Implemented

It's not the random number generator but how the algorithm uses the random numbers that is far from board play in Blitz mode. SMG offered some clues their Blitz algorithm wasn't true to form. My advice is to avoid Blitz.

After reading this thread, I began watching more and more battles. Attackers 6 vs 1, 7 vs 2, 8 vs 3 lose very often. I began rolling dice regularly instead of blitz. There are WAY to many sixes being rolled by defenders, especially when rolling only one dice. The number distribution isn't close to what is being shown on the profile page. This really is a huge problem, and I hope smgstudio takes the problem seriously because its messing up games, especially in the first few rounds when troops are at a huge premium. Love the game, but the dice are off because defenders are rolling too many sixes.

This needs to get fixed. I just had 24 troops wiped out by 6 attackers. Then on the flip side I lost 12 people attacking 1. It happens more often than not. The odds just are not in the favor of the way this is set up. It seems to be an ongoing issue.

That explains why I notice people able to stream right through kills with higher probability and no random mishaps along the way sometimes. It should be truly random. catastrophe is a part of the game. And they still happen to me but I guess not as much and would be nice if a miracle can happen sometimes.

Posted by Ash over at their official Facebook page:
"When we first implemented the code we used random on each individual die. The problem was you got a lot of extreme cases with this. So now we use a probability matrix. There's still edge cases but it's MUCH less than what happens if each dice roll is totally random."
Well, now THAT explains it... SMG is NOT using a true RNG, but rather a "probability matrix" (i.e. the dice are rigged, plain and simple). Extreme cases or not, EACH and EVERY single die roll should be rolled as RANDOM, as this is how it would be with REAL-WORLD dice (and apparently, how it is coded in all of the other Risk and Risk-like games that I play... i.e. NO questionable dice rolls).

Its not the random number generator, but the algorithm that uses it. They seem to have developed thw blitz algorithm apart from summup up independent dice rolls. What they might have done is take the ratio of attacker to defender to weight the outcome. Thats a reasonable approach to save vompute time.

In my most recent game, I had 32 troops blitz 7. I lost all 32, they lost 0. Using an odds generator:
Long-Term Battle Simulation
Attacker Won: 100% of the time (357727 of 357737 trials)
Average Conquering Army Size: 25
Defender Won: 0% of the time (10 of 357737 trials)
Unfortunately for me, this seems to happen much more often than 10 out of 357000 times.
Like 3 out of 5 at least.

SMG - Please use the Merseinne Twister for a RNG.

The dice rolls are completely ridiculous. The amount of times I lose all my troops on a 10 v 2 battles is just plain annoying and in no way reflects the odds. The icing on the cake was just now when I had 23 v 5 and I lost all my troops and they only lost 2.
The game is just unplayable when it's like this.

I believe the dice has gotten better since this problem first started. I don't know if they changed anything or not. But I do feel like there is a higher chance of unexpected results than there used to be. I'm much more paranoid about taking on likely to win but not definite to win odds. A good example is 10 to 2. In the old algorithm you could still get burned with this but I'll watch games where several people in the same round get burned by this but then it doesn't happen for the rest of the game. I wish they'd just say what the algorithm is based on and I think everyone would be at ease. Everytime I roll all 1's I get annoyed because I do see it pretty often when i'm doing roll for roll.

I read through many of the previous posts and can provide a more thoughtful statement than what I posted last night.
The biggest issue objection to a more predictable algorithm seems to be the lack of luck. That is easily addressed with statistical concepts that have been in place to randomize clinical trials since the 1970's. The algorithm can work as true dice 'x' percent of time, and then other times work disproportionately.
For example, 90% of the time the algorithm would work as true dice odds, and randomly 10% of the time it could work disproportionately in favor of the underdog. This would allow luck, but preserve rewarding smarter strategy over luck/desperation (it is a strategic game).
Plus, true dice odds has a huge element of luck anyway (ask anyone who has played craps). The point is to avoid obscene outcomesmes like 10 losing to 2 on an attack that are nearly statistically impossible and punish the successful strategy to create such an imbalanced face off.

Not true dice rules - please update.

More proof that the dice rolling is screwed... it is almost statistically impossible to loose more troops than you have killed when your total win-to-loss ratio is almost 2:1. THAT IS, unless that you are rolling more 1's (or other low numbers) than any other number.
I have seen this on two different devices, as well as a buddy of mine telling me that it is EXACTLY the same on his.
And the funny thing is, I have NEVER seen anything of the sort (stats as mentioned above, AS WELL AS numerous questionable dice rolls) on ANY of the other Risk games that I play (on any other Android Risk or Risk like game, se8veral Risk dice rollers that I have installed on my tablet, Risk for Windows 95, Risk II for Windows 98, OR EVEN Risk for DOS - *CIRCA 1988*).
As a matter of fact, I've been so sick and tired of these problems (as well as the problems when attempting to join a multiplayer game) that just about ALL of the Risk playing that I have been doing for the past few weeks is either the REAL board game, or the old classic Risk for DOS.
Come on, SMG, my buddy and I PAID for it, FIX IT.

Yes I was playing with a friend and it was something like 15v4 and she lost all but 4 of her troops of this was an actual battle she would have like 12 left

## Steve Clements

HOW TO VOTE FOR THIS FEATURE?Tap the 'Do you like this idea?' below91 people like this idea