Additional additional. It’s *really* hard to build a bad random number generator these days. At very least Most systems will have a rand() system call which accesses a /dev/random I pseudodevice, which usually takes a bunch of environmental variables such as last seek time or uptime or whatever to seed a pseudorandom sequence which you’d be really hard pressed to tell from true random.
You on the other hand desperately try to justify/dismiss this pretty obvious problem. DB clearly has some problem with RNG seed.
The seed would have nothing to do with the problem you are stating. The seed is just used to start the generator.
Which suggests it can be a source of the problem.
The only time a bad seed would really come into play is if they are reseeding it with the same number and the retrieval is the same, thus making the numbers possibly predictable. (aka a slot machine reset with the same number each time. Given enough spins, you could start to predict the values the next time it was reset with the same seed). But even this is not likely an issue, since the numbers are also being distributed randomly (randomly because of "human" interference aka entropy) between x number of players all playing at the same time. This is not what you are arguing. What you seem to be arguing is the distribution of the numbers. Which like I mentioned in another post, could be affected by a number of other factors.
"If it wasn't for autocorrect, we wouldn't have Tuvok on a Giraffe."
I have the worst luck in the gauntlet! Placed 6th and then 8th to get like 5-6 crates and got nothing higher than a 3* incense. Such a waste! Their shouldn’t be anything LOWER than a 3* item or crewman if it’s a freaking CRATE!
I’ve never gotten a Guinan or Liguria so wouldn’t be surprised if I never get a Banjo man. :-/
So, the problem here is you don't have the evidence to make the assertion you are trying to make. You have a handful of screenshots and declare "this happens too frequently" with no information about how frequently it is actually happening or even specific prameters around your "close roll" effect.
Posters here posted statistical information about why your inference might be misguided.
Now, you should leave and return with additional statistical evidence to back your claim.
What exactly is this phenomenon?
How often does it occur?
Given expected random distributions, how often should it be occurring?
When you have that data, let us know. I will be happy to peer review it.
Additional additional. It’s *really* hard to build a bad random number generator these days. At very least Most systems will have a rand() system call which accesses a /dev/random I pseudodevice, which usually takes a bunch of environmental variables such as last seek time or uptime or whatever to seed a pseudorandom sequence which you’d be really hard pressed to tell from true random.
Bad maths is much more common...
Sure, it can be an algoritm problem, no argue there. It may even be a combination of both.
So, where do you draw the line at "similar"? You're claiming a 209 and 223 roll are "similar. If we're saying +/- 20 on any given roll, then, for someone like Valeris who has a 250-500 span, there is a 41/250 -> 16.4% chance that any two given rolls are "similar". That is NOT a small probability.
What exactly do you mean by +/- 20? You mean delta=40? 223-209=15 and it was biggest delta here. If you assume 250-500 roll range then delta=15 would get you only about 6% if i'm not mistaken.
Just out of curiosity, what are the odds of this scenario where you and your opponent both have the two skills and they are fairly similar (+-25 pts) but you have a 45% crit and your opponent has 5%? You get 0/6 crits and they get 3/6.
I love when you guys that actually have a way higher understanding of stats and probability explain things in a way I can comprehend it. Thanks
Just out of curiosity, what are the odds of this scenario where you and your opponent both have the two skills and they are fairly similar (+-25 pts) but you have a 45% crit and your opponent has 5%? You get 0/6 crits and they get 3/6.
I love when you guys that actually have a way higher understanding of stats and probability explain things in a way I can comprehend it. Thanks
Assuming the displayed numbers are correct, each roll still has 45% and 5%. The problem is you effectively need an infinite sample size to see this.
Let's say you flip a coin 20 times and you get heads every time. What are the odds your next flip is heads?
It is still 50%. It might be heads for the next 100 flips. But given an infinite sample set, the 50% will emerge.
Each time there are posts like this, we are seeing a very constrained sample set (usually a few handfuls).
[Edit]
This was not about calculating the odds of producing the sequential rolls.
"If it wasn't for autocorrect, we wouldn't have Tuvok on a Giraffe."
