Why has Riot removed a lot of RNG, except for crit?

SexyTaylorSwift·Dec 25, 2014·44 votes·8,218 views

Not a discussion on whether or not crit is OP (because it obviously is and only trolling adcs would argue otherwise), but rather a discussion on why they chose to leave this in the game when they have removed other random things?

Some things were: -Biggest one was dodge chance of course. The removal of dodge removed a huge RNG factor in the game and made it more stable and reliable. -Madreds was another one; removing the chance of a proc and putting a flat, consistent amount of extra damage on it was more reliable in the jungle for consistent clear speeds as well as not accidently stealing blue buff because you got a proc while giving it to your mid laner. -Phage was changed so there was no chance of slow, instead it gave a movespeed buff to the user on a consistent basis. -Frozen Mallet was given a consistent, non-RNG related slow. -Master Yi chance of extra 400 damage on Q to minions was changed to a consistent damage to minions.

125 Comments

Pinker 10Dec 25, 201410 votes

Because removing crit is a lot more work than removing dodge

They had to rework a single skill on jax one on fiora and remove some runes

Crit is like... everywhere.

If you seriously expect reworking everything that uses crit to use something else to take less than a couple years...

Sightless66Dec 25, 201415 votes

Just make it non-random. 1% crit = 1% increased autoattack damage vs non-buildings. Apply the same to critting champions and items. Gives the exact same dps as crit does now. Game balance remains the same or very similar to what it is now. The only champion who would need more than cursory modifications is Trynd, because he has a non-damage effect on critting (getting more fury). Other than him, easy fix.

There were multiple items with dodge and multiple champions relied on them as a core part of their builds. Removing it was a much bigger impact than derandomizing crit would be.

Pinker 10Dec 25, 20147 votes

Removing dodge had practically zero impact on the game. New ninja tabi's passive works out to generally the same effect as it's old one, Jax still evades everything, the only thing it actually really changed was the removal of runes and fiora's parry. 90% of tanks still built merc treads and still do.

Crit is worked into multiple champions in different ways, Adc's rely on it, and giving a straight up constant boost to champion attack damage is NOT a solution. Adc's are strong enough as it is, and now you want to buff every single one of them so they crit EVERY attack?

Solutions have been offered in a different thread, and indeed many of them are good solutions. However, crit is so deeply ingrained in not just the game but the minds of the players that removing it is a ridiculous amount of work. Morello commented on crit close to a year ago, and has said multiple times it's not good and they're working on it.

A year ago.

And they're still taking suggestions from the community on how to replace it.

This isn't a walk in the park, and it's not gonna happen over night.

To remove crit without affecting balance in any way is probably not even possible, let alone easy.

Sightless66Dec 25, 20148 votes

Adc's are strong enough as it is, and now you want to buff every single one of them so they crit EVERY attack?

It's not a buff. It's a normalization. DPS would be exactly the same. The only thing that would lower is variability, and variability screws you just as often as it helps. To make this clear: say that I had 100 damage and 50% crit chance. Over 2 attacks, I would expect to crit once for 200 damage, meaning that I would do 100+200=300 damage. However, I could also crit twice or never, giving me 400 or 200 damage. With a %increase system, 100 damage and 50% crit would give me 150 damage for autoattacks, so I'd always do 300 damage

So, it's not critting every attack. It's getting a % of a crit every attack, and never getting a full crit. Mathematically, this would lead to the exact same expected dps.

What do you think specifically would be a balance problem? I've thought about this, and I don't see any area where crit derandomization would be a major problem, given that dps would always be exactly the same.

Solutions have been offered in a different thread, and indeed many of them are good solutions.

Most solutions offer new mechanics. This one doesn't. It's just the exact same damage without any randomization, leading to the same dps. It takes much less effort than putting in a totally new mechanic.

Morello commented on crit close to a year ago, and has said multiple times it's not good and they're working on it.

They talked about it 6-8 months ago. The only major problem they had with removing crit was that they felt it created "memorable moments". Balance concerns were not brought up as a major obstacle. That goes along with that "the minds of the players" thing you brought up earlier. Derandomization is meant to preserve the same gameplay and balance, not necessarily the same feel. Critting once every 10 hits for double damage can be seen as more fun than hitting for 110% damage every time. I personally think the frustrating aspects of crit outweigh this though, which is why I still think a derandomized system is preferable.

Pinker 10Dec 25, 20144 votes

{quoted}

It's not a buff. It's a normalization. DPS would be exactly the same. The only thing that would lower is variability, and variability screws you just as often as it helps. To make this clear: say that I had 100 damage and 50% crit chance. Over 2 attacks, I would expect to crit once for 200 damage, meaning that I would do 100+200=300 damage. However, I could also crit twice or never, giving me 400 or 200 damage. With a %increase system, 100 damage and 50% crit would give me 150 damage for autoattacks, so I'd always do 300 damage

So, it's not critting every attack. It's getting a % of a crit every attack, and never getting a full crit. Mathematically, this would lead to the exact same expected dps.

What do you think specifically would be a balance problem? I've thought about this, and I don't see any area where crit derandomization would be a major problem, given that dps would always be exactly the same.

