Square Enix's RNG

[Moirear]

I understand probability quite well. Did statistics in University. When I frequently see sequences that have less than a 1 in 100,000 chance occurring, every 500 chances or so, and others complain of the same thing, we're either all incredibly unlucky, or the system is prone to excessively long sequences of extreme values.

You're all incredibly unlucky, because some others are incredibly lucky. My results are fairly average.

That's just how it can go. And most RNGs are basically pseudo-RNGs because they're not based on natural phenomenas that are truly random, so they can occasionally be perceived as prone to extreme values.

You're doing the math on a few samples, but you'd have to gather the data from all players, all checks, all the time to get a more reliable picture of how well the system works at acting random.

[Ksenia]

Having gathered up to 800 Mythril Ore, equal numbers on other items, I can confirm that HQ is around 15-20% and it's like that all the time. Proc for Mooch is similar. For Gathering, it's 15-20% of hits though so if you are getting nothing 40% of the time, you get 15% of the 60 times you hit.

Crafting is different. If I can get 50% or 75% or 88% with all Normal condition and do it over 100 Synthesis, my true HQ will come up half on those synthesis. The number fusges based on Good quality but the true value of my crafting percentage is almost always half. It doesn't matter what the bar says.

[GilbertLapine]

Anyone else feel like SE's RNG is a complete joke?

I have been crafting all night getting about 80-90% chance of HQ and swear I have failed about 50% of the time.

Meanwhile I barely get any HQ with 15-20% chance HQ.

Same goes for gathering. Several misses in a row on 80% chance.

They really need to update their RNG formula.

/rant

Haven't had an issue with the crafting, but the RNG sure gets screwy when they unleash it on the loot tables. Had me convinced for a while it was picking drops by day instead of by run.

[Velox]

And the ugly truth to %-based rates is that, mathematically speaking, there exists a statistical probability that someone will never get an item they desire. Game companies won't sell you that info though. :P

Obviously the odds are very low assuming an infinite number of attempts, but it's not 0.

[Rivienne]

What did you do 300 times? All I saw was you mentioning AK, which any RNG factors therein are far more obscure than 98% HQ success rate, so I am not sure what one has to do with the other.

AK? No idea what you think you read, but I brought a specific test in my first post on thread of mining 100% chance nodes, and using unearth to increase HQ percentage. with a stated chance of 25%, I saw instead 38HQ over ~300 attempts (75 nodes total, so 300 is the minimum. It may be closer to 310 but for simplicity I rounded).

I am btw, neutral on this subject, I am not complaining, or defending the system. I am defending the possibility that the system is flawed however. Based on my experience and testing, the system seems flawed. That doesn't mean it is, but my one stated example was large enough to be meaningful, though not definitive. However for personal play, the system works good enough for me.

Regarding loot drop: I don't care at all. When stuff drops, better than my gear it is an oooh free upgrades moment. I don't go looking for them however, so I am not interested either way in chance rates of loot drop.

But gathering is very clean and easy test for the RNG system, which other people can easily test to backup or refute my numbers. Mine any 100% chance item with a 15% chance HQ over a sufficient sample size, of 50+ nodes. Use unearth consistently at every node to ensure that every node is given the same 25% chance at HQ. (25% is a good number of this type of test, because the sample size doesn't have to be prohibitively large to be meaningful). Count the total number of NQ and HQ at the end of the test. I would expect results of 18-25% HQ.

[Inquisitor]

My own personal experience is that, over time, HQ gathering percentages do tend to match the listed percentage. Not that my sample size is statistically significant, but NO single player (or even a couple of players comparing notes) is going to have a large enough sample size avoid being an outlier on the bell curve.

The displayed percentages may, in fact, be incorrect and I've been lucky. My experience is no more or less valid than others.