5% in the real world is rare to never

hephaestus · July 2019

DNA3000 said:
Just to expand a bit, if there's a 3% chance for something to happen and a 97% chance for something to not happen, then the fundamental counting principle would work something like this. In the first swing, there are one hundred possibilities: EEEHHHHHHHH....H. Three evades and ninety seven hits. For each of those possibilities you'd expand one swing more: each "E" could be followed by any of the "EEEHHHHH...HH" and so on. No one would actually do that and count them up, but it may now seem more obvious that in the first swing 97% of the swings are "H". After two swings 97% of the second swings are also H, which means 3% of all the original "H"'s become "HE" and there's only 97% of 97% that is "HH". After three swings, the number of sequences that is "HHH" is 97% of 97% of 97% - the number of "survivors" keeps getting cut by 3%. This is where the M^N math in statistics comes from.

Thank you for explaining that. So what is the actual formula for what we are talking about then? How many evades should one truly expect in, say, 100 swings? Or if it's easier to do the math, reduce the sample to something manageable.

DNA3000 · July 2019

hephaestus said:

DNA3000 said:
Just to expand a bit, if there's a 3% chance for something to happen and a 97% chance for something to not happen, then the fundamental counting principle would work something like this. In the first swing, there are one hundred possibilities: EEEHHHHHHHH....H. Three evades and ninety seven hits. For each of those possibilities you'd expand one swing more: each "E" could be followed by any of the "EEEHHHHH...HH" and so on. No one would actually do that and count them up, but it may now seem more obvious that in the first swing 97% of the swings are "H". After two swings 97% of the second swings are also H, which means 3% of all the original "H"'s become "HE" and there's only 97% of 97% that is "HH". After three swings, the number of sequences that is "HHH" is 97% of 97% of 97% - the number of "survivors" keeps getting cut by 3%. This is where the M^N math in statistics comes from.
Thank you for explaining that. So what is the actual formula for what we are talking about then? How many evades should one truly expect in, say, 100 swings? Or if it's easier to do the math, reduce the sample to something manageable.

On average, you should expect 3. If the chance for an evade is 3% and each evade happens independently, then the average is easy to calculate: its 3% of 100. But what complicates this is that the average isn't necessarily what anyone sees. Let's simplify a bit and talk craps.

You roll two six sided dice, what value would you expect on average? On average, you should expect to see a seven. But what do you see? Well, on average you'd sometimes see seven, but sometimes see twelve, sometimes see eight, etc. There's a mathematical formula that can calculate the distribution of results, how often you see two, how often you see three, etc. In the case of the evade situation above, you'd see zero evades 4.8% of the time, for example.

But here I should point out that something else having nothing to do with probability can affect what you see. Kabam has never described the mechanics of evade, but testing and investigation has shown that what the game calls "evade" is often a hidden "buff" that triggers actual evades. For example, players often see evading champs like Spiderman seemingly evade two hits in a row in rapid succession. In technical terms that might not be two evades triggering, that might be one evade "buff" triggering, and causing Spiderman to evade all normal attacks for a fraction of a second. If you count that as two evade triggers, it can seem like Spiderman is triggering evade more often than his stated probability.

hephaestus · July 2019

OK, I don't disagree with how the math works on a per-swing basis. Really, I never figured it worked any other way. But your argument seems to be, "Well, since the computer doesn't care about past or present instances, you shouldn't be surprised that 7% happens nearly every time." And that still doesn't seem like a valid argument.

When I fight a mesmerize node, every time, without exception, nearly all hits result in a stun. And nearly all people I've mentioned this to say the same thing. In real life you would not expect that to happen on 7%/instance odds.

So you're saying the math doesn't compound, that's fine. I'm saying that the math may be off in the programming though. Like maybe someone got the decimal wrong and multiplied by .7 instead of .07. Because 70% is way more inline with the actual real-life occurrences than 7%.

And since I know that Kabam will never copy and paste the actual line of code showing the math, this will never be resolved.

(I'm using mesmerize as my example because I really don't find that BW evades that often for me and I don't really hear people complain about her much. But 100% of everyone I have personally talked to about mesmerize complains. Yes, I know people are now going to say they don't have issues with it just to spite me!)

DNA3000 · July 2019

hephaestus said:
OK, I don't disagree with how the math works on a per-swing basis. Really, I never figured it worked any other way. But your argument seems to be, "Well, since the computer doesn't care about past or present instances, you shouldn't be surprised that 7% happens nearly every time." And that still doesn't seem like a valid argument.

When I fight a mesmerize node, every time, without exception, nearly all hits result in a stun. And nearly all people I've mentioned this to say the same thing. In real life you would not expect that to happen on 7%/instance odds.

I'm just explaining the math, the math doesn't explain what you're describing. But if things really do happen as you describe, that would be easy to prove. Just video record all your fights against mesmerize.

