Please be truthful with us...

Quote:

---

Originally Posted by MistakeNot

Any good RNG will have clumps/streaks. Otherwise it is not a good RNG.

---

To expand on this, RNG without clumps/streaks would be bad RNG because the lack of streaks would mean that the system of RNG is taking into account previous attempts and augmenting your next attempt in order to keep the results constantly close to the displayed amount. In other words, your chance of success will increase if you had bad luck and decrease if you had good luck. This adds a pattern to the RNG, thus making it less random and a bad system of RNG.

In a good system of RNG, the chances of every single gather attempt is taken in a bubble. Your success and fail rate history don't matter. Your previous gathering attempts don't change the chance of success of your next attempt. No matter if you've been having a good luck streak or a bad luck streak, the chance of your next attempt is still the displayed value. This "each attempt's chance is taken in a bubble" is what will result in streaks, and it's a sign of properly implemented RNG.

Edit: Note that what I mean by "good" and "bad" RNG is how true to being random it is. I'm not saying it's a bad thing to implement an augmented RNG system, only that it's "bad" in that it's not actually random.

I do know that WoW added some bad luck streak protection, but FFXIV has done no such thing, opting for more pure pRNG.

Quote:

---

Originally Posted by Vandril

To expand on this, RNG without clumps/streaks would be bad RNG because the lack of streaks would mean that the system of RNG is taking into account previous attempts and augmenting your next attempt in order to keep the results constantly close to the displayed amount. In other words, your chance of success will increase if you had bad luck and decrease if you had good luck. This adds a pattern to the RNG, thus making it less random and a bad system of RNG.

---

A non-gathering example, during the first ATMA relic phase, there much hue and cry for the RNG system to be changed to one like this and SE stuck to the pure RNG from the beginning on everything.
Know the pain, endure the pain, embrace the pain.

Quote:

---

Originally Posted by Vandril

To expand on this, RNG without clumps/streaks would be bad RNG because the lack of streaks would mean that the system of RNG is taking into account previous attempts and augmenting your next attempt in order to keep the results constantly close to the displayed amount.

---

On the subject of this, clumps/streaks happen in real life too. The gambler's/monte carlo fallacy has been brought up in threads about this before; a really good example of streaking comes from the exact situation that gave it the name "monte carlo" from the casino where it happened: the ball fell on black 26 times in a row, and people lost millions betting red, assuming that red HAD to happen after that.

For a more mundane example, I see it in my job (I'm a tech support rep at a call center - glamorous, i know); I would say that of the 30 or so calls I worked on today, 15 of those were simply people calling in to make an appointment to work with someone in person, and nearly all of those were in the morning. I had 3 calls in a row where someone refused to provide me with an e-mail address, a situation that rarely happens. Everything has to be taken in a bubble if you want it to be as realistic as possible.

You don't have to believe in RNGesus, RNGesus believes in you.

Quote:

---

Originally Posted by MistakeNot

Any good RNG will have clumps/streaks. Otherwise it is not a good RNG.

---

"Good RNG will have clumps" does not adequately explain why I have seen both events of [failed a 98% 5 times in a row] (0.02^5=0.000000032% chance of occurring) and [failed a 75% chance 20 times in a row] (0.25^20=0.0000000000009094947% chance of occurring). Especially that second one (occurred while trying to do the first time reveal of one of the logs in HW) is just such a low chance of that set occurring that it makes me have trouble believing that FFXIV's RNG is really as pure as they want us to believe.

Quote:

---

Originally Posted by Elamys

On the subject of this, clumps/streaks happen in real life too. The gambler's/monte carlo fallacy has been brought up in threads about this before; a really good example of streaking comes from the exact situation that gave it the name "monte carlo" from the casino where it happened: the ball fell on black 26 times in a row, and people lost millions betting red, assuming that red HAD to happen after that.

---

This is why to this day many(most?) casino's will put up a history board at a roulette table showing the results of the last 10-20 spins. When a random streak occurs the people who dont understand probability will start placing more and bigger bets on the other side because "red is due to hit" or whatever. They win/lose at the same rate they do on every other spin. And since the odds are the same on every spin, more and bigger bets means the house takes in more money.

