More Probability Whining (Legendary Token Pulls)

areyes81

IceIX wrote:

Pylgrim wrote:

Forgive me if I keep plugging my thread but it already has 57 upvotes and several pages of comments, so I think it's important to keep it visible and growing. LTs NEED to change and we need to make sure the devs see this is a concern of a significant part of the playerbase.

We've seen it. Feel free to keep the discussion going if you like. What you appear to be asking for is something to make the game say "Oh, you don't need X-Force anymore? Let's give you less of those". Which would naturally alter the odds in the pack dynamically for you. This would need to be reflected across all packs on a per user basis. It would be a base fundamental change in how covers are provided to the player base, as well as make things orders of magnitude more difficult to balance cover outflow to make sure that (most) players aren't getting covers too quickly. It's not an impossible task, but being 100% honest it's not something that I would expect that we would ever take on for implementation. The alternative that we could do *today* in-game would be to make the Legendary Packs a Vault instead. There are real downsides to that though:

- The Vault would necessarily reset when we add a new 4*/5*. You would have at max 1 month (a season) to pull everything you needed before it may disappear/odds reset - This isn't great for covering characters. Either the Vault is small enough that you could reasonably expect a certain character (100 covers) and probably only 3-4 covers of each character, or it contains enough to cover all the 4*s and the Vault is huge, so odds changes never happen. Then refer back to point 1 on not being able to chew through that. - Even with a Vault, you're still going to have the same issue as current: Out of a 100 piece Vault, players pulling 5 times could very well pull 3 of the 4 available Invisible Woman covers that happen to be hanging out. Chances aren't high of that happening, but it's a near surety that someone will hit that combination with the number of people making pulls. Meaning, it may not even actually change the *feeling* that things are rigged and you're getting screwed even if it does 100% objectively lower the odds of that occurring.

In regards to the more general feeling: We've looked at our odds. We've looked at drops. We continue to look at odds and drop runs. We have yet to find anything anomalous. The numbers of covers of all types from all cover packs have been and still are dropping in the displayed ratios, give or take very much within statistical variations. I totally get that it sucks that you see 3-4 covers in a row, but that's random being random. As was stated above, you never notice when you pull 12 Standards and they're all varied. You *do* notice when you pull 5 Venoms in a row though. Looking at it from a 100 foot view, that's a little streak blip in the otherwise very homogenous statistical run. Just as often, Venom doesn't pop out of 300 pack pulls in a row. But for you personally, that was signs that something is rigged. It really really does come down to being human nature to notice that. I mean, I sure as heck know better and *I* still have a hard time fighting down that feeling when Moonstone pops up 4 times on my 10x Heroic in a season pull.

All this being said, we are working on something that should obviate these feelings some/most of the time. Not ready to announce what that is yet, but all of us (and I'm sure nearly 100% of the player base) totally agree that pulling yet another cover on a finished character is a substandard experience.

I'm not sure if suggested, but I think an easy fix would be implementing a skill option on cover pulls. So instead of pulling a specific color, the user gets a cover for a character and then can choose which color they want or need. So for instance, the leg token reveals Wolverine XForce. The user then gets to choose which color they want (black, yellow, green). It may not change the "I constantly get Star Lord" or anything like that, but at least it gives users the option to get the actual cover they need for a character that is sitting at 5 covers of one color and nothing else but constantly pull that one color. Just a thought

PLBIV

IceIX wrote:

Pylgrim wrote:

Forgive me if I keep plugging my thread but it already has 57 upvotes and several pages of comments, so I think it's important to keep it visible and growing. LTs NEED to change and we need to make sure the devs see this is a concern of a significant part of the playerbase.

