|
On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok.
Like I have eluded to before, the "critical strike generator" (I want to avoid using the word pseudorandom because it is actually random) prevents lucky streak of crits in a row. This clearly solves the problem with critical strike everyone is complaining about. The problem is how much this "critical strike generator" is doing - we don't know.
I would not be surprised if professional teams in Asia has figured it out already. It's not hard; just requires a lot of time to collect data.
As to how it works, let's look at two ways to generate 50% crit:
Method 1: you toss a coin; head = crit, tail = no-crit. Then this is entirely random and can get series of lucky crits.
Method 2: first hit is crit, second is non-crit, third is crit, fourth is non-crit, and so on. I don't want to get too technical here, but suppose you repeat this for 10 hours and forget your numbers, then the next hit sort of have a 50% chance to crit.
Obviously, the ACTUAL "critical strike generator" is probably somewhere in between Method 1 and Method 2. If crit becomes a problem, Riot can simply change their "critical strike generator" to behave more like Method 2 than Method 1, which solves the problem.
|
On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok.
But it doesn't matter! Whether or not any form of RNG effects a fight a binary distribution with a value p where p is a function of the psuedo random generator code and the fight circumstances which exists on the range of [0,1].
So long as p is not always zero. Which it cannot be. Then the limit in probability that the RNG will effect a fight as the number of fights increases is 1.
Which is to say its basically guaranteed that its going to effect fights. Its basically guaranteed its going to effect professional games. Given that we are pretty sure we have seen fights in professional games determined by crit we can be reasonably positive that not only has it already occurred and will continue to occur in the future that it is likely happening in ways we cannot as easily notice.
This.is.not.a.good.thing
On October 01 2015 08:21 Sufficiency wrote:Show nested quote +On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok. Like I have eluded to before, the "critical strike generator" (I want to avoid using the word pseudorandom because it is actually random) prevents lucky streak of crits in a row. This clearly solves the problem with critical strike everyone is complaining about. The problem is how much this "critical strike generator" is doing - we don't know. I would not be surprised if professional teams in Asia has figured it out already. It's not hard; just requires a lot of time to collect data. As to how it works, let's look at two ways to generate 50% crit: Method 1: you toss a coin; head = crit, tail = no-crit. Then this is entirely random and can get series of lucky crits. Method 2: first hit is crit, second is non-crit, third is crit, fourth is non-crit, and so on. I don't want to get too technical here, but suppose you repeat this for 10 hours and forget your numbers, then the next hit sort of have a 50% chance to crit. Obviously, the ACTUAL "critical strike generator" is probably somewhere in between Method 1 and Method 2. If crit becomes a problem, Riot can simply change their "critical strike generator" to behave more like Method 2 than Method 1, which solves the problem.
But i already showed you that even if its the second one then the problem of critical strike still exists! It exists 50% of the time in that scenario!
|
On October 01 2015 08:22 Goumindong wrote:Show nested quote +On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok. But it doesn't matter! Whether or not any form of RNG effects a fight a binary distribution with a value p where p is a function of the psuedo random generator code and the fight circumstances which exists on the range of [0,1]. So long as p is not always zero. Which it cannot be. Then the limit in probability that the RNG will effect a fight as the number of fights increases is 1. Which is to say its basically guaranteed that its going to effect fights. Its basically guaranteed its going to effect professional games. Given that we are pretty sure we have seen fights in professional games determined by crit we can be reasonably positive that not only has it already occurred and will continue to occur in the future that it is likely happening in ways we cannot as easily notice. This.is.not.a.good.thing
Really, it's not. You are talking about a possibility - one which Riot already has code for the "critical strike generator" to minimize. For all we know, it might be more likely for a meteor to strike Berlin and kill everyone in range than Cop critting 5 times in a row to totally ruin the game.
Also, like I have already said, Riot already has the code. They can actually just tune the code if they think it's a problem.
|
On October 01 2015 08:20 wei2coolman wrote:Show nested quote +On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok. that's pretty much my argument, there's enough auto attacks in the game at each common breakpoint for crit chance that the other strategics and skill variable play dwarves outliers in stacked crits.
problem with that is there is a big difference in an auto on a creep and an auto on a priority target
Just because rng doesn't entirely decide games doesn't make it a not garbage mechanic. If you think about how rarely there are even games in league for rng to only decide you wouldn't be surprised.
|
On October 01 2015 08:20 wei2coolman wrote:Show nested quote +On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok. that's pretty much my argument, there's enough auto attacks in the game at each common breakpoint for crit chance that the other strategics and skill variable play dwarves outliers in stacked crits.
Why are we only looking at the winner of entire games, though?
Even if it averages out enough over the course of a game, surely crit can ruin smaller experiences within the game (laning, fights, etc). If both players had equally shitty experiences and the end result is something that doesn't affect who won the game very much, then both players still had shitty experiences.