They talked about it 6-8 months ago. The only major problem they had with removing crit was that they felt it created "memorable moments". Balance concerns were not brought up as a major obstacle.

You're using math. To compare RNG to non-RNG.

Are you serious? Mathematically normalizing the effects of removing RNG is great, but RNG and math don't belong in the same sentence. It has as much probability of being a class wide buff or a class wide nerf as it does normalizing.

You're also using a sample size of two.

Normal attacks @ 100

40% crit chance

Sample size of 1000 hits

RNG is RNG

You can crit anywhere from 0 times, to 1000 times, depending entirely on how lucky you are. You aren't more likely to crit for 400 than you are for 1000. Each individual attack has a 40% chance of critting. If you flip a coin 600 times, and get heads every time, are you gauranteed heads on the 601st? No, it's still 50%. It's great to say oh, just say that crit adds 1% damage, and theoretically, damage ouput will be the same. If crit wasn't already random, this would work. It's not going to work like that. You can't normalize from Random.

The recent thread with community suggested ideas has morello considering a couple of the more realistic ones.

Sightless66Dec 25, 20147 votes

Mathematically normalizing the effects of removing RNG is great, but RNG and math don't belong in the same sentence.

Why? We can use math to predict the damage dealt by any number of attacks with any % crit chance with any AD. RNG is just probability, and probability is very simple math. It's easy to make an expected value, and we can easily determine variance ranges from that expected value.

We play a game based around numbers. Using math to compare numbers isn't out of line I think.

You're also using a sample size of two.

So? It was just a demonstration of the principle. Are you saying that you doubt that expected DPS will always be the same because I used a low sample size? I can prove this with any sample size, any crit chance, and any AD value you choose.

It has as much probability of being a class wide buff or a class wide nerf as it does normalizing.

It's 50% because there are two possibilities... that's not how probability works. If you want to say there would be a class-wide nerf or buff, then you have to explain why. Please do, by the way. Tell me what you think will change. I am willing to discuss this change in relation to any champion, item, role, etc. I do not think that there will be any significant changes, and I am I very willing to discuss that.

You can crit anywhere from 0 times, to 1000 times, depending entirely on how lucky you are.

Two things: First, so what? What effect does this randomness have on balance? I submit that has virtually none. People play based on the expected values, and mathematically, the points of major variation from the expected values are by far the exceptions. It's a rare outcome. Additionaly, no one plays around the extremely small possibilities. I don't start a fight with an IE knowing I can only win if I crit 8 times in a row. Likewise, I don't avoid a fight because I might lose it if my 50% crit chance build doesn't crit at all. People always play to the expected value, and the expected value is by far the most common outcome. What champion or tactic is based on variance? I can't think of a single one (expect maybe very arguably Trynd's laning). Why anyone base their tactics on RNG? It has just as much potential to screw them as help them. So, my question is, who do you think would be affected in a way that alters their balance state?

Second... no you can't. Not in League at least. League doesn't actually use RNG. It uses pRNG (psuedo RNG). This is a system used to decrease variability in random number generation. The way it works is by increasing the chance of an event every time the event does not happen. For League, this means that if I have 50% crit chance, my first hit may actually have 30% crit chance, then my second will have 40%, then 50%, and so on (the actual algorithm is a lot more complex, but you get the idea). Getting a crit lowers it back to the initial chance. Multiple crits in a row lower the chance even further (in some pRNG systems) After a point, the odds of not getting a crit or getting a streak are mathematically insignificiant. Your always crit or no crit event is so improbable that it probably hasn't happened even once in the entire time the game has been out. So, we aren't even shifting from true RNG to a normalized system, so we aren't even getting as big a variance decrease as it would initially appear.

Pinker 10Dec 25, 20141 votes

Why? We can use math to predict the damage dealt by any number of attacks with any % crit chance. RNG is just probability. If I have x chance of an event, and y

That's not finished. Probability is theoretical for a reason, just because something SHOULD happen by all accounts, doesn't mean it will. You can normalize randomness over a large enough sample size, but i'm assuming you don't want to have to attack your enemy several hundred times to start approaching the limit of your original damage output.

So? It was just a demonstration of the principle. Are you saying that you doubt that expected DPS will always be the same because I used a low sample size? I can prove this with any sample size, any crit chance, and any AD value you choose.

It won't always be the same. Over a large enough sample size it could get close, but if it ever does reach the same value, that in itself is a form of RNG, flip a coin 1000 times recording outcomes twice. theoretically you should get 500 heads and 500 tails, twice. You won't. You also won't get the same numbers. It will be close, but theoretical probability very rarely effects real outcomes.

It's 50% because there are two possibilities... that's not how probability works. If you want to say there would be a class-wide nerf or buff, then you have to explain why. Please do, by the way. Tell me what you think will change. I am willing to discuss this change in relation to any champion, item, role, etc. I do not think that there will be any significant changes, and I am I very willing to discuss that.

Attack speed based champions that build high crit chance would be effected the most, as the game system stands attack speed steroids on high crit chance champions are one of the most broken features in the game, and indeed are the entire argument for removing the randomness in the first place. Yasuo would be effected the least due to his very rare less than 100% crit chance.