We've seen it. Feel free to keep the discussion going if you like. What you appear to be asking for is something to make the game say "Oh, you don't need X-Force anymore? Let's give you less of those". Which would naturally alter the odds in the pack dynamically for you. This would need to be reflected across all packs on a per user basis. It would be a base fundamental change in how covers are provided to the player base, as well as make things orders of magnitude more difficult to balance cover outflow to make sure that (most) players aren't getting covers too quickly. It's not an impossible task, but being 100% honest it's not something that I would expect that we would ever take on for implementation. The alternative that we could do *today* in-game would be to make the Legendary Packs a Vault instead. There are real downsides to that though:

- The Vault would necessarily reset when we add a new 4*/5*. You would have at max 1 month (a season) to pull everything you needed before it may disappear/odds reset - This isn't great for covering characters. Either the Vault is small enough that you could reasonably expect a certain character (100 covers) and probably only 3-4 covers of each character, or it contains enough to cover all the 4*s and the Vault is huge, so odds changes never happen. Then refer back to point 1 on not being able to chew through that. - Even with a Vault, you're still going to have the same issue as current: Out of a 100 piece Vault, players pulling 5 times could very well pull 3 of the 4 available Invisible Woman covers that happen to be hanging out. Chances aren't high of that happening, but it's a near surety that someone will hit that combination with the number of people making pulls. Meaning, it may not even actually change the *feeling* that things are rigged and you're getting screwed even if it does 100% objectively lower the odds of that occurring.

In regards to the more general feeling: We've looked at our odds. We've looked at drops. We continue to look at odds and drop runs. We have yet to find anything anomalous. The numbers of covers of all types from all cover packs have been and still are dropping in the displayed ratios, give or take very much within statistical variations. I totally get that it sucks that you see 3-4 covers in a row, but that's random being random. As was stated above, you never notice when you pull 12 Standards and they're all varied. You *do* notice when you pull 5 Venoms in a row though. Looking at it from a 100 foot view, that's a little streak blip in the otherwise very homogenous statistical run. Just as often, Venom doesn't pop out of 300 pack pulls in a row. But for you personally, that was signs that something is rigged. It really really does come down to being human nature to notice that. I mean, I sure as heck know better and *I* still have a hard time fighting down that feeling when Moonstone pops up 4 times on my 10x Heroic in a season pull.

All this being said, we are working on something that should obviate these feelings some/most of the time. Not ready to announce what that is yet, but all of us (and I'm sure nearly 100% of the player base) totally agree that pulling yet another cover on a finished character is a substandard experience.

I'm curious about the first point you make regarding the drawback to changing to a Vault-type structure. Why would the Vault *have* to refresh with new character additions? What would be the problem with just slipping 3 new covers into the vault? They would be out of order? I'm fairly certain that it's not a programming issue. It would not be as easy as a refresh, but there are append clauses.

Unknown

PLBIV wrote:

I'm curious about the first point you make regarding the drawback to changing to a Vault-type structure. Why would the Vault *have* to refresh with new character additions? What would be the problem with just slipping 3 new covers into the vault? They would be out of order? I'm fairly certain that it's not a programming issue. It would not be as easy as a refresh, but there are append clauses.

I can imagine a few things.

Lets say you have a 10 cover Vault, with say, in it.

Now, hey, lets add to the rotation, without resetting it!

If you just say, replace with , everybody who has already drawn Deadpool can't draw X23 without a reset.

But what if we just add X-23 to the end, and make it an 11 cover vault?

Then, well, anybody who has already emptied that vault, or has gotten close to it, has a guaranteed X-23 on their next draw!

Thus, resetting a vault would give everybody the most fair chance at getting the new covers and junk.

PLBIV

colwag wrote:

I can imagine a few things.

Lets say you have a 10 cover Vault, with say, in it.

Now, hey, lets add to the rotation, without resetting it!

If you just say, replace with , everybody who has already drawn Deadpool can't draw X23 without a reset.

But what if we just add X-23 to the end, and make it an 11 cover vault?

Then, well, anybody who has already emptied that vault, or has gotten close to it, has a guaranteed X-23 on their next draw!

Thus, resetting a vault would give everybody the most fair chance at getting the new covers and junk.

Oh, I totally understand that! But why is that a problem? Why is someone who has made their way through the vault getting a better chance at the new covers an issue? If you've been playing long enough to diligently clear out the vault, having new items placed on your shelf shouldn't be a bad thing.

Arimis_Thorn

Last night from the PVE LT I got an Invisible Woman yellow, which I already had 5 of.