Just out of curiosity, what are the odds of this scenario where you and your opponent both have the two skills and they are fairly similar (+-25 pts) but you have a 45% crit and your opponent has 5%? You get 0/6 crits and they get 3/6.
I love when you guys that actually have a way higher understanding of stats and probability explain things in a way I can comprehend it. Thanks
You have (0.55)^6=0.0277=2.77% chance to get 0 crits in 6 rolls with 45% crit chance.
You have 20*0.95*0.95*0.95*0.05*0.05*0.05=0.0021=0.21% chance to get 3 crits in 6 rolls with 5% crit chance.
Just out of curiosity, what are the odds of this scenario where you and your opponent both have the two skills and they are fairly similar (+-25 pts) but you have a 45% crit and your opponent has 5%? You get 0/6 crits and they get 3/6.
I love when you guys that actually have a way higher understanding of stats and probability explain things in a way I can comprehend it. Thanks
1. The rolls actually have no influence on your chance to crit, so it doesn't matter if they are similar or not.
- Similar rolls just mean that the player has to choose to take a gamble on it or not. No matter what the crit rating is, if you have similar scores for your 6 rolls and you opt to go for it, it is actually really risky, because if it does go "wrong" (i.e. you lose) then you'll be left with the feeling as stated by the OP.
Why is it risky? ... because you're only making 6 rolls ... not an infinite amount ... there can be massive amounts of variance in six rolls, and only once you've done an infinite number does the crit percentage of all your rolls approach the percentage the crew has.
2. You can assume that each roll in unique ... so if you have a crit of 5%, then for each roll that is always 5% ... if you crit the first it does not mean that the second roll has less chance to crit ... it is also just 5% ... the same goes for 45% ... if you do not crit your first 3 rolls then that says nothing about if you will crit the next three ... a person will have the expectation that they will see crit rolls in the next 3, but that certainly doesn't need to be the case (and in more cases than not that is actually the case ... because 45% is less than half!).
This is therefore the main issue here ... there is an expectancy to always get 45% of the rolls as crit ... but over 6 rolls that is 2.7 crits ... and unfortunately you cannot get 0.7 rolls, either 2 or 3.
Lesson Learnt
This is also the mistake the OP has made ... he/she basically really banked on luck ... the lesson here should not be "oh ol' RNG is out to get me" but ... do not try to take on similar stat opponents and expect to win (even if you take crit chance into account, which was not the case because it was two identical crew going against each other). The OP would have lost if the opponent just crit once more than the OPs crew, and the chance of that happening is really big.
Additional additional. It’s *really* hard to build a bad random number generator these days.
In this context, bad RNG means "I predicted I would win and I lost, therefore RNG is bad."
So much this. I wish more people would actually understand statistics, probability, and confirmation bias. I have yet to see one of these 'RNG is broken' posts or arguments on this board that can't be rephrased as 'I don't understand probability, so it must DB's fault'.
I'd encourage people who have strong opinions on the matter of the gauntlet RNG to collect data. Having done so myself I found it to at least be an interesting experiment in my own negativity bias, and hopefully useful for the community. For those interested: Gauntlet Data Set: Mirror Matches
I think the mirror matches would still be worthwhile to track, though more difficult now due to all the different bonuses. On the old forums there was also a gauntlet data collection done that was not constrained to mirror matches which collected and analyzed a lot more data: Gauntlet bias data results and some discussion on the two approaches, Am I the only one who thinks the Gauntlet is working just fine?.
We are all downstream from each other and ourselves, therefore choose to be relaxed and groovy. Consider participating in civil discourse, understanding the Tardigrade, and wandering with the Subspace Eddies.
I'd encourage people who have strong opinions on the matter of the gauntlet RNG to collect data. Having done so myself I found it to at least be an interesting experiment in my own negativity bias, and hopefully useful for the community. For those interested: Gauntlet Data Set: Mirror Matches
I think the mirror matches would still be worthwhile to track, though more difficult now due to all the different bonuses. On the old forums there was also a gauntlet data collection done that was not constrained to mirror matches which collected and analyzed a lot more data: Gauntlet bias data results and some discussion on the two approaches, Am I the only one who thinks the Gauntlet is working just fine?.