Two things: First, so what? What effect does this randomness have on balance? I submit that has virtually none. People play based on the expected values, and mathematically, the points of major variation from the expected values are by far the exceptions. It's a rare outcome. Additionaly, no one plays around the extremely small possibilities. I don't start a fight with an IE knowing I can only win if I crit 8 times in a row. Likewise, I don't avoid a fight because I might lose it if my 50% crit chance build doesn't crit at all. People always play to the expected value, and the expected value is by far the most common outcome. What champion or tactic is based on variance? I can't think of a single one (expect maybe very arguably Trynd's laning). Why anyone base their tactics on RNG? It has just as much potential to screw them as help them. So, my question is, who do you think would be affected in a way that alters their balance state?

Randomness has a huge effect on balance. It IS imbalance. https://www.youtube.com/watch?v=Exdil7N1Mdo I'm fully on board with fixing crit, in it's current state it never belonged in a moba in the first place. However, doing so requires not only extensive testing to ensure that damage actually IS similar, not just theoretically similar, but also multiple champion reworks. True, it would be easier than say, removing penetration as a mechanic, which besides being unnecessary would require full revamps of multiple systems, but it's not easy, and it can't possibly happen over night.

Second... no you can't. Not in League at least. League doesn't actually use RNG. It uses pRNG (psuedo RNG). This is a system used to decrease variability in random number generation. The way it works is by increasing the chance of an event every time the event does not happen. For League, this means that if I have 50% crit chance, my first hit may actually have 30% crit chance, then my second will have 40%, then 50%, and so on (the actual algorithm is a lot more complex, but you get the idea). Getting a crit lowers it back to the initial chance. Multiple crits in a row lower the chance even further (in some pRNG systems) After a point, the odds of not getting a crit or getting a streak are mathematically insignificiant. So, we aren't even shifting from true RNG to a normalized system, so we aren't even getting as big a variance decrease as it additionally appeared.

Again, odds aren't a reflection of what actually happens, only what should happen. People win the lottery. pRNG itself is a more reliable system than RNG, but having the same odds as being hit with a lightning bolt the second after you hear you won the lottery while sitting in a basement, as getting a crit on your next attack, does not make that crit impossible.

This is why anything removing crit's randomness requires extensive testing and not just plugging in numbers.

Sightless66Dec 25, 20143 votes

That's not finished.

Ugh, I knew I was forgetting something. My bad. Long day. The point was just that expected damage and variance ranges at any given probability can easily be calculated.

theoretically you should get 500 heads and 500 tails, twice. You won't.

True. Then, do that 100 more times. You'll find that you'll vary above 500 tails about as often as you vary below 500 tails. That's the important point. Variance is not impactful on balance because it is just as likely to screw someone as it is to help them, and it can't be predicted, meaning the only thing anyone plays around is the expected result. The doublelift clip you linked is a good example of this (or would be, except that it wasn't actually a non-expected result. He didn't crit for most of his earlier hits. He just noticed the later hits more because he was about to die. Someone mathed it out on the Reddit thread back when that first showed up, and it was mathematically a pretty predictable result). Doublelift played around the expected result, and was surprised when he didn't get it. However, it is just as likely that the enemy could have had more crits, and killed him much quicker. Variance could just as easily have favored his opponent as it could have favored him. In terms of balance, something that has no systematic benefit is not actually affecting overall results. For example variance did not exist, we'd expect people to win and lose the exact same number of lanes, just without ever getting screwed or screwing someone else (which again, we expect to happen at equal rates).

The point is, he didn't play around that option, and neither his champion nor his role has any systematic benefits from the existence of variance. I'm not saying the game wouldn't change at all. Of course it would change. I'm saying that balance wouldn't change because variance has no systematic advantage or disadvantage for anyone.

However, doing so requires not only extensive testing to ensure that damage actually IS similar, not just theoretically similar,

This makes no sense. Damage in League is a math equation. This is just another math equation. DPS would be the same. We can state that as a fact. We could test it, but that's just plugging numbers into the game and letting the equation act out instead of plugging the same numbers into an equation and getting the same outcome.

The important question isn't whether damage is the same, but whether or not variance is a quality that systematically benefits or harms certain champions (for example, if "fishing for crits" is actually central to the way some champions lane). I don't think there is, and I've never seen a good example of anything that would actually change.

Attack speed based champions that build high crit chance would be effected the most, as the game system stands attack speed steroids on high crit chance champions are one of the most broken features in the game, and indeed are the entire argument for removing the randomness in the first place.

I actually think that the low attack speed values are where the randomness affects the game the most, because with higher attack speed values the variance tends to normalize in most situations. Easier to normalize over 10 attacks than over 4 in the same period of time. Higher sample size and all that.

To wrap it up, of course this would all be tested. There is no reason not to test it even if we can mathematically predict an outcome. However, testing would likely be quick and easy unless we find a systematic benefit to some champion or role from variance (something I don't think we're likely to find), and it promises to be much less impactful on balance and gameplay than any other proposal made.