This morning, I got 15 CP from the Daily Alliance Rewards which put me over 25.

I got an Invisible Woman yellow, which I already had 5 of.

There is a 1 in 72 chance for any given cover from LTs and I pull the same useless one twice in a row.

Guess I should buy a lottery ticket.

OneLastGambit

Just for edifications sake - the people who are keeping track of how many times they have received the same pulls are you also keeping track of how many times the entire userbase does not?

It can feel as though the RNG is personally skewing your odds but it is not however, because the RNG does not operate on a single user level, it works on the entire user base and therefore in order to accurately decide whether or not it is broken you need to track the entire data set and not just 9 pulls from one user.

Incidentally even IF it did function per user (not per user base) then you would still need to track every single legendary token you have ever pulled to decide if the RNG was broken... and even then you would have to track it over a sufficient enough timespan to determine true random generation (in reality this time period is infinity - however this isn't practical).

Just food for thought.

I work in the field of statistical data analysis so I'm used to looking at these kind of issues from a research stand point... and even still the game convinces me that it's picking on me sometimes when it gives me a near 75% MHawkeye pull rate on Heroic tokens

Konman

Arimis Thorn wrote:

<snip> I got 15 CP from the Daily Alliance Rewards <snip>

What "Daily Alliance Reward" gives you 15 Command Points?

Arimis_Thorn

Konman wrote:

Arimis Thorn wrote:

<snip> I got 15 CP from the Daily Alliance Rewards <snip>

What "Daily Alliance Reward" gives you 15 Command Points?

Sorry, the SHIELD Resupply you get when you win your first match of the day. Sorry, couldn't think of the right phrase when I wrote this and wasn't near my phone.

15 cp has apparently been rotated in as one of the possible rewards. This was for day 516. Not sure how often it rolls around.

Pylgrim

bobbyfish wrote:

fair enough if you say progression has stalled (I guess that's why 5* has been brought in, all discussions of that aside). I'm not suggesting there's not a valid need for something for the top rosters to aim for; both from the players' POV and a game sustainability POV that makes sense. I just think the element of randomness should still be part of the game at all levels, rather than guaranteed progression/cover selection for one level of player over another. If there is an option added for 4* transitioners to pick/reject covers, for example, then there should be a similar option for a 2* transitioner. I think the devs would need to be a bit careful about how much they improve pulls, even if 99% of this board think they should do so.

The reason why we want better odds, or a lesser impact of luck is precisely because the progression is almost nonexistant compared to progressing from 1* to 3*. For example, in a week, in average, a player can get up to around 32 certain 3* covers and as many, if no more, random chances of getting more. In that same time, only up to around 8 certain 4*s (and for most of those you need to place top 1-2 in PVE and PVP) and up to around 7 random ones (again, most of those by reaching the final progression reward in each event). So not only there are several times fewer opportunities to grab a 4*, those opportunities also come at the expense of great effort and expenditure. So when you open a 3* cover that you already have in a random token, the loss is minimal; you have dozens more to open and if you still don't get lucky you just need to wait a bit until the corresponding PVP, PVE or DDQ event that rewards exactly the covers that you need comes around. But when one of the precious, rare, hard-to acquire chances to open a 4* fizzles, it's a significant loss and a terrible feeling.

OneLastGambit wrote:

Just for edifications sake - the people who are keeping track of how many times they have received the same pulls are you also keeping track of how many times the entire userbase does not?

It can feel as though the RNG is personally skewing your odds but it is not however, because the RNG does not operate on a single user level, it works on the entire user base and therefore in order to accurately decide whether or not it is broken you need to track the entire data set and not just 9 pulls from one user.

Incidentally even IF it did function per user (not per user base) then you would still need to track every single legendary token you have ever pulled to decide if the RNG was broken... and even then you would have to track it over a sufficient enough timespan to determine true random generation (in reality this time period is infinity - however this isn't practical).

Just food for thought.

