Game's RNG seems whack

I don't mean that it generates long term averages that aren't statistically close to the average chances stated. Over a very long time of collecting and crafting, I've not seen the long term averages deviate to any great extent from the stated chances of success/failure.

No, what seems whack about it is the exceptionally long strings of extreme values that occur from time to time. Now, it's an RNG, as we all know, ANYTHING is possible, but in the last week I've been hit with chains of failures (more noticeable because I care about them more) that start to call into question the nature of the RNG at work.


I've had, on 3 separate occasions, chains of fails that statistically have less than 1 in a million chance of occurring, yet they occurred, three times, over a sample size of probably about 3000 events. This is not just I, many in my FC report similar things, that the long term averages seem okay, it's just the exceptionally long chains of extreme unlikeliness, that hit semi-frequently.

I'd like to see some discussion on this. I know it's an issue that's annoying a lot of people, and no, I haven't gathered explicit information, I just stop and calculate odds from time to time, and something does seem awry.

I can tell you that I've done Garuda HM over 80 times and I've only seen a bow drop once. Not an exaggeration one bit. I have however seen a plethora of SCH books...back to back to back. I'd be curious to know how SE programmed RNG for anything that uses it to determine drop or success rates etc.

I, along with many others, share your pain. The fact that consistent fails occur in large doses to pretty much everyone (which after is back to what should be higher amounts of success), is a bit of a red flag to something being wrong with the algorithm of the system. But maybe that's intentional (as there's never been any mention on it), or maybe that's how "random" works? /shrug Never figured random to be predictable though.

Quote:

---

Originally Posted by EinSoFZ

I can tell you that I've done Garuda HM over 80 times and I've only seen a bow drop once. Not an exaggeration one bit. I have however seen a plethora of SCH books...back to back to back. I'd be curious to know how SE programmed RNG for anything that uses it to determine drop or success rates etc.

---

Loot drops should be considered to run on a different system IMO. You can control success with crafting/gathering, you can't control loot drops (yet... THF? purty please!). Meaning the system has to adjust itself and rely on manipulable variables before it applies RNG. Plus, loot usually won't be 80-99% drop chance, whereas crafters/gatherers frequently find 80-99% success chance chain fail 4-8 times in a row. I have firsthand experience on that 8x fail with 95% success.

I've noticed this as well. I have both strings of failures and strings of success - but it seems a rare event that failure and success is evenly distributed.

There are tons of instances of 'broken math' in this game. The other day gathering I decided to log each attempt and do the math, at 72% success rate and 1000 attempts you would expect around 720 items gathered ... was 513 thats 51.3% way off from 72%, it gets worse when I do the math for my HQ rates.

Quote:

---

Originally Posted by Welsper59

I, along with many others, share your pain. The fact that consistent fails occur in large doses to pretty much everyone (which after is back to what should be higher amounts of success),

---

One example I had was 8 fails out of 9, on Hasty Touch, with Steady Hand 2 up (80% success rate). There's a 0.000256% chance to have 8 or more fails out of 9 tries.
On another long crafting attempt, I had 9 SH2/HT fails out 13 tries, which in itself there is a 0.0000279% chance to have 9 or more fails out of 13.
Or just today, mining, 7 fails in a row at 73% chance, which is a 0.01% chance of occurring.

I've also had 8xSH2/HT fails out of 12. And 6 fails out of 11 occur, and 5 x SH2/HT fails in a row.

An RNG is such that the any one or two of those events occurring is to be expected, but to have all 6, over the period of a week, says to me that something is wrong.

Quote:

---

Originally Posted by Castillan

One example I had was 8 fails out of 9, on Hasty Touch, with Steady Hand 2 up (80% success rate). There's a 0.000256% chance to have 8 or more fails out of 9 tries.
On another long crafting attempt, I had 9 SH2/HT fails out 13 tries, which in itself there is a 0.0000279% chance to have 9 or more fails out of 13.
Or just today, mining, 7 fails in a row at 73% chance, which is a 0.01% chance of occurring.