This can probably be reduced to "what is anti-fun according to riot", but I think it's silly to look at crit exclusively over long periods of time.
|
On October 01 2015 08:21 Sufficiency wrote:Show nested quote +On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok. Like I have eluded to before, the "critical strike generator" (I want to avoid using the word pseudorandom because it is actually random) prevents lucky streak of crits in a row. This clearly solves the problem with critical strike everyone is complaining about. The problem is how much this "critical strike generator" is doing - we don't know. I would not be surprised if professional teams in Asia has figured it out already. It's not hard; just requires a lot of time to collect data. As to how it works, let's look at two ways to generate 50% crit: Method 1: you toss a coin; head = crit, tail = no-crit. Then this is entirely random and can get series of lucky crits. Method 2: first hit is crit, second is non-crit, third is crit, fourth is non-crit, and so on. I don't want to get too technical here, but suppose you repeat this for 10 hours and forget your numbers, then the next hit sort of have a 50% chance to crit. Obviously, the ACTUAL "critical strike generator" is probably somewhere in between Method 1 and Method 2. If crit becomes a problem, Riot can simply change their "critical strike generator" to behave more like Method 2 than Method 1, which solves the problem.
It's most likely a Markov Chain Monte Carlo.
On October 01 2015 08:27 nafta wrote:Show nested quote +On October 01 2015 08:20 wei2coolman wrote:On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok. that's pretty much my argument, there's enough auto attacks in the game at each common breakpoint for crit chance that the other strategics and skill variable play dwarves outliers in stacked crits. problem with that is there is a big difference in an auto on a creep and an auto on a priority target Just because rng doesn't entirely decide games doesn't make it a not garbage mechanic. If you think about how rarely there are even games in league for rng to only decide you wouldn't be surprised.
I was talking about autos on priority target as well.
Also playing around known random variables is also a skill. Now whether or not you think this is a skill that should be rewarded is another thing altogether.
|
On October 01 2015 08:30 wei2coolman wrote:Show nested quote +On October 01 2015 08:21 Sufficiency wrote:On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok. Like I have eluded to before, the "critical strike generator" (I want to avoid using the word pseudorandom because it is actually random) prevents lucky streak of crits in a row. This clearly solves the problem with critical strike everyone is complaining about. The problem is how much this "critical strike generator" is doing - we don't know. I would not be surprised if professional teams in Asia has figured it out already. It's not hard; just requires a lot of time to collect data. As to how it works, let's look at two ways to generate 50% crit: Method 1: you toss a coin; head = crit, tail = no-crit. Then this is entirely random and can get series of lucky crits. Method 2: first hit is crit, second is non-crit, third is crit, fourth is non-crit, and so on. I don't want to get too technical here, but suppose you repeat this for 10 hours and forget your numbers, then the next hit sort of have a 50% chance to crit. Obviously, the ACTUAL "critical strike generator" is probably somewhere in between Method 1 and Method 2. If crit becomes a problem, Riot can simply change their "critical strike generator" to behave more like Method 2 than Method 1, which solves the problem. It's most likely a Markov Chain Monte Carlo.
You COULD call it that. But in the simplest scenario it's just a Markov Chain where the transition probabilities are calculated such that the stationary distribution is the actual critical strike chance. You also need a "mixing parameter", if you will, which you can adjust how close the generator is to Method 1 or 2. It's dead simple: if Riot is unhappy with critical strikes they can just tune the mixing parameter and no one will notice.
I guess it is possible that the actual generator is not just a simple Markov Chain and it's not completely memoryless. In which case it will probably be more like what you have said (it needs latent states), although I still prefer the phrase "Hidden Markov Model". If this is the case, the computation for stationary distribution will become a bit more complicated - I am not sure how to do it on top of my head.
|
Lol there is no skill in playing around crit %. Anyone competent can look at it and decide if it is a good risk to take.
Honestly riot have said a lot of things so not really sure if I believe such a thing exists. They also added being able to flash inbetween being ccd in s2(as in there were patch notes that said they added it intentionally) and later said that it was a bug.
|
On October 01 2015 08:41 nafta wrote:
Lol there is no skill in playing around crit %. Anyone competent can look at it and decide if it is a good risk to take.
Honestly riot have said a lot of things so not really sure if I believe such a thing exists. They also added being able to flash inbetween being ccd in s2(as in there were patch notes that said they added it intentionally) and later said that it was a bug.
Sure, Blizzard implemented something similar for Warcraft 3, but they implemented it really badly and you can only get critical strike chance in 5% increments AND that critical strike chance is bugged (http://dota2.gamepedia.com/Pseudo-random_distribution).