Here's what I'd like, from you or anyone. I'd like a specific example of a champion, role, item, whatever, and how having variance removed would benefit or harm it. I have never seen a good example of this. Let's start with ADCs. You said earlier that they would be buffed by this. How so? I'd still like that answer.

Pinker 10Dec 25, 20141 votes

I'm not gonna quote that, because I think this line is getting long enough already~ sorry to any random that had to read through all that.

The biggest problem that I see happening with turning crit chance into %damage, in terms of item balance is the IE passive.

You say you don't see a scenario where it would nerf anyone/thing, but how do you compensate for 250% crit damage every however many attacks determined by crit chance where normal crits are 200% when you turn crit chance into a 1%:1% damage bonus rather than a chance for increased damage without altering the damage output of every ad champ that buys it? You can't be critting for 250% every attack, but 30% crit chance vs 90% crit chance with current and proposed setups would fundamentally be required to do different damages than their respective counterparts.

I.e. 130% damage with 30% crit over, since were going for simplicities sake in this thread for demonstrative purposes, three attacks is 390%, as opposed to 350%. That's already a buff of 40% damage. On the other hand, 90% crit loses 450% damage on your setup over 10 attacks.

Sightless66Dec 25, 20145 votes

Edit: the delays on my flight finally cleared, so I'm out. It will be a few days before I'm back, so I'm probably not going to make a response. If you think you've got better things to do than write for someone who won't be here to see it, I'll understand.

Fair enough not quoting. It is going on quite long.

IE isn't hard. Right now, IE passive means that you get a 1.5% increase to dps for every point of crit instead of 1%. For example, if I have IE and 50% crit chance and 100 AD (numbers chosen for simplicity, although I can replicate this with realistic numbers and game scenarios if you would like)., I would have one attack for 100, and one for 250, for 350 damage. A derandomized system would have 100 ad, increased by (50% * 1.5 = 75%), for 175 AD, which would be 350 over two attacks.

Another example: We can do the same for 25% crit chance and 200 AD. Random: 3 attacks for 200, one for 500 = 1,100 damage. Derandomized: 200, increased by (25% * 1.5 = 37.5%) for 275 damage. 4 attacks at 275 = 1,100 damage.

We can do this for any crit damage modifier. As a rule, 1% extra crit damage is a 0.01% multiplier to the damage given by crit chance. IE gives an extra 50% damage to crits, so you multiply the crit chance by to get the crit damage.

So, as I see it, expected dps would be exactly the same with an IE with a random or derandomized system.

I'm not sure what exactly you did in your example. Can you explain it a bit more thoroughly?

Are there any other specific problems you can think of?

Pinker 10Dec 25, 20143 votes

Dunce moment there on my part, didn't do 1.5% for +150% x.x

The only thing that would require testing in this now without having any more fails in logic (almost mentioned crits not being reduced by armor x.x) would probably be the coding of programming every hit to partially ignore armor completely, up from only the crits themselves. Given the sloppy way the base of this game is coded, where everything interacts with everything else and nerfing Elise buffs magic pen or other random such events, I see that creating lots of problems.

However, as to (most) of my original arguments, well done sir.

Still gonna require testing though, they break something new every patch.

APrinnyDoodDec 25, 20142 votes

In spaghetti code we trust. I think I'd love to see the funny interactions that would come from it. Maybe crit is tied up with walking and Yasuo would gain terrain-crossing powers. Maybe crit would be tied in with magic resistance and we'd see Galio doing 1,000 damage autos. With Riot, every patch is a new adventure into the unknown!

MakersFDec 26, 20143 votes

OMG, that is one the most ignorant thing I ever read about probability. Here you can go talk with other people that have no idea what probability is and still defend their ignorance.

If you don't know a thing, you shouldn't talk about it.

If you throw a coin 1000 times, the number of heads are described by a normal distribution with mean = 500. This is not an opinion, this is THE ONLY TRUTH. Please avoid to spread further ignorance.

If you don't believe me, take a coin and throw it 10 times, write the number of heads you got. Repeat as many times as you want. The more you repeat the more the average will get close to 5. You don't need to believe me, just take a coin and see it for yourself.

PS: I re read the post multiple times and I'm even most baffled each time..

Pinker 10Dec 26, 20143 votes

Yes, it is possible to get 800 heads. It is possible to get 0 heads. It is also possible for the coin to land on its edge 1000 times in a row, though we usually ignore this possibility and say the only possible outcomes are heads and tails.

We call the number of heads a random variable. Let's say we call it H. The possible values of H are any number from 0 to 1000. H has a probability distribution, which means we can calculate the probability of any of those outcomes. And if you calculate the probability that H = 800, you'll find out it isn't zero, it's just really, really small.

So we can't guarantee that you won't get 800 heads from a fair coin, we just say it's very unlikely. And the probability is so small we're confident in predicting that it won't happen. Nevertheless, it is technically possible.

You're a moron. you will NOT get 500 heads every time you flip a coin 1000 times. Who the fuck taught u statistics They sucked at it.

As did whoever taught u to read.