I think it’s definitely working better than event shuttle percentages.
Ouch, thats a bummer. I just lost a 60 win streak to a similar result where I should clearly have won, all being equal. Still no Banjoman which is annoying as a fleetmate got him for a 6 win streak.
Jim
I’m the far end of the bell curve on Banjoman rewards...got him on my first 3-win streak after the gauntlet restart last week. I almost felt bad about it because so many people played thousands of rounds without getting Locutus and/or Guinan and it took me only three rounds to get The Caretaker...
This is all very interesting and I think I'm going to look into a class on it. The reason I asked is that I never take mirror matches. I rarely do ones that I don't know I have the best chance at winning. I'd rather just spend the merits on a refresh. Though I did decide to try one today for the red shirt approach (and just the heck of it) and not to my surprise at all, I got trounced. I just laughed. Thanks for the insight guys
I'd encourage people who have strong opinions on the matter of the gauntlet RNG to collect data. Having done so myself I found it to at least be an interesting experiment in my own negativity bias, and hopefully useful for the community. For those interested: Gauntlet Data Set: Mirror Matches
I think the mirror matches would still be worthwhile to track, though more difficult now due to all the different bonuses. On the old forums there was also a gauntlet data collection done that was not constrained to mirror matches which collected and analyzed a lot more data: Gauntlet bias data results and some discussion on the two approaches, Am I the only one who thinks the Gauntlet is working just fine?.
Gauntlet is working fine from a math based perspective. Plenty of well informed threads about mathematical randomness, chance, Gambler's fallacy and the programming of RNGs have been posted in the past.
Plenty of suggestions elsewhere for a format change though.
I'd encourage people who have strong opinions on the matter of the gauntlet RNG to collect data. Having done so myself I found it to at least be an interesting experiment in my own negativity bias, and hopefully useful for the community. For those interested: Gauntlet Data Set: Mirror Matches
I think the mirror matches would still be worthwhile to track, though more difficult now due to all the different bonuses. On the old forums there was also a gauntlet data collection done that was not constrained to mirror matches which collected and analyzed a lot more data: Gauntlet bias data results and some discussion on the two approaches, Am I the only one who thinks the Gauntlet is working just fine?.
Gauntlet is working fine from a math based perspective. Plenty of well informed threads about mathematical randomness, chance, Gambler's fallacy and the programming of RNGs have been posted in the past.
Plenty of suggestions elsewhere for a format change though.
As for an approach suggestion: Expect to lose.
I have a motto, If something smells bad it is bad. How long do you play this game? How many times a day you spin dabo wheel? How many times you hit top row reward? How many times you encountered highly improbable RNG outburst in Gauntlet?
I'd encourage people who have strong opinions on the matter of the gauntlet RNG to collect data. Having done so myself I found it to at least be an interesting experiment in my own negativity bias, and hopefully useful for the community. For those interested: Gauntlet Data Set: Mirror Matches
I think the mirror matches would still be worthwhile to track, though more difficult now due to all the different bonuses. On the old forums there was also a gauntlet data collection done that was not constrained to mirror matches which collected and analyzed a lot more data: Gauntlet bias data results and some discussion on the two approaches, Am I the only one who thinks the Gauntlet is working just fine?.
Gauntlet is working fine from a math based perspective. Plenty of well informed threads about mathematical randomness, chance, Gambler's fallacy and the programming of RNGs have been posted in the past.
Plenty of suggestions elsewhere for a format change though.
As for an approach suggestion: Expect to lose.
I have a motto, If something smells bad it is bad. How long do you play this game? How many times a day you spin dabo wheel? How many times you hit top row reward? How many times you encountered highly improbable RNG outburst in Gauntlet?