But I've seen no evidence, not even circumstantial evidence (like the 5% increment thing) that Riot screwed it up. So unless someone has experimented and can demonstrate otherwise, I am calling your statement a conspiracy theory at best.
|
Which is why I said I believe and not stated it as a fact. I mean I still have ptsd from blademasters maybe that why I get so salty.
|
On October 01 2015 08:41 nafta wrote:
Lol there is no skill in playing around crit %. Anyone competent can look at it and decide if it is a good risk to take.
Honestly riot have said a lot of things so not really sure if I believe such a thing exists. They also added being able to flash inbetween being ccd in s2(as in there were patch notes that said they added it intentionally) and later said that it was a bug.
Are you saying poker takes no skill?
|
Poker % aren't so simple and playing actual poker has a lot more shit to do than just calculating the %.
Also poker is a lot of hands not so few.
|
On October 01 2015 08:56 nafta wrote:
Poker % aren't so simple and playing actual poker has a lot more shit to do than just calculating the %.
Also poker is a lot of hands not so few.
So I guess the issue just comes down to a matter of perception. After all, LoL rarely comes down to a 1v1 brawl and whoever crits first wins the game - I am not saying this NEVER happens, but that it RARELY happens.
Kind of similar to Poker if you ask me!
|
On October 01 2015 09:00 Sufficiency wrote:Show nested quote +On October 01 2015 08:56 nafta wrote:
Poker % aren't so simple and playing actual poker has a lot more shit to do than just calculating the %.
Also poker is a lot of hands not so few. So I guess the issue just comes down to a matter of perception. After all, LoL rarely comes down to a 1v1 brawl and whoever crits first wins the game - I am not saying this NEVER happens, but that it RARELY happens. Kind of similar to Poker if you ask me!
On October 01 2015 08:27 nafta wrote:Show nested quote +On October 01 2015 08:20 wei2coolman wrote:On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok. that's pretty much my argument, there's enough auto attacks in the game at each common breakpoint for crit chance that the other strategics and skill variable play dwarves outliers in stacked crits. Just because rng doesn't entirely decide games doesn't make it a not garbage mechanic. If you think about how rarely there are even games in league for rng to only decide you wouldn't be surprised.
Also riot removed dodge because they said they don't want rng in a competetive game and next season added 2% crit in the offensive tree mastery. Just saying.
Not gonna go into if crit is fun since that is subjective but let's think about the pro scene for a second. How many tournaments do the teams actually play? Regular splits don't actually matter at all as long as you get top 6. Then you get to play 2 playoffs and 1 worlds or 1 invitational IF you are one of the really top tier teams(and you can potentially go to some iems if you are popular). Don't you think even one of those games being decided by a dice roll is too much?
|
On October 01 2015 09:03 nafta wrote:Show nested quote +On October 01 2015 09:00 Sufficiency wrote:On October 01 2015 08:56 nafta wrote:
Poker % aren't so simple and playing actual poker has a lot more shit to do than just calculating the %.
Also poker is a lot of hands not so few. So I guess the issue just comes down to a matter of perception. After all, LoL rarely comes down to a 1v1 brawl and whoever crits first wins the game - I am not saying this NEVER happens, but that it RARELY happens. Kind of similar to Poker if you ask me! Show nested quote +On October 01 2015 08:27 nafta wrote:On October 01 2015 08:20 wei2coolman wrote:On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok. that's pretty much my argument, there's enough auto attacks in the game at each common breakpoint for crit chance that the other strategics and skill variable play dwarves outliers in stacked crits. Just because rng doesn't entirely decide games doesn't make it a not garbage mechanic. If you think about how rarely there are even games in league for rng to only decide you wouldn't be surprised. Also riot removed dodge because they said they don't want rng in a competetive game and next season added 2% crit in the offensive tree mastery. Just saying. Not gonna go into if crit is fun since that is subjective but let's think about the pro scene for a second. How many tournaments do the teams actually play? Regular splits don't actually matter at all as long as you get top 6. Then you get to play 2 playoffs and 1 worlds or 1 invitational IF you are one of the really top tier teams(and you can potentially go to some iems if you are popular). Don't you think even one of those games being decided by a dice roll is too much?
I don't remember the patch notes regarding removal dodge that clearly, but if I remember correctly, the biggest problem was Jax was able to have an insane amount of dodge chance with the ninja tabis + dodge runes, and his ability to have dodge chance.
|
On October 01 2015 08:21 Sufficiency wrote:Show nested quote +On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok. Like I have eluded to before, the "critical strike generator" (I want to avoid using the word pseudorandom because it is actually random) prevents lucky streak of crits in a row. This clearly solves the problem with critical strike everyone is complaining about. The problem is how much this "critical strike generator" is doing - we don't know. I would not be surprised if professional teams in Asia has figured it out already. It's not hard; just requires a lot of time to collect data. As to how it works, let's look at two ways to generate 50% crit: Method 1: you toss a coin; head = crit, tail = no-crit. Then this is entirely random and can get series of lucky crits. Method 2: first hit is crit, second is non-crit, third is crit, fourth is non-crit, and so on. I don't want to get too technical here, but suppose you repeat this for 10 hours and forget your numbers, then the next hit sort of have a 50% chance to crit. Obviously, the ACTUAL "critical strike generator" is probably somewhere in between Method 1 and Method 2. If crit becomes a problem, Riot can simply change their "critical strike generator" to behave more like Method 2 than Method 1, which solves the problem.
