A Random Comment

[Felessan]

But when do you do such a large amount of crafting? I only when I got a huge amount of low level normal ingredients.

I craft 2* food for sale, usually 1-2 stacks per day.
Dealing with randomness is tough, but the first important thing ppl should know that if a random doesn't cluster good/bad results, it's not a random.

With atma I want to keep this random thing. I don't want guaranteed win, this is about patience and the will to go on.
While very frustrating, I don't want it to change.

I would suggest a general tips with randomness. Make breaks often, change a zone if no drops for a long time, don't let it burn you.

[Teakwood]

That's the point.

What do you expect? That the game realizes you've had a bad run and eventually just gives you a win so that you feel better? Is that the kind of game we all play these days? Happy go lucky fun time where prizes are handed out just so you don't ever know the slightest amount of frustration?

I was being sarcastic, sweetie.

[hqdm]

I appreciate the dev reply but well, I don't understand it still, I guess.
RNG relating to spiritbond results, because I do a lot of that, I can see it's more or less even result over a great many results.
But relating to other things, it doesn't feel like that. I'm aware of cognitive bias, that part I did understand.
A few days ago I failed ~12 infusions in a row on my Det IV sphere scroll.
A few days before that, first overmeld at 29%, this

I mean, it was expected to some extent that there would be failure.
and I suppose it might be part of my bias that I notice it more because Det IV are really kinda expensive
but it just feels like the listed %s don't match up with the results at all?
The most striking example I can remember quite clearly:
I basic synthed an ingot at 2% HQ chance and it became HQ. When crafting the result, I only reached 98% HQ and it became NQ, back to back.

[LineageRazor]

Dealing with randomness is tough, but the first important thing ppl should know that if a random doesn't cluster good/bad results, it's not a random.

Unless it randomly happens to NOT cluster good/bad results.



Anyway, OP was an interesting read and pretty much confirmed what I already believed to be true about the RNG in this game: Random is random. Nothing is "broken", there are no "tricks" for getting better winning streaks, etc., etc. All such instances are just people spotting momentary patterns.

This post should be stickied to the front page, so every time someone comes up with a harebrained "If you desynth while facing west during sunset you're 10% more likely to get demimateria!1!" theory, we can point them to this post.

[Zaero]

1 post to go WAATMA. Stay classy ff14 forums.

[VitaminMT]

What totally boggles my mind is that I can craft HQ items with NQ materials using quick synth... if i make a big batch of an item I will often get 1 - 2 HQ items. That just tells me that the % of quality when doing normal synths is broken surely it shoudl be 0% chance of HQ?. Too many times I have had 96 - 98% HQ and failed! Failing on that same 1-2 % chance soo many times is impossible! nobody can be that unlucky!

Starting a craft with all NQ materials puts HQ chance at 1%.

I do think we remember the failures more often than the successes, and that sticks with our bias. I know for a fact that I've produced HQ many times off of low percentages before, but those don't stick with me as long as a 95/98% NQ that I'll be stewing over for a while, especially if it was an important craft.

Ya'll gotta learn to control ya'll biases.

[Waeksyn]

I really don't challenge that SE's RNG algorithm works correctly. Random is random. The impact, though, of being on a bad "losing streak" is really not good for the game. For some actions in the game (melding, atma, other) the RNG method should be tossed in favor of one that DOES look at history.

Let's say the chance of success at an overmeld is 25%. If you fail, the next attempt at the same action would have a higher chance of success - say 35% or whatever. If you fail again, the chance increases again. Eventually, if you continue to fail, you will hit 100% success probability. When you succeed, the series gets reset. This would totally eliminate the crushing effect of really bad luck - farming atma for hours at end, spending millions of gil to get that last overmeld.

[VitaminMT]

I really don't challenge that SE's RNG algorithm works correctly. Random is random. The impact, though, of being on a bad "losing streak" is really not good for the game. For some actions in the game (melding, atma, other) the RNG method should be tossed in favor of one that DOES look at history.

Let's say the chance of success at an overmeld is 25%. If you fail, the next attempt at the same action would have a higher chance of success - say 35% or whatever. If you fail again, the chance increases again. Eventually, if you continue to fail, you will hit 100% success probability. When you succeed, the series gets reset. This would totally eliminate the crushing effect of really bad luck - farming atma for hours at end, spending millions of gil to get that last overmeld.

Nah, how about we don't do that at all.

[LeonBlade]

Not to call you out but... I call BS on this. Not impossible i agree, but very higherly unlikely.

I understand, I was surprised myself, there's no reason to lie about it on here just to make a point. I've seen someone HQ a i95 Garuda weapon with like 15% or something, you can watch the video here: https://www.youtube.com/watch?v=dWGZG5EgMuM

[Tiggy]

Tiggy, the time and cpu clock etc. are all known and constant variables, let's go with the system clock, the seed is 9am on the 1st June 2014, you can (if you know the seed date) know what the 1,000,000,000th number is, it isn't random, it has an end result which can be known, there is a 0% chance that you *know* if a coin is going to be heads or tails next toss, it's true random.

Despite the fact those were just examples I'll go down the rabbit hole real fast. The time is a known variable yes, but it's only valid for a millisecond. Not enough time to do anything about it. You don't know the range they clamp the results from, you don't know if they random a few times before picking the final number, or any other such techniques to improve variability. Just knowing the seed isn't as powerful as you think and even then the seed is so fleeting it won't matter between individual lots. For all intents and purposes it is still random enough for uses in games like this. You will never be able to use knowledge of the temporary seed and predict your outcome.

It seems to me you just want to argue random number generation theory though as opposed to how it practically effects the game.