I'd encourage people who have strong opinions on the matter of the gauntlet RNG to collect data. Having done so myself I found it to at least be an interesting experiment in my own negativity bias, and hopefully useful for the community. For those interested: Gauntlet Data Set: Mirror Matches
I think the mirror matches would still be worthwhile to track, though more difficult now due to all the different bonuses. On the old forums there was also a gauntlet data collection done that was not constrained to mirror matches which collected and analyzed a lot more data: Gauntlet bias data results and some discussion on the two approaches, Am I the only one who thinks the Gauntlet is working just fine?.
Gauntlet is working fine from a math based perspective. Plenty of well informed threads about mathematical randomness, chance, Gambler's fallacy and the programming of RNGs have been posted in the past.
Plenty of suggestions elsewhere for a format change though.
As for an approach suggestion: Expect to lose.
I have a motto, If something smells bad it is bad. How long do you play this game? How many times a day you spin dabo wheel? How many times you hit top row reward? How many times you encountered highly improbable RNG outburst in Gauntlet?
Star Trek Enterprise, DR. PHLOX:
There was a time when Denobulans believed they were the only intelligent species in the galaxy. Needless to say, it was quite a surprise when the B'Saari made First Contact. Many refused to accept the truth even with the evidence standing right in front of them.
I'd encourage people who have strong opinions on the matter of the gauntlet RNG to collect data. Having done so myself I found it to at least be an interesting experiment in my own negativity bias, and hopefully useful for the community. For those interested: Gauntlet Data Set: Mirror Matches
I think the mirror matches would still be worthwhile to track, though more difficult now due to all the different bonuses. On the old forums there was also a gauntlet data collection done that was not constrained to mirror matches which collected and analyzed a lot more data: Gauntlet bias data results and some discussion on the two approaches, Am I the only one who thinks the Gauntlet is working just fine?.
Gauntlet is working fine from a math based perspective. Plenty of well informed threads about mathematical randomness, chance, Gambler's fallacy and the programming of RNGs have been posted in the past.
Plenty of suggestions elsewhere for a format change though.
As for an approach suggestion: Expect to lose.
I have a motto, If something smells bad it is bad. How long do you play this game? How many times a day you spin dabo wheel? How many times you hit top row reward? How many times you encountered highly improbable RNG outburst in Gauntlet?
Star Trek Enterprise, DR. PHLOX:
There was a time when Denobulans believed they were the only intelligent species in the galaxy. Needless to say, it was quite a surprise when the B'Saari made First Contact. Many refused to accept the truth even with the evidence standing right in front of them.
You have pointed to 3 screen shots. That is not evidence. I've provided to you suggestions on how you might make your argument stronger and less easily refutable. You are welcome to take that and strengthen your position or continue to scream into the wind. Do what you will.
I'd encourage people who have strong opinions on the matter of the gauntlet RNG to collect data. Having done so myself I found it to at least be an interesting experiment in my own negativity bias, and hopefully useful for the community. For those interested: Gauntlet Data Set: Mirror Matches
I think the mirror matches would still be worthwhile to track, though more difficult now due to all the different bonuses. On the old forums there was also a gauntlet data collection done that was not constrained to mirror matches which collected and analyzed a lot more data: Gauntlet bias data results and some discussion on the two approaches, Am I the only one who thinks the Gauntlet is working just fine?.
Gauntlet is working fine from a math based perspective. Plenty of well informed threads about mathematical randomness, chance, Gambler's fallacy and the programming of RNGs have been posted in the past.
Plenty of suggestions elsewhere for a format change though.
As for an approach suggestion: Expect to lose.
I have a motto, If something smells bad it is bad. How long do you play this game? How many times a day you spin dabo wheel? How many times you hit top row reward? How many times you encountered highly improbable RNG outburst in Gauntlet?
Star Trek Enterprise, DR. PHLOX:
There was a time when Denobulans believed they were the only intelligent species in the galaxy. Needless to say, it was quite a surprise when the B'Saari made First Contact. Many refused to accept the truth even with the evidence standing right in front of them.
You have pointed to 3 screen shots. That is not evidence. I've provided to you suggestions on how you might make your argument stronger and less easily refutable. You are welcome to take that and strengthen your position or continue to scream into the wind. Do what you will.
