RWG / POIs - Linked Loot Parts

[zztong]

Since there are Parts with tiered loot in them, it becomes possible to have multiple loot locations in a POI. What would be handy is if we could place multiple "end loot" parts and link them such that only one of the parts would spawn. In this way, the location of the final loot could be different from quest to quest.

Right now, for instance, I could have 5 x Tier5 loot parts with a 20% chance each of appearing, but they are not linked, so a quest might end up with 2-3 of them, or even none.

 

[zztong]

I haven't tried it, but it occurred to me today we already have this feature. Parts can have a name and it seems if multiple parts share a name then only one of them will be used. If so, this should work with Parts designed to only contain loot objects. It has worked with other Parts that I wanted to place as alternatives in the same location. I don't see why it wouldn't work with Parts in different locations.

The only trip up I can see is that you'd want the last Part in the list to have a 100% chance of placement in case none of the other Parts were selected. We don't control the order of the list so we don't know which one to make 100%. If we don't make an entry with a 100% chance, might there not be any loot placed?

 

[zztong]

I have now tried to use this technique in a test case. It generally works, but could be improved. Details...

TEST 1: I made a POI. I placed 4 "Loot T1" Parts in different locations. Gave them all the same GroupName of "loot." Set the probability of each to 20%.

Results: I observed 5 different outcomes. I either saw 1 of the 4 picked and placed or I saw none of the 4 picked. This makes sense, as the combined probability of placement for 4 Parts at 20% probability represent a combined probability of 1.0 - (1.0-0.2) ^ 4 = 0.5904 or 59.04%. Thus there was approximately a 40% of none of the parts being picked.

TEST 2: I modified TEST 1 so that the probabilities were 20%, 20%, 20% and 99%. The idea being that the 4th choice would always get picked if the first 3 were not.

Results: This worked as I did not see any results where none of the Parts were placed. However, the 4th choice (with 99%) was picked most of the time, which makes sense.

TEST 3: I attempted to tweak the percentages such that all 4 Parts might have roughly the same percentage of placement. To do this, I needed to pick a percentage for the first 3 Parts that would leave roughly that same percentage for the 4th Part even though I had to set the fourth part to 99%.  I ultimately settled on 30%, 30%, 30%, 99% because 1.0 - (1.0-0.3) ^ 3 = 0.657 or 65.7%. That left a 100% - 65.7% = 34.3% chance of the last Part being picked.

Results: While I did not see a case where none of the Parts were picked, I still saw the majority of the time the 4th Part was picked, rather than a more-even distribution.

Conclusion: Since we're dealing with probabilities, it is possible TEST 3 really could lead to more-even distributions. More testing would be needed to see that. Still I suspect POI designers are not likely to want to go through the probability mathematics to balance out the odds of even placement. While configuring Parts, the Probability slider has 20%, 40%, 60%, 80%, and 100% as easy stops. You can select any probability by clicking in the track. (Click-Drag in the probability slider produces NREs.)

 

[zztong]

Suggestion / Use Case: (see also, Testing found above) @Laz Man @Kinyajuu

I would like to randomize the location of final loot within a POI. In this way, there might be 4 or 5 different locations that the final loot may be placed each time the POI is initially placed, or regenerated by a quest. What currently does not seem possible is that POI Designers cannot ultimately arrange the probabilities of placement for each of those Parts in a group such that they can be certain of an even chance of placement for each location. It appears you can get close, and you may end up with no loot having been placed.

Discussion:

A notable complication is that Groups of Parts are not defined discretely, but are collectively defined by their members. That means if the configuration of 2 of the Parts had a flag set indicating "Even Placement Within Group" was desired, but 2 other Parts did not, then which approach should RWG use?

Ideally, choosing a Part to be placed in this manner does not depend on a Probability setting in the UI, but is the equivalent of picking a chip from a hat.