I was talking about the absolute specifics not broadly how it works, yes.
Also, for people thinking they don't use psuedorng, with 10 seconds of googling I saw Xypherous stating specifically they use a bounded geometric model for crit. It's an old post, but I don't see any reason to assume they've changed it other than, well, conspiracy theories or nonsense anti-Riot bias.
Crit as an RNG element is not about the number of games played; it's about the number of deciding AAs. You don't judge poker by tournaments, you judge it by significant hands. The majority of high-skill players are in primarily cash games.
|
On October 01 2015 09:19 wei2coolman wrote:Show nested quote +On October 01 2015 09:03 nafta wrote:On October 01 2015 09:00 Sufficiency wrote:On October 01 2015 08:56 nafta wrote:
Poker % aren't so simple and playing actual poker has a lot more shit to do than just calculating the %.
Also poker is a lot of hands not so few. So I guess the issue just comes down to a matter of perception. After all, LoL rarely comes down to a 1v1 brawl and whoever crits first wins the game - I am not saying this NEVER happens, but that it RARELY happens. Kind of similar to Poker if you ask me! On October 01 2015 08:27 nafta wrote:On October 01 2015 08:20 wei2coolman wrote:On October 01 2015 08:12 Caiada wrote:
Nobody knows exactly how their psuedorandom generator works or how actual crit% works on a per-fight basis, so the technical discussions are pointless. Analysis is the only thing that's going to answer shit here, and until somebody argues or properly shows that an individual fight is going to completely turn a game (outside of soloq; fuck soloq), crit looks a-ok. that's pretty much my argument, there's enough auto attacks in the game at each common breakpoint for crit chance that the other strategics and skill variable play dwarves outliers in stacked crits. Just because rng doesn't entirely decide games doesn't make it a not garbage mechanic. If you think about how rarely there are even games in league for rng to only decide you wouldn't be surprised. Also riot removed dodge because they said they don't want rng in a competetive game and next season added 2% crit in the offensive tree mastery. Just saying. Not gonna go into if crit is fun since that is subjective but let's think about the pro scene for a second. How many tournaments do the teams actually play? Regular splits don't actually matter at all as long as you get top 6. Then you get to play 2 playoffs and 1 worlds or 1 invitational IF you are one of the really top tier teams(and you can potentially go to some iems if you are popular). Don't you think even one of those games being decided by a dice roll is too much? I don't remember the patch notes regarding removal dodge that clearly, but if I remember correctly, the biggest problem was Jax was able to have an insane amount of dodge chance with the ninja tabis + dodge runes, and his ability to have dodge chance.
http://clgaming.net/redtracker/914-so-dodge-was-removed-because-riot-was-trying-to-move-away-from-rng-reliance
"crit is cool"
too lazy to search for actual patch notes but I remember them quite well
|
For the people defending Crit, do you actually think having Crit (pseudo random or otherwise) is preferable to just having "Increases auto-attack damage by X%" where X is the Crit chance of the item? Would the game be better or worse if Crit was changed in that way?
Assuming you changed procs that work off Crit to scale correctly (IE, Frenzy mastery, crit damage runes) I don't see how it could be anything but an improvement. So why defend Crit?
|
Whether the crit averages out or not is irrelevant because not every attack is equally important. That auto on a minion? irrelevant. That jinx auto which could crit the guy who just flashed out of the teamfight? Game deciding.
On October 01 2015 10:01 Amarok wrote:
For the people defending Crit, do you actually think having Crit (pseudo random or otherwise) is preferable to just having "Increases auto-attack damage by X%" where X is the Crit chance of the item? Would the game be better or worse if Crit was changed in that way?
Assuming you changed procs that work off Crit to scale correctly (IE, Frenzy mastery, crit damage runes) I don't see how it could be anything but an improvement. So why defend Crit?
Because they enjoy being contrarian
|
On October 01 2015 10:01 Amarok wrote:
For the people defending Crit, do you actually think having Crit (pseudo random or otherwise) is preferable to just having "Increases auto-attack damage by X%" where X is the Crit chance of the item? Would the game be better or worse if Crit was changed in that way?
Assuming you changed procs that work off Crit to scale correctly (IE, Frenzy mastery, crit damage runes) I don't see how it could be anything but an improvement. So why defend Crit?
Only downside I can see is you'd have to buff Ashe.
|
|
|
|
|
|
EPT