I have had this issue since the launch of 2.0. Yes, white knights will champion the cause that the RNG is fine. But it simply isn't. They like to regurgitate abstract non-sense as if theory, alone, answers for everything. But you know what.... I have yet to see anyone explain my repeated counter to their abstract garbage:

How can they explain the clear lack of the opposite possibility? Yes, RNG can spike low and spike high. No one is arguing that. But if it consistently spikes low (which it does! Missing between one and two times out of ever four with 90%, for example), why does it never spike high?

When I uncover items at a new node, I only have a base chance of 25%. I'll go through five to six nodes before I uncover the item. That's 0/20-24. That is hardly a result of 25%. If the white knights are to be believed, then why have I never seen myself uncover new items 4/4 times? If it spikes low, why does it never spike high?

This same issue is present in crafting. I have three different insanely dedicated crafters in my FC. They all laugh at me when I ask why I always fail to HQ an item at 97%. "If it's not 100%, it's 0%. Just accept it." They have all witnessed the same issue with their endless experience in leveling all of their gathering and crafting classes. Sure, technically and mathematically, with a 97% success, I'm bound to get a failure now and again. But not at the rate at which I fail, in practical application. And yet, I have never seen a successful HQ when I only had a 3% chance. When I asked my known crafting associates if they ever had, they all laughed me right off the internet. They have never seen that happen to them.

When these vomit-spewing RNG white knights can adequately answer that, I'll reconsider their claim that the RNG is correct in this game.

Anecdotal evidence is merely that : anecdotal.

If you want a good anecdote, how's this for you : quickly manually synthing refined materials for my weaver, I HAVE had 1% HQ (no Quality added, just Progress) come out as HQ rather than NQ. I've also had bad luck with Hasty Touch and only come out with 19% chance, or 10% chance, and still get HQ.

Or how about this one for gathering : on a +1 Gathering Attempt node, with 14% chance of HQ, getting 4 of the 5 swings to be HQ.

Or how about having a 6% chance of the fourth overmeld on crafting gear, and getting it first try?
Or how about having a 35% chance on the first overmeld on crafting gear, and taking 57 materia before one finally sticks?

What I am saying is - your sample sizes are WAY too small to mean anything. Come back and complain when you miss 25% things more than 250,000 times out of 1,000,000.

Flip a coin. Its got a 50/50 shot at being Heads or Tails, right? If you flip a coin 2 times and come up 2 Heads and 0 Tails, or 2 Tails and 0 Heads, would you call that "bad RNG"? Would you call it "good RNG"? Or would you understand that merely flipping it twice is two separate calls to the RNG and can go either way regardless of any previous outcome?

When you flip a coin, you are NOT guaranteed to have an exact, 50/50 split between Heads and Tails. If you were, it would not truly be "Random" would it? Because guaranteeing an exact 50/50 split would mean if you had 2 Heads and 1 Tails after 3 flips, your fourth would have to be Tails. Yet, you can easily get streaks of 4-6 in a row.

Now, imagine that you are guaranteed that 50/50 split, but you are one of a thousand people all grabbing for that coin for their own personal coin flips. Are you guaranteed, given that you have limited control over how soon you can regain control of the coin for another flip, to have a personal 50/50 split? Or could the coin come up Heads 100% of the time for half the people and Tails for the other half? The COIN is still giving its 50/50 split, but none of the thousand people are seeing it under those circumstances, are they?

Final question : you DO understand that the RNG in any video game is a lot more complex, and in any MMO gets called upon by a lot more than merely 1000 people, right? Your view of the situation, even that of your circle of 100 friends, is extremely limited. For another analogy, you literally cannot see the forest for the trees. You see the results you personally get. You hear about the results that others get. But you cannot, I cannot, no one can see ALL the results of ALL the RNG calls, even for a certain action such as discovering an Unknown item on a gathering node.

Oh, for what its worth, the specific questions you asked are the kinds of things that are answered either anecdotally, or through that same theorycrafting you dismiss.