Simplest possible terms, since you apparently can't read even high school level terms

Toss a coin 1000 times. Write results. Repeat. compare results. Are they the same? No.

ScalyManChardDec 26, 20144 votes

He ain't saying it will always be 500. Mean = average. In your case it's also the most probable outcome, though its still not very likely.

Think his point is the same as mine.

Simple thought experiment to illustrate:

Coin has 50% chance of heads or tails. Toss the coin twice. What is the probability of: Two heads Two tails 1 of each?

Pinker 10Dec 26, 20142 votes

He didn't read mine, I didn't read his. It's only fair.

My original post had absolutely zero to do with the average over the two series, it was simply pointing out that u would neither get 500 heads/tails each time OR the same number of heads/tails each time.

MakersFDec 26, 20143 votes

I wont' lose other time with a self entitled brad who doesn't even understand the difference between independent trials and repeated trials.

The statistic I'm using is the only official statistic that ever existed, is the only one that works and the only one used in real world, and the only one proved correct. The one you are professing, well, I have no idea what it is, because it's clearly wrong.

And since you are so self entitled that you don't want to try tossing 10 coins 5 times, I'll do it for you

http://repl.it/7BT

You just need to press the play button at the top of the text editor. As you can so, doing as many trials as you want, the average is always around 0.5.

Press that button 20 times, and look that you'll get numbers close to 0.5

I can't do more for a person who doesn't want to learn. If you still refuse to learn, then stay in your ignorance

ScalyManChardDec 26, 20142 votes

You aren't more likely to crit for 400 than you are for 1000.

Ummm...

Using coin tosses, chance of 400/1000 heads - 4.6339E-11 1000/1000 - 9.3326E-302

Binomial calculator: http://stattrek.com/online-calculator/binomial.aspx

The chances are vastly different. May be easier to think of with 4 coins - for all 4 coins to land heads they ALL have to land heads but for 2 heads, there are a lot of combinations (tails tails heads heads, heads heads tails tails...) That output that result. Each combination has the same probability but there are more combinations for 400 than 1000.

The probability of a probability like that will follow normal distribution. Once you already have 600 heads the probability of the 601st is indeed still 50% but the probability of 601 consecutive heads is less than the probabilty of 600 consecutive heads - there's 1 more chance for a tails to show up.

Edit: I may mean permutation instead of combination. Its one of those two.

Pinker 10Dec 26, 20142 votes

Coin tosses aren't RNG. RNG as a system is fundamentally flawed in that you actually aren't more likely to get 400 than you are 1000. http://en.wikipedia.org/wiki/Random_number_generation It's called random for a reason. whether you get 400 or 1000 or 80 or 268 is just that, random.

as pointed out below the post you responded to and above this one, league uses pRNG, which, while better, still won't lower you below one percent crit chance. While the odds there are much greater against it, it can still happen.

ScalyManChardDec 26, 20142 votes

Typical Pseudorandom generators utilize some number that, while not "technically" random is close enough. Its because computers can't flip a coin or use particle decay for true randomness so they use something like clock time and apply an algorithm to it to get a function that is effectively random.

Distribution for a good pseudorandom function will be mostly the same as the true random function it is trying to replicate.

RNG as a system is fundamentally flawed in that you actually aren't more likely to get 400 than you are 1000.

If you're talking about LoL being pseudorandom in that it modifies crit chance based on previous results, I don't know if it does or doesn't. Some games do (easy difficulty on XCom) but I don't know why LoL would, it'd take far more effort than just using some RAND function and just make crit harder to understand.

Even if it did, it probably wouldn't work in the way you describe.

It's called random for a reason. whether you get 400 or 1000 or 80 or 268 is just that, random.

On a single roll of a random function with an equal probability of all numbers between 1 and 1000, true. But when you take a bunch of those randomly generated numbers (every crit check / coin toss) the distribution becomes normal.

https://en.m.wikipedia.org/wiki/Normal_distribution

Maybe im not really getting what you mean. But you seem to be confusing the term "random" with "evenly distributed".

Pinker 10Dec 26, 20142 votes

What i mean really has nothing to do with league in the first place, as it only applies to RNG, and not pRNG.

A true RNG doesn't weight a value. In a true RNG scenario, number less than or equal to crit chance is crit, number greater is not, to a maximum of 100. If every individual attack has a 30% chance of being a crit, theoretically at least, one in three should be a crit. With RNG, 92 in a row could be crits, and 60 in a row could be normal hits. There's no weighting, it simply picks a number, completely at random. It could pick 25 12 times in a row and 7 other numbers below 30, or it could pick 19 completely different numbers over 30. Both are equally likely. It just picks. You'd be equally likely to get 300/1000 crits as 600/1000 crits or 1000/1000 crits. There's no weighting, no rules, no reason the machine couldn't just endlessly pick the same number. The problem with comparing RNG to real world scenarios is that RNG has only one variable ~ which number happens to randomly be generated. That's the only variable. If it's under a certain value, it's a normal hit, if it's above, it's a crit, but that has nothing to do with an RNG system, and only with the system that takes that randomly generated number and decides to do something with it. To get close to being 1/3 hits as a crit with a system like that would be based on two things, sample size, and luck ~ both of which, in the case of RNG, can change. If you got luckier, you might get it in 27 hits, if not, it might take 500.