I have pointed out to three screenshots in this thread but you are welcome to look for other similar threads on this new and old forum. You will find more screenshots there. My position is strong enough you must be that wind you are talking about.
I'd encourage people who have strong opinions on the matter of the gauntlet RNG to collect data. Having done so myself I found it to at least be an interesting experiment in my own negativity bias, and hopefully useful for the community. For those interested: Gauntlet Data Set: Mirror Matches
I think the mirror matches would still be worthwhile to track, though more difficult now due to all the different bonuses. On the old forums there was also a gauntlet data collection done that was not constrained to mirror matches which collected and analyzed a lot more data: Gauntlet bias data results and some discussion on the two approaches, Am I the only one who thinks the Gauntlet is working just fine?.
Gauntlet is working fine from a math based perspective. Plenty of well informed threads about mathematical randomness, chance, Gambler's fallacy and the programming of RNGs have been posted in the past.
Plenty of suggestions elsewhere for a format change though.
As for an approach suggestion: Expect to lose.
I have a motto, If something smells bad it is bad. How long do you play this game? How many times a day you spin dabo wheel? How many times you hit top row reward? How many times you encountered highly improbable RNG outburst in Gauntlet?
Star Trek Enterprise, DR. PHLOX:
There was a time when Denobulans believed they were the only intelligent species in the galaxy. Needless to say, it was quite a surprise when the B'Saari made First Contact. Many refused to accept the truth even with the evidence standing right in front of them.
You have pointed to 3 screen shots. That is not evidence. I've provided to you suggestions on how you might make your argument stronger and less easily refutable. You are welcome to take that and strengthen your position or continue to scream into the wind. Do what you will.
I have pointed out to three screenshots in this thread but you are welcome to look for other similar threads on this new and old forum. You will find more screenshots there. My position is strong enough you must be that wind you are talking about.
To reiterate:
What exactly is this phenomenon? (You haven't given a precise definition)
How often does it occur? (You haven't quantified this)
Given expected random distributions, how often should it be occurring? (You haven't quantified this)
You don't have an argument until you have defined those items.
Comments
Which suggests it can be a source of the problem.
Bad maths is much more common...
The only time a bad seed would really come into play is if they are reseeding it with the same number and the retrieval is the same, thus making the numbers possibly predictable. (aka a slot machine reset with the same number each time. Given enough spins, you could start to predict the values the next time it was reset with the same seed). But even this is not likely an issue, since the numbers are also being distributed randomly (randomly because of "human" interference aka entropy) between x number of players all playing at the same time. This is not what you are arguing. What you seem to be arguing is the distribution of the numbers. Which like I mentioned in another post, could be affected by a number of other factors.
I’ve never gotten a Guinan or Liguria so wouldn’t be surprised if I never get a Banjo man. :-/
Posters here posted statistical information about why your inference might be misguided.
Now, you should leave and return with additional statistical evidence to back your claim.
What exactly is this phenomenon?
How often does it occur?
Given expected random distributions, how often should it be occurring?
When you have that data, let us know. I will be happy to peer review it.
Sure, it can be an algoritm problem, no argue there. It may even be a combination of both.
In this context, bad RNG means "I predicted I would win and I lost, therefore RNG is bad."
What exactly do you mean by +/- 20? You mean delta=40? 223-209=15 and it was biggest delta here. If you assume 250-500 roll range then delta=15 would get you only about 6% if i'm not mistaken.
I love when you guys that actually have a way higher understanding of stats and probability explain things in a way I can comprehend it. Thanks
Assuming the displayed numbers are correct, each roll still has 45% and 5%. The problem is you effectively need an infinite sample size to see this.
Let's say you flip a coin 20 times and you get heads every time. What are the odds your next flip is heads?
It is still 50%. It might be heads for the next 100 flips. But given an infinite sample set, the 50% will emerge.
Each time there are posts like this, we are seeing a very constrained sample set (usually a few handfuls).
[Edit]
This was not about calculating the odds of producing the sequential rolls.
You have (0.55)^6=0.0277=2.77% chance to get 0 crits in 6 rolls with 45% crit chance.