In practice, in the past when people have said "this always happens" what tends to happen is when asked to provide evidence of it happening first they back down a bit and say "it doesn't really happen all the time, just most of the time" and then "it happens more often than it should" and then "it doesn't happen often enough for me to be able to video it, but it still happens way too often." Which is damaging to the credibility of the assertion.

If anyone could show that all, or nearly all hits result in stun, with a large enough sample size, this would be evidence something might be bugged. Maybe the way the random roll was happening, or maybe something in the way mesmerize was implemented having nothing to do with the random roll. But I've never seen this and I haven't seen video evidence of it. I've only heard anecdotes and lots of other people nodding their heads. In the past, the complaint was that video evidence was hard to get without special software. But today, pretty much all supported platforms for MCOC directly support recording gameplay.

Even if this were inconvenient for you to do, there's thousands of players out there: this bug has managed to avoid basically everyone catching it on tape. The odds of that happening are lower than the odds of mesmerize actually doing it.

hephaestus · July 2019

OK, I don't disagree with how the math works on a per-swing basis. Really, I never figured it worked any other way. But your argument seems to be, "Well, since the computer doesn't care about past or present instances, you shouldn't be surprised that 7% happens nearly every time." And that still doesn't seem like a valid argument.

When I fight a mesmerize node, every time, without exception, nearly all hits result in a stun. And nearly all people I've mentioned this to say the same thing. In real life you would not expect that to happen on 7%/instance odds.

So you're saying the math doesn't compound, that's fine. I'm saying that the math may be off in the programming though. Like maybe someone got the decimal wrong and multiplied by .7 instead of .07. Because 70% is way more inline with the actual real-life occurrences than 7%.

And since I know that Kabam will never copy and paste the actual line of code showing the math, this will never be resolved.

DNA3000 said:

hephaestus said:
OK, I don't disagree with how the math works on a per-swing basis. Really, I never figured it worked any other way. But your argument seems to be, "Well, since the computer doesn't care about past or present instances, you shouldn't be surprised that 7% happens nearly every time." And that still doesn't seem like a valid argument.

When I fight a mesmerize node, every time, without exception, nearly all hits result in a stun. And nearly all people I've mentioned this to say the same thing. In real life you would not expect that to happen on 7%/instance odds.
I'm just explaining the math, the math doesn't explain what you're describing. But if things really do happen as you describe, that would be easy to prove. Just video record all your fights against mesmerize.

In practice, in the past when people have said "this always happens" what tends to happen is when asked to provide evidence of it happening first they back down a bit and say "it doesn't really happen all the time, just most of the time" and then "it happens more often than it should" and then "it doesn't happen often enough for me to be able to video it, but it still happens way too often." Which is damaging to the credibility of the assertion.

If anyone could show that all, or nearly all hits result in stun, with a large enough sample size, this would be evidence something might be bugged. Maybe the way the random roll was happening, or maybe something in the way mesmerize was implemented having nothing to do with the random roll. But I've never seen this and I haven't seen video evidence of it. I've only heard anecdotes and lots of other people nodding their heads. In the past, the complaint was that video evidence was hard to get without special software. But today, pretty much all supported platforms for MCOC directly support recording gameplay.

Even if this were inconvenient for you to do, there's thousands of players out there: this bug has managed to avoid basically everyone catching it on tape. The odds of that happening are lower than the odds of mesmerize actually doing it.

First, thank you for reasoning through all this calmly and logically without insults and attitude. That is rare on this forum.

Second, the post of mine that you are quoting isn't showing up for me. I was actually in the process of re-typing it when you replied to it. I've shift-F5-ed my browser several times now. Weird.

Third, I have not found a screen recorder for Android that works well yet. Every one I've tried causes an offset in input vs screen output, making it nearly impossible to fight. Do you know of a free one that works well?

Fourth, IOS screen recorder does work well and I would like to try to capture evidence later when I am on my iPad. Is there a mesmerize node that you can think of that is easy / cheap to get to that I can play on?

Wubbie075 · July 2019

DNA3000 said:
But here I should point out that something else having nothing to do with probability can affect what you see. Kabam has never described the mechanics of evade, but testing and investigation has shown that what the game calls "evade" is often a hidden "buff" that triggers actual evades. For example, players often see evading champs like Spiderman seemingly evade two hits in a row in rapid succession. In technical terms that might not be two evades triggering, that might be one evade "buff" triggering, and causing Spiderman to evade all normal attacks for a fraction of a second. If you count that as two evade triggers, it can seem like Spiderman is triggering evade more often than his stated probability.