The closest RNG can actually get to a real world scenario is a virtual coin toss, it either picks 1, or 2. It is still not, however, entirely true to a real world scenario. There's no gravity. There's no spin.

League most likely uses pRNG because that can at least be controlled and/or weighted, if necessary. (which it definitely is, in terms of the crits i used in the above example, to actually make 300/1000 more likely than 1000/1000) True RNG has no likelihood of getting one thing more than the other.

And you people thought pRNG in league was bad.... Imagine how fcked crit would be if it was true RNG.

ScalyManChardDec 26, 20143 votes

What do you mean by "RNG"? You use that term like its some function or system I am unaware of.

If every individual attack has a 30% chance of being a crit, theoretically at least, one in three should be a crit.

That's not true, it instead means that over a large enough sample size about 30% of them will be crits. You can describe the chance of each combination of crits and non-crits but there is no single " expected" number of crits, unless you want to refer to the most probable (likely integer) number as the expected value.

You'd be equally likely to get 300/1000 crits as 600/1000 crits or 1000/1000 crits.

You seem to have some grasp of statistics but the above is simply not true and I'm not sure where the disconnect is. If by RNG you mean the " assisted rng" as described elsewhere in these comments, itd actually make 1000/1000 even less likely than "true" rng. If possible please try the 2 coin toss thought experiment I posted in another comment.

Edit: again here for reference -

Coin has 50% chance of heads or tails. Toss the coin twice. What is the probability of: Two heads Two tails 1 of each?

Pinker 10Dec 26, 20141 votes

If i meant the pseudorng, i would refer to the pseudorng. A random number generator is just that. A random number generator. It has a set of numbers to pick from and it picks. At random. keyword, random. Hence why i said, at the start of the post, it had nothing to do with league. And at the bottom, that league uses pseudorng so 300/100 is more likely than 1000/1000 Which it isn't in unassisted RNG. It can't be. By definition.

To clarify True RNG such as i'm referring to, isn't actually used by anything. It doesn't have seeds or keys, and is actually random I.e 25x19 could happen equally as likely as 25 seperate entities. I've gone so far off topic i'm not even sure how i got here.

ScalyManChardDec 26, 20141 votes

OK but "RNG" as you talk about doesn't even exist as a concept.

And at the bottom, that league uses pseudorng so 300/100 is more likely than 1000/1000 Which it isn't in unassisted RNG. It can't be. By definition

It is, in true RNG, because you are talking about multiple trials. If you're talking about some magical "true random" then get out of here because you are just as likely to get "5th dimensional aqua-green cat alien" as you are to get "25 19s"

I'll go ahead and do the coin toss experiment for you -

2 coin toss. Chance of following: 2 heads - .25 2 tails - .25 1 heads, 1 tails - .5

The way you explain things, it'd be 33.333...% for each which is not true and I'll continue:

The 2 coin tosses have 4 permutations of equal probability (because each outcome of a single toss has the same probabilty). Heads heads, heads tails, tails heads, tails tails. Of these, two of them (heads tails, tails heads) are the same combination.

In a coin toss all permutations have the same probability but combinations have different probabilities. The chance of the 400th toss being heads or tails is the same but the chance of getting 400 total heads or crits is far different than the chance of 1000 heads or crits.

Basically you are stating that in some magical nonexistent RNG function you describe all combinations of a set of trials have an equal probability. Which is a ridiculous thing to state.

Pinker 10Dec 27, 20141 votes

Seriously. Unless ur going to attempt to discuss what I'm actually talking about instead of spamming coin tosses, not reading my posts, and pretending statistics is useful outside of teaching statistics and internet forum arguments, stop spamming my notifications.

ScalyManChardDec 27, 20141 votes

Word, my bad, I'll try to lay off. Godspeed in your endeavors.

StephenizgodDec 27, 20141 votes

Ok i agree it wont be easy, but it needs to be done regardless of whether it is easy or not. Crit is just unhealthy for the game, ADC's need a more reliable and fairer form of damage to all the other roles. That downvoted comment by Teemohunter is a perfect example of why it needs to be changed, by all right he should have lost and yet he didnt. Is it fair that he won simply because luck was on his side? Not at all. Luck in League should be something like when you flash over a wall running from one enemy only to realize there was another enemy waiting in the bush next to the wall and if you didnt flash you would have died. Thats healthy luck, not buying an item that ramps up your damage randomly.

I dont think it would take a couple years tbh, maybe a 3-4 months. There are a ton of good ideas, like you said, that can be a good start for removing crit. Once they pick or create the perfect solution, it shouldn't take too long to change Crit to something less toxic. I mean every preseason they make huge changes to the game, this would just be one of the hugest changes.

Kuroi86Dec 25, 20141 votes

The only champion who would need more than cursory modifications is Trynd, because he has a non-damage effect on critting (getting more fury).

Tryndamere Ashe Yasuo

Also what's this crap about damage vs non-buildings? =/ Are you seriously talking about taking away one of the ADC main purposes, pushing towers?