I've also had 8xSH2/HT fails out of 12. And 6 fails out of 11 occur, and 5 x SH2/HT fails in a row.

An RNG is such that the any one or two of those events occurring is to be expected, but to have all 6, over the period of a week, says to me that something is wrong.

---

Everyone put on your tinfoil hats!!!

Quote:

---

Originally Posted by DSN

There are tons of instances of 'broken math' in this game. The other day gathering I decided to log each attempt and do the math, at 72% success rate and 1000 attempts you would expect around 720 items gathered ... was 513 thats 51.3% way off from 72%, it gets worse when I do the math for my HQ rates.

---

You do know just because it 72% success rate doesn't mean that you will get 720/1000 right? It's a 72% chance each time you make a swing. But, I do agree, RNG can be rough in this game. lol

Quote:

---

Originally Posted by DSN

There are tons of instances of 'broken math' in this game. The other day gathering I decided to log each attempt and do the math, at 72% success rate and 1000 attempts you would expect around 720 items gathered ... was 513 thats 51.3% way off from 72%, it gets worse when I do the math for my HQ rates.

---

I honestly think HQ rates run on a different algorithm from general success percentage, at least when dealing with HQ gathering. I'm assuming it's normal to frequently HQ lower level gathering nodes, as the one time I did that, I had what seemed to be ~50% HQ rate, despite it saying 15%. So if level/gear really does matter that much, despite what game statistics are visibly shown to us, that means there are additional factors besides baseline RNG.

Failures seem to run on a different formula though, as regardless of how low a % of failure it might be, it will always happen at higher rates than a low % of success. Give it a try, see how often you'll encounter a 5-10% fail, compared to a 5-10% success. Not exactly something you can do with gathering though, I think, but rather something you can do with crafting (or comparing percentages between both). It's still probably an RNG thing mind you, but there just might be additional RNG factors in place.

Quote:

---

Originally Posted by Welsper59

I honestly think HQ rates run on a different algorithm from general success percentage, at least when dealing with HQ gathering.

---

This is definitely the case. Someone out there wrote the algorithm so that the actual percentage chance is 15% lower than what is written, but it says that the success rate is normal.

#RNGConspiracy

Quote:

---

Originally Posted by Bizniztyme

You do know just because it 72% success rate doesn't mean that you will get 720/1000 right? It's a 72% chance each time you make a swing. But, I do agree, RNG can be rough in this game. lol

---

While that's true, over a large enough sample size it should regress back to the 72% success rate. If you were to double DSN's sample size (original being 513/1000 or 51.3%), that would mean there would have to be 927 successes out of the next 1000 (92.7%) to get back to the proper success rate (1440/2000 or 72%). Even if you double that sample size to 4000, DSN would have had to get 2367 successes out of the next 3000 (79%) to get back to 72%.

Further, if you evaluate the 207 extra fails from the original use case in a vacuum, the odds of that many fails at a 72% success rate are astronomically high. So high in fact that you have a better chance of winning the PowerBall lottery (1 in 175,000,000) 13 times in a row. [Disclaimer: While I did do the math to get the odds, I'm not statistically versed enough to determine if I incorrectly evaluated the 207 fails]

The point is, it doesn't seem like the math is right. It could be either the algorithm to determine success or fail or the way the game calculates the stated success rate to you.

I agree with the OP. For some time now I've been thinking that this game has one of the worst RNG I've ever seen.

RNG is put into games to piss people off and make them keep trying to do stuff. So the longer it takes the more you play = more money for the company.

RNG will never be of any good to player's.

Tin-foil hats maybe. But ever thought maybe this is something to do with how the random values are generated?

Ask yourself "what is random, how can I make a random number?" - oddly it's not an easy as it sounds, even a thrown dice is seeded from something (the trajectory of the throw, to momentum and torque, the initial position of the die, the surface it lands upon etc). All these little factors can be traced back to a non-random source, a careful dice thrower can in fact weight the outcome in his favour. A computer is no different, it cannot generate a "truly" random number, the random value is seeded from somewhere, whether pulling values out of memory, looking at the system clock, grabbing pixels off a photograph of yoshi-p, whatever.