I work in the field of statistical data analysis so I'm used to looking at these kind of issues from a research stand point... and even still the game convinces me that it's picking on me sometimes when it gives me a near 75% MHawkeye pull rate on Heroic tokens

Every time you post this, every time I post this same rebuttal, and every time, apparently, you ignore it. We all (or most of us) understand luck, odds and chance. We understand "runs" and we understand statistics and probability. The problem is that the highest, rarest, harder-to-get tier of rewards should not rely on luck for prize handling (or at least minimise its impact). Yes, some people will draw the expected averages, some people will enjoy "good luck runs" and some wretched people will suffer of "terrible luck runs", and yes, that's how luck works. But NO ONE at this level of play and effort should get to feel what it is felt in a "terrible luck run". It's simply not fair, even if statistically correct. So yeah, chance works like this, there's no way to change it... except by removing chance from the equation or minimising the part it plays! That's what we are asking for here.

PLBIV

OneLastGambit wrote:

Just for edifications sake - the people who are keeping track of how many times they have received the same pulls are you also keeping track of how many times the entire userbase does not?

It can feel as though the RNG is personally skewing your odds but it is not however, because the RNG does not operate on a single user level, it works on the entire user base and therefore in order to accurately decide whether or not it is broken you need to track the entire data set and not just 9 pulls from one user.

Incidentally even IF it did function per user (not per user base) then you would still need to track every single legendary token you have ever pulled to decide if the RNG was broken... and even then you would have to track it over a sufficient enough timespan to determine true random generation (in reality this time period is infinity - however this isn't practical).

Just food for thought.

I work in the field of statistical data analysis so I'm used to looking at these kind of issues from a research stand point... and even still the game convinces me that it's picking on me sometimes when it gives me a near 75% MHawkeye pull rate on Heroic tokens

While this is true, to be truly random, an RNG should function like a roll of the dice: no memory and no bias (except, in this case, for the 4* v 5* bias). Therefore, the odds of getting 2 of the same cover in a row, as an individual and as a group, is still going to be 1 in "however many possible covers are out there adjusted for * bias". That doesn't change whether you open one token immediately after another of 4 weeks later. So it's equally cruddy. If it was keeping track of historical distribution in any way, it ceases to be an RNG. That's why anyone with a math background is going to avoid slot machines at all cost: they have a bias built in. Whether you're charting your own experience or the population's experience, the distribution will be the same over time.

And with that, I say that the Vault concept is better, if they just adjust it to not refresh with the introduction of a new char.

OneLastGambit

The reason why we want better odds, or a lesser impact of luck is precisely because the progression is almost nonexistant compared to progressing from 1* to 3*. For example, in a week, in average, a player can get up to around 32 certain 3* covers and as many, if no more, random chances of getting more. In that same time, only up to around 8 certain 4*s (and for most of those you need to place top 1-2 in PVE and PVP) and up to around 7 random ones (again, most of those by reaching the final progression reward in each event). So not only there are several times fewer opportunities to grab a 4*, those opportunities also come at the expense of great effort and expenditure. So when you open a 3* cover that you already have in a random token, the loss is minimal; you have dozens more to open and if you still don't get lucky you just need to wait a bit until the corresponding PVP, PVE or DDQ event that rewards exactly the covers that you need comes around. But when one of the precious, rare, hard-to acquire chances to open a 4* fizzles, it's a significant loss and a terrible feeling.

OneLastGambit wrote:

Just for edifications sake - the people who are keeping track of how many times they have received the same pulls are you also keeping track of how many times the entire userbase does not?

It can feel as though the RNG is personally skewing your odds but it is not however, because the RNG does not operate on a single user level, it works on the entire user base and therefore in order to accurately decide whether or not it is broken you need to track the entire data set and not just 9 pulls from one user.

Incidentally even IF it did function per user (not per user base) then you would still need to track every single legendary token you have ever pulled to decide if the RNG was broken... and even then you would have to track it over a sufficient enough timespan to determine true random generation (in reality this time period is infinity - however this isn't practical).

Just food for thought.