Sightless66Dec 25, 20143 votes

Crit doesn't work on towers. This would just be replicating current functionality. The point is that nothing changes.

Yasuo is easy. Each point of crit would increase his autoattack damage by 1.9%. That's exactly the same as it is right now. All he has is an increase in crit chance and a lower crit damage. That's just math. Out of all the crit-champions, he is by far the easiest fix.

Ashe is also easy. She builds up focus stacks until she does double damage (increased by crit damage modifier). Starting point is determined by crit chance.

LordOfTheLeagueDec 25, 20144 votes

Tryn is really easy too. His passive is obvious so I won't talk about it.

Spinning Slash The current mechanic : The cooldown is reduced by 1 seconds whenever Tryndamere critically strikes (doubled against champions).

With the % aa reworked crit : The cooldown is reduced by 0-1 (scale from 0 to 1 depending on his crit chance) seconds whenever Tryndamere auto attacks (doubled against champions).

If it's not clear he just reduce the cd for 0 seconds when he has 0% "crit chance", 1 seconds when he has 100% (the same as live). When he has 40% crit, he reduce the cd of the skill for 0.4 seconds.

Not so hard.

I don't think the % crit => % auto attacks bonus damage change is hard at all. Riot is just lazy for that.

wolfaxDec 26, 20143 votes

Except every crit item just became a whole lot more appealing to every bruiser since they don't need to stack several to make them effective. The item literally read take this and deal X% more damage. The fact is crit is only really viable as the needed ADC steroid because it requires several pieces to make it consistent enough.

JustMyBassCannonDec 25, 20145 votes

Well, actually, Fiora didn't have a Dodge Chance. It was Jax, Sivir and Udyr who had innate Dodge Chance. But the principle was still the same; you either stacked a high % Dodge on one of them with runes and Ninja Tabi, or you just went with what you got out of the runes+Tabi on another champ. Either way, you were sitting on at least a 20% chance to randomly nullify an attack.

Crit is equally wide-spread in usability, as everyone can Crit because everyone has basic attacks. There are, however, many more champions with abilities that can Crit or are modified by Crit modifiers. A systemic rework of Crit Chance is especially going to affect all Marksmen and all champions with kit Crit effects.

On top of all that, Marksmen are kind of balanced around the fact that they have Crits. They need that extra damage output so that their general weakness to anything that can punch their faces in for 2 seconds doesn't make them irrelevant.

Pinker 10Dec 25, 20141 votes

Her parry was her dodge, albeit on a long-ish cooldown.

JustMyBassCannonDec 25, 20145 votes

Riposte still is a one-shot Dodge. And Jax still has Counter Strike.

Fiora wasn't even around during the Dodge removal, iirc.

Pinker 10Dec 25, 20141 votes

3 patches later ~ Stoner memory doesn't quite last 3 years i'm afraid.

Still though, can't believe people think crit would be easier to remove than dodge.

like, ten people used dodge, tops.

I think it hurt more removing FoN than it did dodge.

1st HanarSpectreDec 25, 20143 votes

[{quoted}](name=Pinker 10,realm=NA,application-id=3ErqAdtq,discussion-id=L1XEKzEj,comment-id=00000001000000000000,timestamp=2014-12-25T19:57:51.652+0000)

3 patches later ~ Stoner memory doesn't quite last 3 years i'm afraid.

Still though, can't believe people think crit would be easier to remove than dodge.

like, ten people used dodge, tops.

I think it hurt more removing FoN than it did dodge.

I used dodge seals, on all my characters, so did a lot of people, they were the best yellow runes by far.

DurzakaDec 25, 20141 votes

[{quoted}](name=Pinker 10,realm=NA,application-id=3ErqAdtq,discussion-id=L1XEKzEj,comment-id=0000,timestamp=2014-12-25T18:24:50.453+0000)

Because removing crit is a lot more work than removing dodge

They had to rework a single skill on jax one on fiora and remove some runes

Crit is like... everywhere.

If you seriously expect reworking everything that uses crit to use something else to take less than a couple years...

Not to mention, as it has been explained NUMEROUS times, that dodge isnt fun RNG. You either dodge or you dont, and you dont notice it. And you cant do anything to counter someone with dodge, they just dodge your attacks and you get upset.

Crit is fun RNG. You feel the power when you hit someone in the face with a crit. And unlike dodge, you can still counter crits with armor.

EDIT: Pinker 10 really understands. Crit is absolutely a problem that needs to be fixed. But it is a nearly impossible problem to fix properly, as he has explained several times in this thread.

Sightless66Dec 25, 20143 votes

Crit is fun RNG. You feel the power when you hit someone in the face with a crit.

I'd argue that this benefit is eclipsed by the number of frustrating instances where you lose a fight that you had every expectation to win.

Pinker 10 really understands. Crit is absolutely a problem that needs to be fixed. But it is a nearly impossible problem to fix properly, as he has explained several times in this thread.

I don't think his explanations have been sufficient. He seems to have it as an assumption that variance benefits certain classes or champions, which would mean that any solution would have to take that into account and balance accordingly, but I don't think he's justified that point, just asserted it. I see no evidence that variance is a beneficial or harmful point to any class to a different degree than it is for any other. I think fixing crit would require much, much less work than he thinks it would.

