Start a new topic
Implemented

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

Is it really that bad for the server to just use the PRNG for each roll, even in blitz? Isn't that the most accurate approach?
Guys, its not the random number generator itself, but the Blitz algorithm that uses the numbers. SMG has said it uses a "transition matrix" algorithm. This algorithm short cuts the number of manual rolls required to finish an opponent. The TM embeds probabilities for every attacker-defender combination and the random number between 0-1 is rolles to see what "matrix element" is tested. Then another random number is thrown to see if the test is good or not. So at least two random numbers are used in the algorithm, one to decide the interaction probability and one to test the win/lose criteria for that particular interaction. This is called a "Markov Chain". I think SMG doesnt want to tax their servers and use tbis simplistic yet inaccurate or nonrealistic approach to save compute time. However, the intelligent programmer could devise a scheme to utilize parallel computing in Blitz mode such that more transition probabilities are determined and tested simultaneously. This is the way they do it in real life with chemical kinetics, radiation transport, and ocean/environmental science. SMGs opted for a quick and simplistic approach to the algorithm makes the game not enjoyable in the long run. Why dont they fix the algorithm? The random number stride is long enough, but tbe way they use the numbers is highly suspect. I use the Merseinne Twister and even though we use the same generator, my algorithms that use the numbers alway sample from a true probability distribution.

2 people like this
Before SMG changed up the dice weren't they random roll for roll (Or acceptably random anyway)? I never understood why they were changed. Now my accounts have their percentages skewed to the higher numbers how does that reflect real life dice? Before they were all 16's. Of course I remember people clamoring for change back then but I think it used to be a much more satisfactory algorithm. On spectator mode it is a good way to spy on suspicious players. I took down the #2 player Wednesday by joining his game and reporting him (He played two of three humans and had a bot). His secondary was ranked 118 so he was getting points from beating that every game. As far as I can tell SMG deducts massive amounts of ranking points from cheaters so it is almost like banning them. Besides if they keep their old name we can remember them so that is nice. I am confident that the rest of the top ten and many other top accounts are cheaters as well but they are inactive or smarter. Still if I see any top accounts playing I think I will crash their party.

2 people like this
(continuation of "Moral of the story") ,,,as Risk is PREDOMINATELY a DICE game, hence, the dice need to function properly (like they do in all of the other PC/Android Risk and Risk-like games that I play).
Side Note: The thing about pseudo random number generators... No, they are not 100% random, but they are damned close. The same can be applied to Blitz mode. Really weird how all of the other PC (DOS, as well as Windows) versions of Risk, as well as the other Risk clone and dice rolling apps that I have played/used are WAY MORE random than SMG's Risk. The DOS and Windows 3.x versions of Risk have "blitz" types of options, as do versions for Windows 95 and Windows 98, not to mention clones such as Border Siege. IN ALL OF THESE games the dice are dead on. HARDLY EVER are there "questionable rolls", single rolling and "blitz" type rolling alike. I ALSO never see "questionable rolls" in the dice rolling app "Dynamic Dice", or several JavaScript based rolls that can be found around the Internet. Morale of the story, THE DICE NEED TO BE FIXED.

1 person likes this
I have mostly seen the advantage to the attacker. Sometimes it doesn't work out that way. Sometimes it's catastrophic. Only a few times have I been burned with catastrophic Blitz rolls several times in one game. It can happen with real dice rolls but shouldn't happen often. It hasn't happened often for me. I feel mostly a clean run through most of my games. But I will say that I do remember the ones where I get burned hard.

I agree they should strive to make a truly random, roll for roll. I would prefer that. But I don't think it kills the game if everyone has the same type of rolls.

More important to me is the anti-cheating. I am so glad they finally implemented spectator mode. It allows me to stick around and see if what looked like cheating was in fact cheating so I can report them. But all they do is punish them by removing the score gained. They need to be more harsh. They need to delete accounts when caught cheating. They also need to ditch the idea that players should be able to play with friends in games mixed with random people. Friend on Risk? Ok. Friends at your house? Not ok. It's usually the same guy using many devices. And even if not, that puts bias into a ranked game. Make games not ranked / work on a ladder system with strict rules? ok no problem then.