Quote:

---

Originally Posted by Roth_Trailfinder

What I am saying is - your sample sizes are WAY too small to mean anything. Come back and complain when you miss 25% things more than 250,000 times out of 1,000,000.

---

Don't be ridiculous, that's an absurd requirement. Depending on how tight you want the error bars to be, the quality of an RNG can be determined with orders of magnitude fewer samples. For example, a mere 665 carefully collected samples was enough to establish the rate of G/E procs for L58 mats to within +/-2.7% with 93% confidence. (http://forum.square-enix.com/ffxiv/t...ar-L60-recipes)

So please don't make statements about things you don't understand. You're not going to convince anyone that the FFXIV RNG is a good one by demanding impossible numbers of samples for proof. Instead, you should encourage them to carefully collect a few thousand samples and share the results. Then we'll have some good scientific data to discuss, instead of dueling beliefs.

For example, if someone wants to craft a large number of items to 50% HQ chance and record how many come out HQ, whether the actual odds are near 50% (say, between 48 and 52%: close enough, I think) can be determined with 95% confidence by a mere 2500 craftings. Alternatively, you could determine if the true odds are in the 47 to 53% range with a 95% confidence with under a thousand craftings -- the less tight a bound you want and the less confidence you require, the fewer samples you need.

Quote:

---

Originally Posted by Roth_Trailfinder

You see the results you personally get. You hear about the results that others get. But you cannot, I cannot, no one can see ALL the results of ALL the RNG calls

---

If the RNG is worth a damn, this critique is irrelevant. Any valid sample of the RNG, any at all, should have the same statistical qualities as any other sample or of all other samples of the same size. Not the same results, of course, but the same statistical qualities. That's basic statistics.

Quote:

---

Originally Posted by Ramath

They like to regurgitate abstract non-sense as if theory, alone, answers for everything.

---

It is theory, but it isn't abstract. It's as straightforward as statistics gets without delving into the actual math. The explanations given are how RNG works. I couldn't care less about whiteknighting for SE by coming to their defense, because I couldn't care less about what you or others feel about the company. The fact is that the statistics gathered in sufficient quantities in the past all point toward there being no quantifiable issue with the systems of RNG in use. Over a statistically significant quantity of attempts, the results trend very closely to the displayed chance.

Quote:

---

Originally Posted by Ramath

How can they explain the clear lack of the opposite possibility? Yes, RNG can spike low and spike high. No one is arguing that. But if it consistently spikes low (which it does! Missing between one and two times out of ever four with 90%, for example), why does it never spike high?

When I uncover items at a new node, I only have a base chance of 25%. I'll go through five to six nodes before I uncover the item. That's 0/20-24. That is hardly a result of 25%. If the white knights are to be believed, then why have I never seen myself uncover new items 4/4 times? If it spikes low, why does it never spike high?

---

Fact: it does also spike high. If a system of RNG were only spiking low then the overall result of the RNG would not end up close to the displayed value, yet every test that records statistically significant numbers of attempts shows that it does. Since you end up having the chance equal to the displayed value over a very large number of gathering attempts AND you do get trends of failures, there MUST also be trends of success to accomplish this. So, assuming you believe what you're saying, a few scenarios are possible:

1: Your personal low spikes really haven't been countered so far, and your current results are lower than the displayed value. You're having bad luck. Eventually, you are statistically likely to run into a similar chain of good luck, but you simply haven't yet. It's theoretically possible to never get that high spike to counter the low spike no matter how many attempts you make, but that's possible in the same way that flipping a coin an infinite number of times and always getting heads is possible.

2: Your personal low spikes are being countered, but you just haven't noticed. As I've said before, the human brain is notoriously bad at intuitively grasping probability. People also tend to remember either negative or positive events more than the other, rather than in equal measure. It's more than likely that your brain and memory are working against you in how you perceive your RNG. Alternatively, even if your brain is an enigma that has a perfect intuitive grasp on probability, you may still not have noticed the high spikes if they weren't pronounced enough. You can have extremely strong low spikes, and then otherwise experience a very mild, constant high "spike". For example, you could fail at 50% chance 10 times in a row, then succeed at 50% chance 51% of the time over the next 200 attempts. Both the scenario of bad luck and good luck are "spikes", but the bad luck in this case is far, far more pronounced, and that's the one you'll notice and remember. You're much less likely to even notice that spike of good luck because it's so spread out and doesn't stand out.