There is always the potential the random source is less random than might've been intended and creates unintended patterns in it's results.

Stochastic is stochastic.

I along with multiple FC mates have experianced the Primal "books" being a dime a dozen as well. It does feel like the "Random" aspect of it isn't quite so random.

RNG is RNG

It is why people are still playing

just deal with it

Yeah I've failed a 95% chance in gathering 8 times in a row before. During a leve of course. -.- It got to the point I don't even bother trying to gather without 100% success because even 99% is a waste. But I know I'm Murphy's Law's bitch so I never really gave it much clout until I saw the amount of topics on forums around the internet.

I don't have a recorded sample size but I've lost count of how many times Hasty Touch + Steady Hand II has failed. 20% shouldn't happen that much. I was almost convinced that it was a legitimate bug that Hasty Touch wouldn't work on the first attempt, it failed so many times in a row.

Quote:

---

Originally Posted by Rhindas

I don't have a recorded sample size but I've lost count of how many times Hasty Touch + Steady Hand II has failed. 20% shouldn't happen that much. I was almost convinced that it was a legitimate bug that Hasty Touch wouldn't work on the first attempt, it failed so many times in a row.

---

Here's a thing, ever noticed how it seems to be WAY worse on 40 Dura crafts? On 60/70/80 crafts, HT+SH2 seems fine, but on 40 Dura crafts, it's almost as if the 30% bonus from SH2 is multiplicative rather than additive (ie. 50 * 1.3 => 65%, rather than 80%).

Quote:

---

Originally Posted by RyuujinZERO

Tin-foil hats maybe. But ever thought maybe this is something to do with how the random values are generated?

Ask yourself "what is random, how can I make a random number?" - oddly it's not an easy as it sounds, even a thrown dice is seeded from something (the trajectory of the throw, to momentum and torque, the initial position of the die, the surface it lands upon etc). All these little factors can be traced back to a non-random source, a careful dice thrower can in fact weight the outcome in his favour. A computer is no different, it cannot generate a "truly" random number, the random value is seeded from somewhere, whether pulling values out of memory, looking at the system clock, grabbing pixels off a photograph of yoshi-p, whatever.

There is always the potential the random source is less random than might've been intended and creates unintended patterns in it's results.

---

Precisely. I've worked in IT for 25 years. In my time, I've seen both good RNGs (the Unix/Linux one) and bad ones (one that just added a prime number incrementally and kept taking the modulo of the result by another prime number to generate a percentile chance, or one that just replayed and cycled a pre-recorded set of numbers). A computerised RNG is wholly dependent upon the underlying algorithm used.

When it comes to computerised RNGs, if something seems whack, chances are, it is.

I think it's a bad seed for the rng as it seems to become static at times but then the seed could be anything server side. Ifrit seems to drop the same thing over and over when doing multiple runs in a row but not always. It's just random i would also check gear condition if that makes any hidden difference to the % chance of gathering success.

It's similar in beast mat drops. Kill 10 mobs, get one drop. Repeat like that for 50 mobs, then get 10 in a row.

The current RnG forces grind at an exponential rate. If they ever raise level cap to 70 the amount of RnG churn is going to be in the tens of thousands.

RNG has no memory. That is why RNG seems wrong.

You guys know that RNG's are flawed, right? or are you just talking about how some are better than others and the FFXIV seems below par?

Yeah here's the problem I have with RNG.

http://img.photobucket.com/albums/v3...ps184e13e0.png

That's an awful log of consecutive failures for 80% success rate. :p

EDIT
Though I guess you can't see I have it up huh. Well, I did, for whatever it's worth. :p

They might be reseeding the random call at every call...that would be a bug.