I work in the field of statistical data analysis so I'm used to looking at these kind of issues from a research stand point... and even still the game convinces me that it's picking on me sometimes when it gives me a near 75% MHawkeye pull rate on Heroic tokens

Every time you post this, every time I post this same rebuttal, and every time, apparently, you ignore it. We all (or most of us) understand luck, odds and chance. We understand "runs" and we understand statistics and probability. The problem is that the highest, rarest, harder-to-get tier of rewards should not rely on luck for prize handling (or at least minimise its impact). Yes, some people will draw the expected averages, some people will enjoy "good luck runs" and some wretched people will suffer of "terrible luck runs", and yes, that's how luck works. But NO ONE at this level of play and effort should get to feel what it is felt in a "terrible luck run". It's simply not fair, even if statistically correct. So yeah, chance works like this, there's no way to change it... except by removing chance from the equation or minimising the part it plays! That's what we are asking for here.

Right so the crux is then that you want an absolute certainty of obtaining the best covers in the game with the appropriate level of effort. Is that correct? You don't want any element of chance in it.

So, for example, if they made it so that you cannot obtain the best covers via token pulls and could only get it via placing first or second in the PVP end of season rankings that would be better? This would be a fair exchange of effort/reward yes? Best cover in game goes to best player in game.

You want to know why random chance is the best and fairest way? Whales. If the best covers only go to the best players guess who they are going to be? Whales. The people who already have the best maxed out rosters that nobody else can compete with. With random chance everyone has the chance to get the best covers.

I can see why people are frustrated, I honestly do. I never ignored your rebuttals, its that everytime this issue pops up the question/complaint is never "please change RNG to a non RNG system", the issue is always "Why do I only get rubbish tokens" - which is the issue that I answered.

Personally if they are to not use RNG (as you suggest) then for me the best and fairest way to obtain these covers would be to hand them out in Shield Re-supply to the people who have played longest. But as annoying as the RNG is (and trust me I find it annoying - I've pulled 5 MHawkeyes today from 3 heroic tokens and 2 event tokens) I'd rather stick with that system because it prevents the top of rankings from completely running away from the rest.

fmftint

But as annoying as the RNG is (and trust me I find it annoying - I've pulled 5 MHawkeyes today from 3 heroic tokens and 2 event tokens) I'd rather stick with that system because it prevents the top of rankings from completely running away from the rest.

Have you looked at top rankings lately? 2.5-3k single event scores.... The top is already running away from the rest

simonsez

OneLastGambit wrote:

Incidentally even IF it did function per user (not per user base) then you would still need to track every single legendary token you have ever pulled to decide if the RNG was broken... and even then you would have to track it over a sufficient enough timespan to determine true random generation (in reality this time period is infinity - however this isn't practical).

I don't agree. If someone posts the results of their legendary pulls, and a dev implies that those results were legitimate, and the probability of this outcome is somewhere around 1 in 13 million, I don't need to do any more data collection. I'm very comfortable asserting that either the odds aren't as advertised, or there's nothing random about the results of token pulls.

And I work in the field of statistical data analysis too.

OneLastGambit

fmftint wrote:

But as annoying as the RNG is (and trust me I find it annoying - I've pulled 5 MHawkeyes today from 3 heroic tokens and 2 event tokens) I'd rather stick with that system because it prevents the top of rankings from completely running away from the rest.

Have you looked at top rankings lately? 2.5-3k single event scores.... The top is already running away from the rest

Yep but imagine how much worse that would be if the best prizes only went to that guy running away? at least this way his advantage is not as significant because of two things...

A. That amazing prize he might have won could be IW
B. I could open the same type of token and get a Surfer

Thus closing the gap very slightly. Without that element of chance it would be - best person gets all the best covers everyone else has to suck it.

OneLastGambit

simonsez wrote:

OneLastGambit wrote:

Incidentally even IF it did function per user (not per user base) then you would still need to track every single legendary token you have ever pulled to decide if the RNG was broken... and even then you would have to track it over a sufficient enough timespan to determine true random generation (in reality this time period is infinity - however this isn't practical).

I don't agree. If someone posts the results of their legendary pulls, and a dev implies that those results were legitimate, and the probability of this outcome is somewhere around 1 in 13 million, I don't need to do any more data collection. I'm very comfortable asserting that either the odds aren't as advertised, or there's nothing random about the results of token pulls.