Quote:

---

Originally Posted by Ramath

This same issue is present in crafting. I have three different insanely dedicated crafters in my FC. They all laugh at me when I ask why I always fail to HQ an item at 97%. "If it's not 100%, it's 0%. Just accept it."

---

These same realities apply to crafting. The RNG is broken in neither. That quote from your FC crafters has been a crafter joke in FFXIV for a long time, but all it's saying is that it's possible to fail unless it's 100% and that you'll just have to accept what you get. Don't take the joke too seriously.

Quote:

---

Originally Posted by Ramath

When these vomit-spewing RNG white knights can adequately answer that, I'll reconsider their claim that the RNG is correct in this game.

---

Start reconsidering, then.

Also, it would be in your best interest to stop with the "white knight" nonsense, as your reliance on slander belies your intentions.

Oh! As an aside:

Quote:

---

Originally Posted by Ramath

And yet, I have never seen a successful HQ when I only had a 3% chance. When I asked my known crafting associates if they ever had, they all laughed me right off the internet. They have never seen that happen to them.

---

I've gotten HQ without increasing quality at all (1% chance) quite a few times. :P

Hell, I've gotten HQ at all sorts of percentages. 7% is actually one of my more memorable "failures turned wonders". I royally screwed up the rotation (pressed a macro by accident) and didn't have Reclaim slotted. Ended up with 7% Quality, and got an HQ out of it. It was one of those 2-star level 60 recipes for a piece of NIN equipment back in early Heavensward, iirc.

This doesn't prove anything, of course, but it's amusing to think about.

Quote:

---

Originally Posted by Silverbane

Don't be ridiculous, that's an absurd requirement. Depending on how tight you want the error bars to be, the quality of an RNG can be determined with orders of magnitude fewer samples. For example, a mere 665 carefully collected samples was enough to establish the rate of G/E procs for L58 mats to within +/-2.7% with 93% confidence.

---

You're right. Statistics are not my strong point, I only took one class of them in college almost 20 years ago and haven't really used them since. One thing I know though - the larger the sample size, the better the results.

HOWEVER - you said something important here. "Carefully collected". That does not describe ANYTHING stated by anyone in this thread, on either side of the discussion, regarding the RNG. Its all anecdotal here.

Quote:

---

Originally Posted by Roth_Trailfinder

You're right. Statistics are not my strong point, I only took one class of them in college almost 20 years ago and haven't really used them since. One thing I know though - the larger the sample size, the better the results.

---

And yet you posted:

Quote:

---

Originally Posted by Roth_Trailfinder

What I am saying is - your sample sizes are WAY too small to mean anything.

---

That is dead wrong.
Even someone with the slightest knowledge of statistics should recall that small samples have meaning.
Actual, useful, meaning, but with large margins of error and/or low confidence (take your pick).

Flip a coin even once, and if it comes up tails, you know that it's not a two-headed coin, with 100% certainty, based on a sample size of one. What you don't know is whether it's a two-tailed coin.

The plural of anecdote is not data. But even one observation of a 98% chance failing, or a 1% chance succeeding, tells us something.
It tells us the dice are being rolled. The RNG isn't stuck at all zeroes or all ones (which can happen with defective HW or SW RNGs). The DM, rolling the dice behind the screen, may be cheating, but he or she is at least trying to hide that fact if they are.

Quote:

---

Originally Posted by Roth_Trailfinder

HOWEVER - you said something important here.

---

I said a lot of important things, all of them based on actual knowledge of what I was talking about.
But thank you for your patronizing commentary. With your admitted and demonstrated expertise in the subject matter, you must realize how much I value it.

And now a quick note based on my 40+ years of programming computers.

