You’ve no idea what you’re talking about.
UK, saying that without some evidence or argument is unhelpful and rude. Please explain things as you understand them if you have something substantive to contribute.
2 Likes
When I saw the title I was hoping to get an explanation on why a 25% rez chance happens 70% for the AI…
“It’s all in your mind”.
In fact Salty’s post is quite simple and perfectly correct. It’s also very careful not to relate at all to the real heart of the matter, viz. “is RNG’s randomness actually random?”
In fact is a further “it’s all in yout mind”, Or “there’s no spoon”, if you prefer. But there is a spoon, and it’s bended.
@Sheba
Pets.
I had two, in two different occasions, and they were completely different and quite interesting.
1 Like
I guess another interesting thing to point out is that “probability” and “randomness” are very related to each other.
Randomness is very vague. It just means “something you can’t predict”.
Probability helps us understand different kinds of randomness. Most people consider a “truly random” system to be one where any event has equal odds of happening independent of previous events. Probability generally calls this a “normal distribution”. But there are ways to make a data set look like a normal distribution while still having predictable patterns! So the real most important question to ask in probability is, “What question am I actually asking?”
People love to trot out that coins can be imbalanced, dice can be unfair, and that computer RNGs are only pseudo-random. All of these things are true and can introduce bias to calculations. Unless we’re talking about very natural phenomenon like “the decay of radioactive isotopes” practially all sources of randomness at our disposal are influenced by Physics or other rules that we can manipulate.
I can make a d6 that’s so heavy on the 6 side, even if you carefully place it with the 3 side up it will thud its way over to a 1. Anyone sane would agree this is not random, since the result is 100% predictable.
Now let’s say I dial it back and weight the d6 so it’s only about 50% likely to land with the 1 side up. That is dramatically different than a normal d6! Normally you expect ~16.7% odds of any given side. Now the odds are 50% 6, and 10% for each of the other 5 sides. But… now it’s random again!
It’s definitely biased. Betting on 1 will effectively be a coin toss. If you were betting, you’d be wisest to bet on that. But you can still roll a 3 and lose your bet. And nothing you can do short of adding more weight to the d6 will make it easier to say what the next roll will be.
So it turns out even a random system with a bias is still a random system. It just might not be the system that we want.
There’s another problem people have with probability and it’s not so easy to illustrate.
Any non-zero probability event can happen, no matter how unlikely.
So when flipping coins, it’s very, very unlikely you’ll get 10 heads in a row. But it’s not impossible. If you were betting on tails, you’ll be very angry about it. But if, for some reason, you start flipping coins right now, I’d almost be willing to wager money you’d see 10 heads in a row by the time you reach 5,000 tosses. Probably sooner, I’m too lazy to work the math.
That applies to this video game and does NOT imply an RNG bias on its own. I don’t know the real odds but I bet 1,000 gem keys are very likely to get you at least one mythic. There are some poor saps out there who have opened 1,000 gem chests and got 0 mythics. Guaranteed. If it wasn’t true, the system would either “not be random” or would be subject to a bias.
This is how gambling works. It’s addictive because our minds don’t understand probability. This is why the devs don’t like disclosing the drop rates: the more users are educated about probability the more they understand it’s never true to say, “Oh, I’ve opened X chests so far, I’m really close!” No number of chests can be opened to make you “closer” to any given outcome. It’s all probability.
RE: “Don’t they have to disclose odds?”
We had a thread about it recently due to the Apple changes. It seems both players and the devs of GoW seem to think they don’t qualify under the policy because you don’t “pay money for a random chance”, you “pay money for a key that represents a random chance”.
I don’t like this stance. Apple is making moves to try and have developers voluntarily disclose odds. One can read it as a canary: “Hey, a lot of legislators are starting to think of classifying F2P games as gambling, the more you do today to look honest the less traction they’ll have.” Every dev that tries to find a cute loophole risks some pretty dramatic consequences for all game developers down the road.
Not to mention it’s well-documented most of the fallacies that fuel addictive behaviors rely on the odds being obscured.
Whether or not absolutely required by Apple’s internal rules, China’s regulations, or fear of whatever regulations may come down the pike in any other jurisdiction, I call on Infinity Plus 2 and 505 Games to voluntarily disclose the odds, as it is the right thing to do.
3 Likes
While for the most part this is true, there is however a calculation that is easily used to find the probability that something that you are after will have dropped in a number of attempts…
1 - ( ( 1 - x ) ^ y )
so, in your example of the 1 in 100 chance to find a unicorn, we have
1-((1-.01)^Y)
Now we’ll assume we ran this 100 times…
The actual % chance that you would have found at least one unicorn in that 100 times is 36.60%
While talk of theory and algorithms is nice, let’s take a look at the practical for a moment. I have not received a mythical from a chest in over 6 months now (Queen Aurora). Regardless of methodology, I am going to get bored at some point and leave the game. Opening chest after chest and getting nothing new or exciting for months on end really does nothing for me. Mythics should be rare, but if they are too rare, then what fun is it for some?