It's easier to observe this with Miles Morales and his evade counter. I noticed that I'd often trigger two evades very quickly but his counter would only go down by one. My assumption for why that happened was pretty much what you described, even if I didn't understand the specific game mechanics that caused it.

hephaestus · July 2019

Wubbie075 said:

DNA3000 said:

hephaestus said:

DNA3000 said:
But here I should point out that something else having nothing to do with probability can affect what you see. Kabam has never described the mechanics of evade, but testing and investigation has shown that what the game calls "evade" is often a hidden "buff" that triggers actual evades. For example, players often see evading champs like Spiderman seemingly evade two hits in a row in rapid succession. In technical terms that might not be two evades triggering, that might be one evade "buff" triggering, and causing Spiderman to evade all normal attacks for a fraction of a second. If you count that as two evade triggers, it can seem like Spiderman is triggering evade more often than his stated probability.
It's easier to observe this with Miles Morales and his evade counter. I noticed that I'd often trigger two evades very quickly but his counter would only go down by one. My assumption for why that happened was pretty much what you described.
I would write that off as the evade taking longer than the animation, or at least the end of the animation happening before the end of the evade, so to make the animation coincide with the hit, it has to trigger the animation twice while the evade only triggers once. At least that's what I was always assuming was happening when I noticed that.

hephaestus · July 2019

Hm. I think my display is somehow being screwed up by every one of my comments suddenly having to be approved. I don't know why that would be?

hephaestus · July 2019

Wubbie075 said:

DNA3000 said:
But here I should point out that something else having nothing to do with probability can affect what you see. Kabam has never described the mechanics of evade, but testing and investigation has shown that what the game calls "evade" is often a hidden "buff" that triggers actual evades. For example, players often see evading champs like Spiderman seemingly evade two hits in a row in rapid succession. In technical terms that might not be two evades triggering, that might be one evade "buff" triggering, and causing Spiderman to evade all normal attacks for a fraction of a second. If you count that as two evade triggers, it can seem like Spiderman is triggering evade more often than his stated probability.
It's easier to observe this with Miles Morales and his evade counter. I noticed that I'd often trigger two evades very quickly but his counter would only go down by one. My assumption for why that happened was pretty much what you described, even if I didn't understand the specific game mechanics that caused it.

I would write that off as the evade taking longer than the animation, or at least the end of the animation happening before the end of the evade, so to make the animation coincide with the hit, it has to trigger the animation twice while the evade only triggers once. At least that's what I was always assuming was happening when I noticed that.

DNA3000 · July 2019

hephaestus said:

Wubbie075 said:

DNA3000 said:
But here I should point out that something else having nothing to do with probability can affect what you see. Kabam has never described the mechanics of evade, but testing and investigation has shown that what the game calls "evade" is often a hidden "buff" that triggers actual evades. For example, players often see evading champs like Spiderman seemingly evade two hits in a row in rapid succession. In technical terms that might not be two evades triggering, that might be one evade "buff" triggering, and causing Spiderman to evade all normal attacks for a fraction of a second. If you count that as two evade triggers, it can seem like Spiderman is triggering evade more often than his stated probability.
It's easier to observe this with Miles Morales and his evade counter. I noticed that I'd often trigger two evades very quickly but his counter would only go down by one. My assumption for why that happened was pretty much what you described, even if I didn't understand the specific game mechanics that caused it.
I would write that off as the evade taking longer than the animation, or at least the end of the animation happening before the end of the evade, so to make the animation coincide with the hit, it has to trigger the animation twice while the evade only triggers once. At least that's what I was always assuming was happening when I noticed that.

This mechanic was first noted with Ultron who can only trigger evade once in a given window of time, but could apparently evade two attacks in rapid succession. The conclusion was that both evades were actually one trigger of evade.

It makes sense as well when you think about a champ like Sparky evading a multistrike special attack. He doesn't have a 100% chance to evade those, but when he evades he generally evades all parts of the attack and the player sees multiple "evade" messages. But those multiple evades have to be part of the same evade, because he doesn't have an independent chance to evade each part of the attack separately. Somehow, the one trigger for "evade" causes Sparky to "mini-evade" every part of the special, and we players see those "mini-evades" as apparently individual evades.

DNA3000 · July 2019

hephaestus said:
Is there a mesmerize node that you can think of that is easy / cheap to get to that I can play on?

The easiest one I can think of to get to would be the Elektra boss in Act 4.4.3. Easy in terms of fights, but a boat load of energy to reach (I think you can reach her without energy refills though, starting from full energy). The quicker node would be the Mordo path in Act 6.1.6, but you have to be able to reach 6.1.6 and then clear the center path. But it is a lot less energy to reach.

If there's a closer/easier one, I can't think of it off the top of my head.

KattohS · July 2019

Catherine said:
just fought a small defender in aw with 5% evade global node. unduped wasp evaded every light and medium i threw at her. 5% in kabam numbers is 80% unless it's a crystal opening.

Hahahaha

Rizz_Dizzle2 · July 2019

All I know is that the 0.1% chance of landing featured hero is pretty correct

5% in the real world is rare to never

Comments