Cyrypto-quality random number generation, the kind that, say, Google would use to create root key pairs for their crypto-based certification service, is very very hard. Any substantial variation from perfection makes the keys easier to crack. Amateurs need not apply.

But game-quality psuedo-RNG is dead simple. Knuth published a simple and adequate algorithm back in the 1960's, if you insist on rolling your own, but why bother? Every major programming language I know provides an adequate and dirt-simple to use pRNG function either built-in or as part of a standard library.

Could SE have screwed it up? Of course, there's nothing you can't screw up in software. But the odds that they've accidentally screwed it up in such a way that a reported 89% chance isn't an 89% chance is very low, and I don't see a reason why they would do it intentionally.

I've got a good system on the %'s.

95%+ gg.
90%-94% you might fail half of them.
<89%? gg you scrub, try again later

Quote:

---

Originally Posted by Silverbane

Every major programming language I know provides an adequate and dirt-simple to use pRNG function either built-in or as part of a standard library.
Could SE have screwed it up? Of course, there's nothing you can't screw up in software. But the odds that they've accidentally screwed it up in such a way that a reported 89% chance isn't an 89% chance is very low, and I don't see a reason why they would do it intentionally.

---

They could be doing something a silly as reseeding the RNG function based on seconds every time you strike. This greatly skews the true RNG nature of those functions. I am pretty sure they did not make such a rookie mistake though.

What I've learned about SE's % system:

under 50% = ~10%
50-89% = ~30%
90-99% = ~65%
100% = 100%


Also, after reading some additional posts --- @ Vandril, your posts are full of assumptions and false equivalencies and thus make no actual sense and follow no actual logic.

Quote:

---

Originally Posted by Ken88

What I've learned about SE's % system:

under 50% = ~10%
50-89% = ~30%
90-99% = ~65%
100% = 100%


Also, after reading some additional posts --- @ Vandril, your posts are full of assumptions and false equivalencies and thus make no actual sense and follow no actual logic.

---

We're glad you've figured out SE's system to be some conspiracy of lies.

I find that people who complain about the RNG being grossly inaccurate are inevitably those who are fairly new to gathering/crafting and have not been doing so for very long. Every single time somebody gathers hard data, the numbers match! How can you possibly still say it's not correct?

Quote:

---

Originally Posted by Ken88

Also, after reading some additional posts --- @ Vandril, your posts are full of assumptions and false equivalencies and thus make no actual sense and follow no actual logic.

---

You dug up this post ignored a Dev's response only to throw shade at another player? What is that about?

Quote:

---

Originally Posted by Elamys

We're glad you've figured out...

---

Of course someone you don't agree with is new and lacks experience! Also can I see all these matching data records, like a link or something... because I've seen several and none of them match. ( I know you can't and don't worry about it, I won't be checking back here anyways )

Quote:

---

Originally Posted by QT_Melon

You dug up this post...

---

Dug up? I wasn't aware that the first page was digging lol... and every post you've made on this thread has been a passive aggressive jab at people you disagree with. And yeah I throw a little shade, but only because like you and Elamys, he degrades others he doesn't agree with.

You three breed toxic communities.

Quote:

---

Originally Posted by Ken88

And yeah I throw a little shade, but only because like you and Elamys, he degrades others he doesn't agree with.

You three breed toxic communities.

---

At no point in my posts did I degrade anyone. I don't care if you agree with me on whatever or not, but don't involve me in your feeling defensive over being accused of something.

Quote:

---

Originally Posted by Ken88

You three breed toxic communities.

---

Each person has been rather bland in their statements except to people who haven't read the statement from the developer. The refuting statements have been extremely mild. If that's being toxic no one should disagree, ever.

It's more toxic however to single out a player who is simply backing up the relative statements from the developers in a thread that has been dead for a month.

FFXIV does not follow normal logic...... In FFXIV if it's not 100% then it might as well be 1%.

Love it or Hate it.... :)

Sounds like someone doesn't know how statistics work~!

Haha, wow, this thread had been dead for like a month, and this joker comes here and pulls it up just to insult us, yet we're the toxic ones. Nice. Let's completely ignore the guy with the superiority complex, though.