1 Like
On a related note but completely pointless to the topic, a coin toss isn’t exactly a 50% chance of getting heads or tails, it’s actually a few points of a percent higher chance of getting a head due to the design of a coin (it’s not perfectly flat on both sides).
There’s some useless trivia for you.
Is that based on an american quarter, or some other coin?
And there is probably a non-zero probability that the coin could come to rest balancing on-edge.
Those ‘useless troops’ are useful to some players. Hey, I like mammoths! 
I would just like to add that Salty made this thread on her own accord (after discussion, of course, but the motivation was her own). I appreciate your concern about the thread’s accurateness but in reality this is just one person’s attempt to bring a little light on the situation. It is driven by her own desire to be bring some understanding on why things are the way they are in the game. You can see by the angle of the post and the fact that it’s not Sirrian posting it (which usually follows with an explicit warning of “we’re getting into maths here!..”) that the topic was not going to be addressed in a complicated manner. That was not the point of the topic. If this is a concern that you have, I suppose perhaps going about it in a more polite way wouldn’t hurt.
4 Likes
1/2 to get heads but What are the probability to get 100 x heads in a row? 
You would need billions… of dollars.
What’s funny is I discussed this with a buddy as he was also talking probability this morning.
As I wrote the dang analogy I was like, “You know, someone’s going to ‘well, actually’ me and point out coins can land on their edge. People hate abstraction”.
I put forth you’ve left out a ton of other possibilities.
- A bird could steal the coin mid-air.
- You could start flipping the coin moments before an explosion that kills you. This will invalidate the result as the coin did not naturally fall.
- The coin could fall into a storm drain and you can’t observe the result.
- You could spontaneously combust.
- A stray bullet could hit the coin and render it unreadable.
If we have to account for all of the things that COULD happen, we can argue it’s a wonder that any coin ever falls with heads or tails clearly visible.
Or: we can adopt a methodology of coin flipping that eliminates it entirely. Standard procedure is:
- Flip into the air.
- Catch in one hand.
- Make sure, by feel, the flat side of the coin is down.
- Optionally slap the coin-hand onto the back of the not-coin-hand and reveal.
This leaves the coin flattest side down in all scenarios.
1 Like
Honestly I’d say different coins have slightly different chances. It’s all very minute though.
Wellactually is my middle name.
I have no doubt that it is very minute. I was just curious about your source since you said ‘heads’ had a slightly higher probability. If that is true, it would depend on the particular coin. For example, a Canadian quarter might be slightly more weighted to tails or something…
Hey, they dont get you wins and enough resources /shrug
This was a very simple explanation as to how probability works in relation to drop rates.
We are not going to be getting into seeds and timestamps. We share a lot with you all in regards to our game, going as far as to run tests, etc. In fact, we share a lot more than most developers do, and always endeavour to do so.
However, this conversation is going nowhere. I only wanted to shed some light on why some players receive better drops than others. Luck and chance.
(PS, being a social media and community manager doesn’t mean I’m clueless when it comes to math. I’m a fully rounded human with a number of surprising skills.)
13 Likes
For me, the problem is not so much the lack of independence ( gambler’s fallacy) which is bothering me. Instead it is the observed frequency of events does not seem to be consistent with what is stated in the cards.
For example the Dragon Soul third trait gives it a 25% chance to resurrect after death. A-priori, the probability that a dragon soul will resurrect 3x in a match is very low. (approx 1.5x out of 100 matches played). This has however been observed to happen much more often than that (e.g. 5 to 10 out of 100 matches). If you ran a statistical hypothesis test, this leads you to not rejecting the hypothesis that the probability of resurrection is > 25%.
It is therefore not so much a bad pseudo random number generator and predictable outcomes but the frequency of outcomes which is causing a disconnect.
The drop rates for mythics has not been published so that there is no way to run a hypothesis test. It is however possible to run hypothesis test for other events (e.g. TDS prob to resurrect, Fizzbang prob to explode, Kraken prob to devour) whre the p is known. Not sure what is causing this but it might be worthwhile auditing the code of some of these cards to determine if the correct probabilities are being implemented.
1 Like
Ok,so in a nutshell…
A gambler has 99 ponies and figures that if the drop rate holds true then a small number of additional tries will probably produce a unicorn. His mistake however is failing to take into account that the sample size is larger than he expected, the lucky beast next to him just got three unicorns in five tries, and if the drop rate holds true there are a whole lot more disenchanted ponies in his future.
The other takeaway seems to be that the chance of the mythic drop rate being 1-in-100 is akin to the chance of me finding a real live unicorn at the end of the next rainbow I see.
This is of course, a simplified hypothetical that ignores the well-known fact that there are pots of gold at the end of rainbows and not unicorns.
“It must be indicative of something, besides the redistribution of wealth” -Guildenstern, Rosencrantz and Guildenstern Are Dead