90% sucess rate and still missing

[Dirtyfinger]

1% was never 1% in FFXI, so this wouldn't surprise me.

[Xbob42]

There have to be other factors that govern the actual percentage, because I can almost guarantee you the number you see, unless it is 0% or 100%, is not accurate. I have 90%s fail multiple times on every single craft, and everyone I know shows the same. Someone needs to log this shit, like, thousands of usages of it at different percentages, and I'm 100% sure it'll show that all percentages unless it's guaranteed to succeed or fail are MUCH lower than they appear to be.

For those arguing that it's totally normal, how many times in a row per day do you get a 10% or under synth/gathering? You might get the occasional HQ, but if I got a 10% HQ as often as I failed a 90% chance to do something in this game I'd have roughly 1/3 HQ items with no effort whatsoever.

[Astarica]

One way to look at probability is that if you make 1000 crafts, then statistically speaking it is extremely likely you'll see an event that should only happen 1 in 1000 times. However you should not see something that's supposed to only happen 1 million in 1 in 1000 attempts let alone anything rarer. To use a concrete example, with 90% chance to gather on a 6 gather node, the chance to fail 3 out of 6 is about 2%. I certainly have seen this happened and I certainly gathered these nodes far more than 50 times, so seeing 3 out of 6 fails with 90% chance isn't anything worth alarm unless it happens repeatedly. Now failing 6 in a row is 1 in 1 million, which I've yet to see and I should not expect to see it happen for a pretty long time. If I fail 12 in a row, that's 1 in 1 trillion and almost certain that the % displayed in the game are wrong, or that I should've bought a lottery ticket instead.

[Geobryn]

the server runs random calculations like trillion times a day, and there is one guy who saw 12 fails in a row, i guess it matches up :P

[Claire_Pendragon]

When it comes to FFs, I've noticed bad luck, where as in other MMOs I have insanely amazing luck.
(FFXI spent 5 years trying to get an item with a 30% drop rate, eventually just quit the game over it)

Here I had 90% chance, and failed 27 times in a row.

I had 20-36% chance to HQ mats, and failed to HQ all 450 mats. (1 mat per attempt)
Then had a 4-13% chance to HQ another mat, did it 300 some times, and no HQs.

But if I get 60% or higher, I HQ 99% of the time.

Technically there is no such thing as random, and I'm starting to wonder about the algorithms designed to give the illusion of random.

But then again, I could question every other MMO as well.

Ive gotten things to drop repeatedly, even if they have a 0.01% chance to drop, or lower.

(My best record, was farming two dragons call swords from WoW, despite my PT consisting of all sword users, who all roll need against me. Its a 1 out of 15,000 chance to drop, now divide that by 5 PT members, and now do it again. Also, any time i run that dungeon as a sword user, I get the sword.)

[Conradus]

Technically there is no such thing as random, and I'm starting to wonder about the algorithms designed to give the illusion of random.

That's a little sweeping. There's no such thing as random in a deterministic process, and as all computer programs are deterministic processes, they cannot by themselves internally generate truly random numbers. Many physical processes are truly random, and there is specialized hardware available if your application needs a reliable source of truly random numbers (many cryptographic protocols need such; if you are without randomizing hardware, your computer will look to such external factors as the exact timing of network packets and key presses to try and gather the entropy it needs).

Pseudo-random number generators have gotten a good deal of study. Many are indeed lacking, but standards for evaluating them and good implementations do exist. It's not my field, so I don't know the details, but it is possible to do well in this area, although many don't.

[Caraway]

That said, most computerized random number generators are actually pseudorandom number generators (they for the computer to do it, there must be a formula or algorithm; usually, it involves a seed value and a function of time from a set reference point), and are prone to streaking, but even out over large samples.

This really touches on the issue to be honest.

You never get proper randomness out of a computer. Random number computations are based on seeds and each seed can provide differing sets of results.

That said, from my anecdotal experience with crafting (and harvesting) it definitely seems like the game has a tendency to streak to the extremes. There may be something to investigate, but the amount of work it would take to really dig into the number generator would prove daunting for us players (and a cake-walk for the developer's provided they accurate log all of these events, which I hope they do).

You can't prove or disprove something by how the system feels but I can definitely say that the randomness in this game feels wrong, compared to the many other games I've played with random chances being so obvious.

[Conradus]

You can't prove or disprove something by how the system feels but I can definitely say that the randomness in this game feels wrong, compared to the many other games I've played with random chances being so obvious.

It's been shown in studies, many, many times, that human "feelings" about probability and randomness are inevitably so wrong as to be laughable. Statistical analysis based on hard data or GTFO.

[Caraway]

It's been shown in studies, many, many times, that human "feelings" about probability and randomness are inevitably so wrong as to be laughable. Statistical analysis based on hard data or GTFO.

Which is why I said you can't prove or disprove by how something feels. It would be interesting though if someone had the time to run the numbers, or if SE could provide them.

[Ilmoran]

That said, from my anecdotal experience with crafting (and harvesting) it definitely seems like the game has a tendency to streak to the extremes. There may be something to investigate, but the amount of work it would take to really dig into the number generator would prove daunting for us players (and a cake-walk for the developer's provided they accurate log all of these events, which I hope they do).

Crafting is a bad place to look at the RNG anyway, because your target numbers change so often (one step your success/fail is based on a 90% chance, another its 80%, the next one is 100% chance so you can't even see what number you generated, etc). Harvesting is in general a better place to look at the RNG, and over the course of gathering a few hundred items, I usually see the rates come out to their expected outcomes (again, RNG is streaky, but in large samples comes out correct).