I havent really had issues with connections that much so I can't really complain about it.

 

No dude, of the millions playing, hardly anyone will come here to complain - they will just get sick of the app and uninstall lol... I dread to think of how much money SMG has lost over this issue!


People are interested in Risk because they've played the board game - so they know how real dice behave... and understand that the the attacker has a slight advantage over the defender when its 3v2 dice etc...


Unfortunately I have yet to see that advantage displayed in most of the games I've played - what I do see however, are a constant stream of virtually impossible rolls... which lose vastly more troops then were defeated! 


Sure there will always be some little kids who will rage-quit if they get one bad roll lol - but it's kinda silly to think grown adults, who've been playing the game for years... are just complaining for no reason lol!


This game has the potential to make millions for SMG - but they are sabotaging it's success by not fixing three major issues, namely:

1) Maintaining 100% Reliable Connection

2) Integrating Anti-Cheating Measures

3) Developing Truly Random Dice 


2 people like this
I agree it's not the same thing. Nothing computerized is truly random anyway. It has to have some numerical basis since there's no gravity pulling the dice down. Hell, even that's not really random to roll dice. Random.org uses a neat technique. they use micro weather data to produce their random results.That's probably as random as you'll get with a computer. As for the dice in the real game, depends on the player rolling them.

Also, before they implemented the matrix, it really burned up processing power on your device. I recall some 50+ battles taking a whole half a second to process. I do personally prefer making it roll out each one, one by one but I get why they did it.

I also never would have expected to get a lousy roll like that on the board game but that's because i've played less than 100 times in my life. But already in 2 years playing in this app, I've easily got 1500-2000 games. Yet, I cannot recall a single time when it was that devastating for me. But I'm sure out of the millions playing, there is bound to be someone with such results who will come on here to complain when it happens ;)

 

Yeah, it's possible, one in a million. I've played enough real-world Risk with real-world dice to know the statistics... never seen it happen though. That being said, the dice in the app needs to be ONE HUNDRED PERCENT random, not using a "probability matrix". P.S. I'm sure thankful that the 49 year old dice in my 1968 edition real-world board game doesn't use this "probability matrix".
It is possible to happen in real odds. A probability matrix might feel unfair because there is a 1 out of xxxxxxxxxx chance that one will get selected. But it's not that much different from rolling each one out. For the sake of not killing your device it's not a terrible trade. But yes, when I have a bad roll I also want to blame the matrix too haha

 

28 vs 2, I lost 27, the opponent lost NONE. Ahhhh... the "probability matrix", GARBAGE!

I have been collecting data on battle outcomes. I agree with SMG (actually can't disagree with their assertion) that dice roles are random. What is specifically not random are battle outcomes. For example, I have observed a 7 attacking 2 battle lose 4 of 19 times (21%).  I should have seen that happen *maybe* 1 time. This happens too frequently and across too many 'dice paring' combinations to be random. 


In short, the dice rolls may be (pseudo)random, but the battle outcomes are EMPIRICALLY not. 


(MATLAB available on request)


1 person likes this
"Implemented" yeah friccin rite!! You've been exposed for being cheap SMG. Apparently the matrix system they use is cheaper than having true dice rolls, that explains it!
Its not the dice rolls thats the problem. The problem is tbe Blitz Transition Matrix probabilities that utilize tge spe ific random number to test against the predefined probability for each transition. SMG has defined the probabilitu for each transition.For example a 10 vs 4 transition probability has been defined as x. If the uniform ra dom number is less tgan x, then its a fail. The issue is their choice of the value x. They need to have a more accurate Transition Matrix.

1 person likes this
I was under the impression that the dice rolls were calculated locally via the device. I believe this because it's possible to lose internet connection, continue rolling and then suddenly find yourself backed up to a previous point where you had already rolled through and have to do it again when it reconnects.

 


1 person likes this
Login or Signup to post a comment