Ifrit (Hard) Weapon drop rate is a joke.

[MarcusY1990]

All I see is PLD blade, WHM cane and Grimoire. mostly the blade a lots. Trying for the axe and haven't seen one yet.

[RoughDivider]

I love people who don't understand probability, sure the 11% is the chance each one drops. However, I could put 9 marbles in a bag and end up picking the same one 50 times, it just means I need more tries to normalize it out to the 11% chance. Soooo theres my simple explanation of probability have fun kids.

[MieyCyruuuus]

lol@ 50 being a legitimate sample size.
you just got unlucky.

[RoughDivider]

lol@ 50 being a legitimate sample size.
you just got unlucky.

I was saying that it isn't a legitimate sample size, people just need to be hit in the head with stats books.

[MieyCyruuuus]

I was saying that it isn't a legitimate sample size, people just need to be hit in the head with stats books.

I was agreeing with you and replying to people that are complaining based on personal feelings.

[GeilLega]

Hey kids! n=20 is the beginning of statistical relevance, and actually above n=5000 or so you need to start performing other tests on your data sets in order to ensure heteroschoasticity and identify other data purity problems. 50 is probably on the low side given the number of possible results in a sample, but it's certainly enough to make some statistically informed observations. An easy one is that grimoires drop more than any other weapon. This actually makes *some* sense since multiple classes use them, while only dragoons use harpoons; it would only make logical sense to drop more of a commonly shared weapon.

To the above posted that made a completely inane marble comment: you sure could put 9 marbles in a bag, randomly select one, and select the same one 50 consecutive times. The probability of this occurring is calculable and is .00000000000000000000000000000000000000000000019% (ok I rounded it off, R will take it out quite a bit further but that illustrates my point I hope). It does not mean you "just need to sample some more to normalize it." It means you got an incredibly rare occurrence that is nearly mathematically impossible. If you've ever done the classic "pennies" problem when this type of probability event is first introduced in math you'll know that even flipping heads on coins 10 times in an unbroken row becomes fairly rare (.097% of the time, or a little worse than 1 in 1000 times)

[Hudskie]

20 runs and i finally got a harpoon. always saw grimoire and book. got already those and a bow. at first I feel like, "do I need to give up hunting a harpoon and just reroll to other class?" the next run. I finally got what I want.

Try to get angry at your lady luck, it somewhat effective.

[reckless76]

Computers are horrible at generating random numbers. The best they can do is produce a string of numbers that only look random to the naked eye. There are, of course systems that purport to produce truly random numbers, but they rely on things like radio-active decay or atmospheric noise; and who here really believe SE has a geiger counter hooked up somewhere to randomize loot drops? The truth is they are likely using the cheapest (computationally) method baked into most programming libraries: the linear congruential generator. This algorithm requires a seed, and it is in this implementation they could have screwed up. The questions should be how often do they instantiate the generator? Once a day? Every instance? And where do they get the seed? Date-Time? Character data? If the generator is started every time a party picks a fight with ifrit, and the seed is poorly chosen, then yes, the loot drops could be perfectly predictable and heavily skewed towards certain items.

[Hysterior]

Took my BLM friend for his first Ifrit run = BLM weapon on first try
I think I had my SCH book within 5 tries as well

on the other side, Garuda never dropped my book...

[MieyCyruuuus]

snip

I'm really not sure if this is even a serious post.