And I work in the field of statistical data analysis too.

Perhaps I mis-read it but it seemed to me like ICE was saying that the token drops are working exactly as intended and are not skewed in any way. He does suggest that to satisfy player demand they would consider a vault based system instead. The dev did not imply that anybodies token pulls were not working as they should or even that they were unfair, he only said he understands why people get annoyed at it but they would not really be willing to alter token odds to suit individual users.

If I have mis read somewhere feel free to correct me.

Mercurywolf

simonsez wrote:

OneLastGambit wrote:

Incidentally even IF it did function per user (not per user base) then you would still need to track every single legendary token you have ever pulled to decide if the RNG was broken... and even then you would have to track it over a sufficient enough timespan to determine true random generation (in reality this time period is infinity - however this isn't practical).

I don't agree. If someone posts the results of their legendary pulls, and a dev implies that those results were legitimate, and the probability of this outcome is somewhere around 1 in 13 million, I don't need to do any more data collection. I'm very comfortable asserting that either the odds aren't as advertised, or there's nothing random about the results of token pulls.

And I work in the field of statistical data analysis too.

So, you're basing a conclusion on one result out of how many? I get that a fair number of people complain about their token pulls, but at the end of the day, those people represent an incredibly small percentage of the total player base or number of tokens being pulled.

Yes, it's like being struck by lightning multiple times (the opposite of winning the lottery) the odds are close to 0, but it can still happen.

Another way to look at this is the old rule of customer service. When someone is happy, they may tell 1 or 2 people. When someone is unhappy, they won't stop telling anyone who will listen because they feel slighted.

simonsez

OneLastGambit wrote:

Perhaps I mis-read it but it seemed to me like ICE was saying that the token drops are working exactly as intended and are not skewed in any way.

He did say that. But he's also said that an unnamed player who reported statistically improbable results, was "lucky". Given the probability of what was reported, "lucky" doesn't cut it.

I don't doubt that Ice believes nothing is amiss. I just don't believe it myself, given what I've seen here and in Line.

OneLastGambit

simonsez wrote:

OneLastGambit wrote:

Perhaps I mis-read it but it seemed to me like ICE was saying that the token drops are working exactly as intended and are not skewed in any way.

He did say that. But he's also said that an unnamed player who reported statistically improbable results, was "lucky". Given the probability of what was reported, "lucky" doesn't cut it.

I don't doubt that Ice believes nothing is amiss. I just don't believe it myself, given what I've seen here and in Line.

Ah I see. I didn't read the part about the user reporting it. Although even still we should still remember that this is only one user that has reported this with empirical evidence, the rest is all anecdotal. If enough of us actually presented the devs with empirical evidence to support this claim then perhaps it would be looked into further, I think until this step is taken the devs will assume it is simply a very minor percentage of disgruntled players voicing an annoyance.

simonsez

Mercurywolf wrote:

Yes, it's like being struck by lightning multiple times (the opposite of winning the lottery) the odds are close to 0, but it can still happen.

Of course it can still happen. But if you're sitting in a room with some people, and one guy tells you that he's been hit by lightning 17 times, you'd be foolish to think that this is completely via random chance, and not because he stands under a tree during every thunderstorm.

Likewise, if someone tells you he pulled 13 5*s out of 20 tokens (fictional example), you'd be foolish to believe that this resulted from random draws, each having a 10% probability of success. Saying, "But it was only 20 tokens" is not a valid rebuttal. Probability is probability. And the probability of a 13/20 success rate on random pulls with a 10% success rate, is negligible.

simonsez

OneLastGambit wrote:

If enough of us actually presented the devs with empirical evidence to support this claim then perhaps it would be looked into further

Not going to hold my breath on that. Ice says he's fine with the data he's seen, and he has access to all the data we'd have. I just don't know what their background is re: statistical analytics. For all I know, all they've done is look at aggregated data that shows something like 100,000 token have been redeemed, and 10,000 5*s have been rewarded. But just because the aggregated hit rate is 10%, that doesn't mean it's random.