A Random Comment

[Yukiko]

Ah yes, I did forget to mention a bit on how this randomness works on the programming level. If you read the OP's opening post with the quoted text, they mention the RNG but add pseudo to it. Yes a computer cannot inherently create randomness. It's an impossible feat for it to do when it's designed to be accurate. If it wasn't computers wouldn't work like they do now. As such, a random list of numbers is used. However it's not just one list, there are a great deal of lists. The calling up of the random function requires a seed to pick a list to use as the call of the function. If you don't want to be unpredictably random you'd use a predictably mutable seed or a fixed seed. I believe this is how older games used the pRNG and you tell this is the case when you can predictably have the same outcome from fresh start of a game from powering it on. Look at games that have an abusing the RNG guide. To prevent an application from having this issue you want a seed that is unpredictably mutable from the onset of the starting of the program you are running. In most cases the random seed used is the current time. This works rather well since the call to current time returns a number that is the number of seconds that have passed since a predetermined date making the result a different seed number every time you call the random function. it's still not completely random but it makes it really hard to predict the seed unless you know exactly what time random was called, how current time returns a number and how random uses a seed that might fall outside the bounds of the number of lists it has to use.
Finally after a number has been generated for use, it's however the coder wants to used it. Since this game uses percentages for success for crafting HQs, Iwouldn't see why it's not the same for atma or alex drops from FATEs. In this case how I've usually see it done is if the resulting number is less then or equal to the percentage odds where both numbers used are in the same scale. Scale begin that if the percentage is a whole number the numbers 1-100 are used.
Computer randomness is is rather complicated and this is mostly a gist of how it works. But even I as a programmer has problems with understanding all of how the random function works because the function is based on what source you are calling it from meaning that in different coding environments will used a different way to call a similar random.

yea, but the problem is not only the seed, you can programm complex code around "random()" but you still use that command to generate the number and so the workaround may result in same randomness as when using another algorythm but much more complex. prng tend to have a pattern, no matter what, what we need is a trng but as the OP in its quote said: you need more trials. A good example is when you fire on a net with a prng and let you generate a number out of the result. The more shoots you do, the better randomness you will get - but thats the problem with the server issue...

Indeed there are some really good prng out there, but i had to learn they all have their weakness... In a MMOG you add the number of players, so even when just having a few bias you can be sure someone is feeling/seeing it. Thats why you need failproof systems - a good analogy would be fire alarm! While in most programms you can do a lot trials to keep bias low that will not happen in a MMOG, too many factors compensating the trials... You have to question if the effort in failproof systems is worth the outcome or not, and sometimes as we see in weekly monsterhunting it can be just one line of code to make it disapear...

I think the discussion is moving a bit in the wrong direction because its not really a problem of randomness but more the miss of failproof systems...
Furthermore, the community isnt angered about some odd bias. The problem is that it tends to happen all the time!
That may be just an imagination, but still when players notice it, there is something wrong -> if you let a "worm" randomly move and it drifts bottom left corner you see there is something wrong...
Its not only the RNG, the problem is we have it in nearly everything... What I wonder is why? In 1.x (Yoshi-P version) they have done a lot to eliminate RNG from game, why in hell they gone back to that stupid RNG gamedesign?

http://www.youtube.com/watch?v=itaMNuWLzJo
http://www.youtube.com/watch?v=_tN2ev3hO14

[XNihili]

Random being server side, if I failed 10 times with 99% chance to win it means that there are also 10 other people who did win instead of me.

[Tiggy]

I'll just leave this here... (how is it still not in the thread xD)

It's funny yes but in computer programming we can be sure. We can pick a range, say 1 to 1000 and generate millions of random numbers in just a few seconds. We can then look at the distribution of those numbers across the range and see if it is even or not. In computer programming we are most certainly sure.

[MiriyaGarisk]

I have been that person who fully melded a ring on the first try and hit that 7% meld for a friend on the first try. I have also been that guy who failed for an hour on something that was almost a guarantee. I can't tell you if the Random Generator is truly random because I don't care to spend the time to do a Monte Carlo analysis or Anovas based around the supposive seed conditions, but I can tell you that I think basing the game on as much range with low probability of success is a pretty shitty design. If Alex drops at 1% from fates, then I much rather need 7500 Alex and make it 100% drop. The Deviation around such critical low drops is just not fair to the player base when there is not enough content to smooth out the experience from player to player.

[MiriyaGarisk]

In 1.0 SE had a system to dampen the range on primal weapons and conceded that if you did Garuda X number of times you earned your prize and not this well to bad so sad crap it is just RNG they keep shoving down our throats since 2.0 began. I don't expect anything to be easy or given to me, but if I can beat T9 10 times I kind of expect to be given my due and not go 40 times and into the next patch without my drop.

[IndigoHawk]

Not to mention the size ratio comparison of car:road and plane:sky. In traffic you may have 100x+ the amount of cars within 1 mile of you than a plane will *ever* have within even 10 miles of it.

This is probably true ... but not relevant. You believe that if planes were as tightly packed as cars, then they would have more accidents. Therefore you feel that planes are more dangerous cars. You're probably right, but that doesn't matter, because that's not people use cars and planes.

In regular use (taking a plane trip from an airport or taking a trip to the grocery store in your car) a single plane trip is safer than a single car trip. This is based on the history of plane and car travel. (Of course, both are actually pretty safe, so there's no need to worry.) Safety statistics are based on how things are used, not how they were intended to be used or could be used.

[IndigoHawk]

Yes a computer cannot inherently create randomness. It's an impossible feat for it to do when it's designed to be accurate.

This problem is overstated. Random functions in programming are very random and good enough for most uses. You are correct that programming doesn't generate true random results, but you rarely need that level of accuracy.

The difference between true random and very random is only important to complex simulations and other very precise operations. For something like FFXIV that wants to hand out ATMAs or determine crafting results, any modern random algorithm will be accurate enough.

People who believe that the system isn't treating them fairly because it uses very random instead of true random are incorrect. Unless the code actually has a bug, then it's random enough, and people are experiencing confirmation bias, not a flaw in the random computation.

[Yukiko]

Unless the code actually has a bug, then it's random enough

And thats where you are incorrect... its just false because you cant say it in general... it depends on how its programmed and wich numbers its using to repeat the code...

[tymora]

The issue here is precisely that for all practical purposes, the RNG is well...random.
While it converges if you have enough samples, for the individual player, that can mean a difference between 1 and 1000 trials. The lower the probability, the more samples needed to converge.

To keep players sane and to maintain perceived fairness, they SHOULD introduce some form of bias, such that it prevents "too good" and "too bad" occurrences.

[Nekodar]

If Alex drops at 1% from fates, then I much rather need 7500 Alex and make it 100% drop. The Deviation around such critical low drops is just not fair to the player base when there is not enough content to smooth out the experience from player to player.

Since Alex from FATEs is just one option out of currently 4 ways of obtaining them, your suggestion would make things even worse for the majority of players.
1) Alex from FATE
2) Alex from Allied Seals
3) Alex from map obtained through daily quest
4) Alex from map obtained with myth tomes