DurzakaDec 25, 20141 votes

{quoted}

I'd argue that this benefit is eclipsed by the number of frustrating instances where you lose a fight that you had every expectation to win.

I don't think his explanations have been sufficient. He seems to have it as an assumption that variance benefits certain classes or champions, which would mean that any solution would have to take that into account and balance accordingly, but I don't think he's justified that point, just asserted it. I see no evidence that variance is a beneficial or harmful point to any class to a different degree than it is for any other. I think fixing crit would require much, much less work than he thinks it would.

Losing a fight isnt supposed to be fun. Thats a fact you are never going to get around. But when you won a fight with dodge, you felt nothing. You probably barely even noticed you dodged that AA.

The point he is making is that crit isnt a flat power scaling. it becomes stronger the more you buy. hence the 10% on triforce is very weak, and doesnt account for much power. But if that was flat out 10% more damage, that is a HUGE deal, and much more beneficial for champions like Corki and Ezreal.

As sample sizes get higher and higher, the difference between Crit and % DMG Amp that you are talking about becames almost nothing. But the smaller the sample size, the bigger the difference. And in League you will never have large sample sizes, because most fights wont be lasting beyond 20-30 AAs from an ADC at max anyways.

The gating on Crit is that it is late game scaling with some use early game. But its true power doesnt shine until you past 50-60% and can actually rely on getting a crit during an AA exchange.

Prometheus856Dec 26, 20142 votes

if they'd get off their butts and just do it...

seriously, for the most played game in the history of the world, riot's development team is a giant dinosaur. It's like there's one guy on the management team that wants to read and sign off on everything from balance changes to skin ideas to skin development to esports interviews.

I have no idea WTF riot's doing all the time. Does the balance team do literally nothing but 3-4 minor number tweaks every 2 days?

TeemoHunter1337Dec 25, 20144 votes

crit saved me so much.

i was once fighting a vayne who had bork and i had infinity edge. i was graves.

she was 50 pct. i was 25 pct.

we both had no mana.

she got me down to 5 pct hp, i got her down to 20 pct hp.

i knew i was about to die as we were both autoattacking each other, then my last autoattack was a crit. and she died instantly. i lived.

i knew without crit id have died but man the RNG saved me when i should have died lol.

Noam BraumskyDec 25, 201435 votes

Is this supposed to make me like Critting? This proves the point of how awful it is, not to mention how Infinity Edge is practically mandatory on every adc in the game. That Vayne should have beaten you. In this situation Vayne has the upper hand presumably because she played better then you earlier or you didn't back off soon enough from a fight, etc. That was her win, but because of this rng crap you beat her in a 1v1 because.... what was the point of crits again?

Done25Dec 25, 20145 votes

{quoted}

Is this supposed to make me like Critting? This proves the point of how awful it is, not to mention how Infinity Edge is practically mandatory on every adc in the game. That Vayne should have beaten you. In this situation Vayne has the upper hand presumably because she played better then you earlier or you didn't back off soon enough from a fight, etc. That was her win, but because of this rng crap you beat her in a 1v1 because.... what was the point of crits again?

I think he's agreeing with you? At the very least his story supports your viewpoint.

The sad truth is that Riot CAN'T remove crit, as it serves for a very vital inhibitor to marksmen. You can't crit towers. It gives them something to dump gold in to increase their killing power against organics, without allowing them to hit towers for a lucky 1000+ damage.

SecretAgentHulkDec 25, 20147 votes

Do you mean to say that there is no conceivable way that Riot could implement bonus damage, but only to non-structure targets?

TeemoHunter1337Dec 25, 20143 votes

{quoted}

Is this supposed to make me like Critting? This proves the point of how awful it is, not to mention how Infinity Edge is practically mandatory on every adc in the game. That Vayne should have beaten you. In this situation Vayne has the upper hand presumably because she played better then you earlier or you didn't back off soon enough from a fight, etc. That was her win, but because of this rng crap you beat her in a 1v1 because.... what was the point of crits again?

what? why are u acting like the toxic feeders in my ranked games lol.

my story was to give further proof to your point. to be short, i was agreeing with u.

i actually expected upvotes LOL.

Noam BraumskyDec 26, 20142 votes

He said "rng saved me lol" so I took it as a point in criticals' favor.

TeemoHunter1337Dec 26, 20142 votes

{quoted}

He said "rng saved me lol" so I took it as a point in criticals' favor.

it did save me. i wasnt really taking sides here. iwas just giving an example of critical strike in your discussion, so you can have further details when people read it.

no needed to downvote me or judge.

DUDE BRODec 26, 20142 votes

Except Crits aren't true RNG in LoL. They changed that around the same time they removed dodge.

LoL has assisted RNG for crits, so if you have 20% crits, LoL will force a crit about twice per 10 hits. True RNG could give you 0 in 10 hits, or even 6 in 10 hits from 20%, but LoL removes a lot of the 'randomness', so 20% crit really is more 20%, and less random.