You have 20*0.95*0.95*0.95*0.05*0.05*0.05=0.0021=0.21% chance to get 3 crits in 6 rolls with 5% crit chance.
1. The rolls actually have no influence on your chance to crit, so it doesn't matter if they are similar or not.
- Similar rolls just mean that the player has to choose to take a gamble on it or not. No matter what the crit rating is, if you have similar scores for your 6 rolls and you opt to go for it, it is actually really risky, because if it does go "wrong" (i.e. you lose) then you'll be left with the feeling as stated by the OP.
Why is it risky? ... because you're only making 6 rolls ... not an infinite amount ... there can be massive amounts of variance in six rolls, and only once you've done an infinite number does the crit percentage of all your rolls approach the percentage the crew has.
2. You can assume that each roll in unique ... so if you have a crit of 5%, then for each roll that is always 5% ... if you crit the first it does not mean that the second roll has less chance to crit ... it is also just 5% ... the same goes for 45% ... if you do not crit your first 3 rolls then that says nothing about if you will crit the next three ... a person will have the expectation that they will see crit rolls in the next 3, but that certainly doesn't need to be the case (and in more cases than not that is actually the case ... because 45% is less than half!).
This is therefore the main issue here ... there is an expectancy to always get 45% of the rolls as crit ... but over 6 rolls that is 2.7 crits ... and unfortunately you cannot get 0.7 rolls, either 2 or 3.
Lesson Learnt
This is also the mistake the OP has made ... he/she basically really banked on luck ... the lesson here should not be "oh ol' RNG is out to get me" but ... do not try to take on similar stat opponents and expect to win (even if you take crit chance into account, which was not the case because it was two identical crew going against each other). The OP would have lost if the opponent just crit once more than the OPs crew, and the chance of that happening is really big.
So much this. I wish more people would actually understand statistics, probability, and confirmation bias. I have yet to see one of these 'RNG is broken' posts or arguments on this board that can't be rephrased as 'I don't understand probability, so it must DB's fault'.
I think the mirror matches would still be worthwhile to track, though more difficult now due to all the different bonuses. On the old forums there was also a gauntlet data collection done that was not constrained to mirror matches which collected and analyzed a lot more data: Gauntlet bias data results and some discussion on the two approaches, Am I the only one who thinks the Gauntlet is working just fine?.
Consider participating in civil discourse, understanding the Tardigrade, and wandering with the Subspace Eddies.
I think it’s definitely working better than event shuttle percentages.
I’m the far end of the bell curve on Banjoman rewards...got him on my first 3-win streak after the gauntlet restart last week. I almost felt bad about it because so many people played thousands of rounds without getting Locutus and/or Guinan and it took me only three rounds to get The Caretaker...
Gauntlet is working fine from a math based perspective. Plenty of well informed threads about mathematical randomness, chance, Gambler's fallacy and the programming of RNGs have been posted in the past.
Plenty of suggestions elsewhere for a format change though.
As for an approach suggestion: Expect to lose.
I have a motto, If something smells bad it is bad. How long do you play this game? How many times a day you spin dabo wheel? How many times you hit top row reward? How many times you encountered highly improbable RNG outburst in Gauntlet?
https://en.wikipedia.org/wiki/Confirmation_bias
Star Trek Enterprise, DR. PHLOX:
There was a time when Denobulans believed they were the only intelligent species in the galaxy. Needless to say, it was quite a surprise when the B'Saari made First Contact. Many refused to accept the truth even with the evidence standing right in front of them.
You have pointed to 3 screen shots. That is not evidence. I've provided to you suggestions on how you might make your argument stronger and less easily refutable. You are welcome to take that and strengthen your position or continue to scream into the wind. Do what you will.
I have pointed out to three screenshots in this thread but you are welcome to look for other similar threads on this new and old forum. You will find more screenshots there. My position is strong enough you must be that wind you are talking about.
To reiterate:
What exactly is this phenomenon? (You haven't given a precise definition)
How often does it occur? (You haven't quantified this)
Given expected random distributions, how often should it be occurring? (You haven't quantified this)
You don't have an argument until you have defined those items.