Socket Chances Revealed!

That 6S doesn't seem right. I crafted a lot of stuff before and I always got a 6S within 2-3 Jew stacks. And I don't mean just 10 items. I probably did this for 100+ crafting items.

I am an idiot, sorry - for some reason, I utterly missed the relevant Chris' comment.

I stand corrected, until now I would swear I have already seen a situation where 5S turned into 5S (~no-op) but if Chris says this cannot happen then my memory is just wrong.

But this is just another case begging for much more opennes from GGG side - we would not need to do this archeology if the underlying mechanics would be disclosed

so 0.33% x 315 jewelers = 100% chance 6 sockkets?? nais

"

morbo написал:

I've 6S with one jew and not 6S with >400 jew.

I despise socketing and linking. This absolute RNG high variance is retarded. Only gambling addicts find this fun.

I've never felt happiness of achievement when I got a 6S or 5L. Just relief it's finally fucking over...

couldnt agree more.. but unlike the OP here, i've spent more with 0 result getting a 5L or a 6L item...

unlucky? apparently..

at one point, i spent more time farming for raw materials to create JEWs orb then anything else i've done..
its a stupid part of the game.. that i despise and needs to be done away with..

make it an option to buy or craft- give a vendor, say, 10 of these and 5 of these and so on and craft a 5 or 6 linked item.. problem solved.. (imo of course)

"

Firecrest5 написал:

"

syrioforel написал:

"

Malone написал:

it means 300 jewellers in average to get 6S ? weird, I think it was easier than that...

300 on average means most people get it in roughly <= 200.

WTF

It's true :D.

If the chance to 6S is 1/300 (which means the expected # of jeweller's orbs spent is 300), then the odds of *not* getting a 6S after 200 tries is

(1-1/300)^200 ~= 51%

so 49% of people are done at or before 200 orbs in.

-----

Median and mean are different. The mean value is 300, the median is 207.

EDIT: Here's a graph of your probability of having a 6S before having spent a total of x jeweller's orbs:

.

The odds are stated at 1/300. That puts the expected # of orbs at 300 and the median number of orbs spent at ~200. Again, median and mean are different.

EDIT: Removed quote since the poster removed it.

I just want to point out there is no such thing as memory.

The game doesn't "remember" how many orbs you've used.

Just because you've failed 500 times doesn't make the 501th more likely.

With a 1/300 chance it will on average take 300 orbs, however each orb is an independent event (1/300).

"

Zaqwert написал:

I just want to point out there is no such thing as memory.

The game doesn't "remember" how many orbs you've used.

Just because you've failed 500 times doesn't make the 501th more likely.

With a 1/300 chance it will on average take 300 orbs, however each orb is an independent event (1/300).

Indeed!

This is how the median (~200) can be so low compared to the expectation (300). There are some poor souls (1 in 50 will need to spend 1200 or more orbs) making up the difference, bringing the mean up to 300.

Most people will do fine (<= 200), and a select few will get shafted, and shafted hard. This reflects the QQ posts we see on the forums regularly.

"

SunnyRay написал:

Actually, I still believe simulated 0,004489% might be more accurate than calculated 0,004529%, because calculation assumes that number of sockets distribution before using jeweller orb is based on "tickets" for 1-5 pool and independent from previous roll, which is true for the very first jeweller orb used on an item after drop, but not true for all the rest (subsequent rolls are depentent on preceding).

I think you've got it backwards, my table accounts for not rerolling the same sockets, which is why the cell at column "1" and row "to 1 socket" is 0. What it fails at is calculating from an item that isn't 6 socket to start with, as it's just a simple probability table.

Also, the calculated amount in that spreadsheet is 0.4491569...%, not 0.4529%, not that it matters much. I ran a simulation, as well (though it was written in c++), and I came up with similar numbers to yours. I actually ran one that stopped after 10 million successful 6s items (which is about 3 billion jeweler's), but that took a long time and I didn't record the results.