I came from FFxi where RNG means ranger. I'm not up to speed in the cool lingo but WTF does RNG stand for? Lol

Quote:

---

Originally Posted by Pride_Xiv

I came from FFxi where RNG means ranger. I'm not up to speed in the cool lingo but WTF does RNG stand for? Lol

---

RNG = Random Number Generation
As to the crafters, what direction were you facing and what was the moon phase? :p

Quote:

---

Originally Posted by Wobi

RNG = Random Number Generation
As to the crafters, what direction were you facing and what was the moon phase? :p

---

Thanks amigo ^^

One of my programming instructors made mention of True Computer RNG, as opposed to an algorithm which often relies on system clock or other influences as was stated earlier.

This is done on a hardware level.

Here is a link, quite an interesting read for those who like this sort of thing.

http://en.wikipedia.org/wiki/Hardwar...mber_generator

I have seen so many books from the primals it is amazing to me. I forgot garuda drops a bow because I never have seen it. Titan seems to have a large stockpile of shields.

Steady hand seems buggered. Gathering is really out there.

One thing for the people who are quick to call the original poster a conspiracy theorist. They admitted the drop rates in coils were bugged and the random generator there wasn't working properly after it was already brought to their attention on this forum. As much as I like the game is it that unlikely other randomly generated items/events might be off?

I also feel sometimes this way,

I have failed HQ crafting numerous times on 85% to even 98% (numerous is like more than ones on 98%)
while i have never ever had a HQ success on 2% or even 10% , my lowest HQ success was 14% and it happened less times than high percentage failure.

in gathering, similar, sometimes i fail all 5 HQ attempts on 50%... which is odd...

Honestly, i think the game should use a formula that checks the parties build before putting out gear.

Gotta love RNG. Ran 3 AKs, only piece of tank gear that drops is the neck, and I already got the Hero's piece. That's not my biggest rage about the runs, but the BLM/SMN body dropped all 3 times >.<

http://i.imgur.com/6tlxxkR.jpg

This was 11% success rate.
A total of 70 fails in a row.
We gave up, and sold the sash.

I do agree though, it does seem a lot less "random"... When I am melding and get a few lucky melds, generally it stays that way for a while.
When I am getting a lot of failures, generally it stays that way..

In addition to this - simply doing the primals - I have definitely seen both SMN and SCH books more then anything else.

agreed, the RNG seems retarded. a friend of mine crafted an item, 2 consecutive times 99% quality to end up with nq. me personally, i crafted 27 items in about an hour span (estimate) all with quality 50% or greater, 12 of which were 80%, and only got 5 HQs out of them.

I propose no longer calling it the RNG (Random Number Generator), and change it to the RFG (Random F***you Generator) because that is really what it feels like.

I dont understand why some people do not understand that random is random.

Quote:

---

Originally Posted by Mayin

I dont understand why some people do not understand that random is random.

---

We understand, it's just that the game does weird things from time to time.
For example, I crafted for about 6 hours yesterday (3 in the morning, 3 before bed), out of the hundreds of synths a lot ended between 80% and 89% to HQ, but only like 2 of them actually HQ'd (that were in the 80s range). I even had synths with 17% or 30% HQ, but if it was around 80% most of the time it just failed. Tell me that isn't weird.

Quote:

---

I'd like to see some discussion on this.

---

First we'll need to know the:
-Time of day you were crafting
-Which direction you were facing
-In game weather
-Outside weather
-Moon phase in game and on Mars (in particular Phobos)
-What kind of pants you were wearing

My issue is mostly with Garuda and probably ifrit. But I don't do Ifrit that much.

But I'm hearing from lots of people and reading it from here as well. That Garuda mostly drops Books and wands most of the time. And I experienced the same thing. It sometimes feel that a certain primal has a higher chance of dropping certain group of gear.

If I were to do an approximate list of what I've seen dropped from Garuda over the +100 runs I did. Then it would look a bit like this.

PLD: 4-5
WAR: 7-10
DRG: I don't even remember seeing one drop. But to be fair I'll write 1-3
MNK: 10
BRD: 5-10
BLM: 10-15
WHM: 20-30
SCH: 20-30
SMN: 20-30

Mage gear seem to drop like mad. Especially books.