacker wins most of the time? So it's all about offense for novices?

I boot novices anyway. And usually beginners. and if SMG would take a moment to program filtering I'd only host games for masters and grandmasters. But you know, your mom.


The question is. How are these stats composed?
Is it from normal rolls only or from Blitz rolls as well, I suspect it is not from all rools as Blitz is probably using some shortened algorithm.

And what Briand is saying is, that if you 1s are rolled less frequently than the defender is getting a higher advantage than he is supposed to, compared to true dice rolls.

There is another very disturbing statement in the Faq, I quote: "The results are within +-5% of the control comparison, which is acceptable."

Uhm no, it's not acceptable at all, when I attack with 8 vs 1 and the chance of winning is 99,9% and instead it's 95%, because the difference is like night and day.

5% deviation is absolutely unacceptable for a trivial algorithm like dice throwing ;/

Well, if everyone is rolling extra ones, it's not "unfair." But it is very surprising to see everyone on the map suddenly lose big attacks that you'd think they'd easily win. I had one game where everyone got their big army, started owning the map and would lose out big time on one of the blitz attacks. I'm talking like 40 armies losing against 10 kind of attacks. I've witnessed a major increase in high losses on some attacks and watch others bite the dust in the same game which has been unusual although not always impossible to happen.

Also, in one game I had 2 manual rolls where both me and the defender all rolled 1's. That doesn't happen often, or ever and it was twice in a game. I'm usually using blitz so that's crazy.

Anyway. It's certainly making for a less predictable game. I used to feel confident blowing through people knowing that on average, the attacker does better than 50% of the kills. Now I'm used to needing well above the average to be sure and I'm still getting burned sometimes. Same as the defender of course but I have no choice in that.


Briand, That level of complexity is not needed to test what I’m talking about. It’s a simple biasing away from rolling 1’s for any user with a Novice rank, doesn’t matter who they are rolling against or in what scenario. Create a new account, play one game as a Novice, check stats — You will find roughly a 12% rolling of 1’s and roughly 17% rolls for every other number. Create any number of new accounts to test this and you will find that distribution bias every time (keep in mind probability still applies so you may not end up with those exact numbers but something close).
Forgot to add, I checked on one of my accounts with 100+ games. All of my accounts in the past have had 16% across the board. This one now has 17% on the 1's after about maybe 10 games. "excited" to see how much this % increases over time :)


Login or Signup to post a comment