Vendor recipe - need some help with statistical analysis(or an answer from GGG)

Base recipe:
Put 5 items of the same type and rarity(magic or rare) in the vendor window and obtain 1x unid item of the same type and rarity, with new sockets.

Modifications:
Add a 6th item in and/or add in jewelers, chromatic, or fusing orbs to modify the socket number, color, and linkage.

Testing:
Started with 6x claidheamn Mor(one equipped to free up more inventory space).
10x chromatic(1 per space)
10x jewelers(1 per space)

I get different sockets with all 6 swords in the trade window, or with any particular distribution of 5(with only 5 I get 1 or 2 sockets, different colors). Socket distribution is always the same given the same input.

I then proceeded to test with the different orbs.

I am only going to list the number of sockets here for simplicity.

When I use the orb in slot 1, I get 1 socket, slot 2 <->2 sockets, slot 3 <-> 2 sockets, slot 4 <-> 2 sockets, slot 5 <-> 1 socket, slot 6 <-> 5 sockets, etc.

If I change the order of the orbs in my inventory, relog, put them in my stash then move them back to the inventory, etc. the same orb always gives the same value.
Combining the orbs also always gives the same value for the same combination, but is not predictable based on the values for the individual orbs. for example, placing 2 orbs that each individually give me 5 sockets into the window together only gives me 2 sockets.

Each orb I add to the window also adds a secondary item(fragments, orbs, scrolls, etc.). In some cases I have added an orb to the window and had it show me an alchemy shard, and when I add a second orb, it now shows 2x scroll fragments.

Preliminary conclusion:
Each individual orb seems to have a fixed value. Each result is replicable. The sockets/colors/links seems to be based on a formula which uses a value for the item and the value of the orb, as changing the resulting item(by swapping out one of the source items in the recipe) changes the expression of each of the orbs, but the new expressions are also consistent and replicable. Figuring out how that fits together is beyond me. If someone wants to help me with the methodology, I don't mind testing it. I am able to create any number of sockets from 1-6, and by marking each variant of the recipe I want to use(including the coordinates of the specific orbs in my inventory so I don't mix them up) I can replicate each expression(at least until I finalise the trade).

Reserved for further information.

Please post if you require clarification or more info on anything I've posted here.

"

Blizzisme wrote:

Please post if you require clarification or more info on anything I've posted here.

I'm not sure exactly what your question is, but the behavior you are talking about makes sense from a programming perspective.

What i can tell in plain words: you won't find a way to generate the exact socket combination you want on demand. For all intents and purposes, every different item or combination of items you put in will randomly give you a new result.

yesterday i spent 2 hours trying to get a good amount of links. Best i got was 6 sockets with 4-2 links (1 link missing in the middle).

I dont know why youre using orbs or other higher value items. I only used scrolls, u can even use crappy fragments, any item will do.

To me it seemed TOTALLY random. I could reproduce some stuff, but sometimes it changed completey, like there's some kinda timer on it. I don't know if there's really formulas, and if some items are better than others (i dont think thats the case).

Either way it seems EXTREMELY hard to make a 5 or 6 linked this way, as i spent 2 hours trying to do it without any result....

"

masteralbert wrote:

Either way it seems EXTREMELY hard to make a 5 or 6 linked this way, as i spent 2 hours trying to do it without any result....

Interesting. I can hear the complaints coming already: "easiest" way to get 6-linked is super tedious.

"

masteralbert wrote:

To me it seemed TOTALLY random. I could reproduce some stuff, but sometimes it changed completey, like there's some kinda timer on it. I don't know if there's really formulas, and if some items are better than others (i dont think thats the case).

The results per orb did not change over the space of 4 hours, As I said above, the results did change if I changed the specific claidheamns I had in the sell box. So I had 6 claidheamns, If I took out claidheamn number 3 and just had 1,2,4,5,6 then it was the same as every other time that I had 1,2,4,5,6... No matter what order the swords were placed in. If I took out 4 and put in 3, then it was the same every time that I had that combination.

To me, it looks like something working like a hashcode. A somewhat random function which always give the same results with the same inputs. So the only way to know how to make 6 linked items is to know the hashing function. And they won't tell you ;)

To be clear, it's not less neither more random that the other random stuffs. As you might know, true random doesn't exist, it's always mathematics ;)

I feel you left out a critical piece of info.

/itemlevel

There's too many variables in your testing. You need a control.

Start with 6 items of all the same base type, rarity, quality, socket layout and /itemlevel. It might help of all of these items were 'found' in the same condition as opposed to being modified with scouring orbs, jeweler's, chromatics, etc.

The socket results need to be hidden when using this recipe. Allowing it to go into open beta or beyond like this would be a huge mistake. "Vendor bots" will be made, and 6-linked items will be about as tough to find as "high" runes and sojs were in d2.
Nobody really wants to sit in front of a vendor for hours clicking and checking socket combinations... so there's no reason the ability to make huge profit doing that should be added/allowed, as it will only open up huge potential for abuse, and adds no fun to the game.

I find this feature very fun and rewarding.