RNG is nothing else then changing a fully straightforward solution to acquiring items into an amplitude in a wave-function.
Which means it becomes chance-based but overall every single person will get the respective item at roughly the same time unless they’re extra lucky to be on the good side of said wave, hence the amplitude giving them the item much sooner… or the other direction.
Issues generally arise when the RNG is adjusted too much in any of those directions, so an example of that:
We imagine we get… let’s say 10 uniques an hour (example number solely). Now to get a… glove base we have a 5% chance. That means under normal circumstances a player gets a glove unique every 20 hours. Done and gone.
The RNG factor of it rolling every single time again what exactly can drop though could also give you a glove base every single time during the farming time and nothing else… it’s vastly less likely then you getting a meteorite to fall right on top of your head… but it’s there.
The same goes for not getting it in 20 hours… or 40… or 60.
Determinism hence is needed to alleviate the extreme circumstances of any RNG system. You’re not supposed to drop the same item 50 times in a row and you’re also not supposed to miss the drop-window by 20 times over the baseline.
Hence:
Does basically not something you as a game dev want to have especially.
Choosing to reduce the loot pool from which an item can be chosen is one thing… but actively changing the loot-chance inside a specific pool is a dangerous direction.
Imagine the axe example. There’s a few rare drops in the pool, let’s say while you have a 95% chance to drop a ‘normal’ one then the rare ones are 5% chance drop in comparison. So you get one every 20 times any other are otherwise dropped.
Now commonly LP is as much of a roll as choosing the base type and the drop inside that pool, hence a RNG number which simply needs to be above a specific threshold to cause it to drop.
Enforcing a LP item hence needs to make the minimum roll the one which aligns with the number needed to allow 1LP in the first place… unless you want it to simply roll up to ‘1’… which is nigh useless as it doesn’t change the ‘2’, ‘3’ and ‘4’ chance at all. So it need to affect them all.
So in that case if we have a 1 billion pool for the numbers it can roll then we need to automatically cause the roll to add the minimum requirement for 1 LP to the roll. Let’s say something has a 15% chance to get 1 LP then that means instead of 1 billion we only have 850 million rolls possible from then on.
Which is a vast difference already.
This causes your follow up 0,7% chance to get 2 LP to become something entirely else though it becomes far higher, the same with 3 and 4 LP.
Which is all still fairly fine in itself to a degree… but besides being a problem with integer overflows from a coding standpoint (since the numbers are already so high) it also can’t simply increase the pre-existing LP by ‘1’ since that skews the chances in one utterly ridiculous way, removing the lowermost pool which is around an e-7 chance to acquire for the more rare items and bringing it hence up to the 3 LP chance… which is usually not even 1 in 100k. That’s a difference of 10000% acquisition rate for 4 LP items hence.
So overall not as easy as it’s made out to be since LP is a fairly nicely made RNG system.
Which brings us back to causing prices for prophecies needing to be in line with the respective drops as our upside is ‘free choice of pool’ currently.
If you want to then go the next step it’s ‘deterministic choice of the item dropped in the pool’ instead of LP, costing respectively more.
Imagining we get 1 unique for 1k (ease of use for the numbers as an example) and we have said axe as a 5% drop available then the favor cost needs to go up to 20k.
If you then want to even get LP on it it would need to go to the respective LP pricing on top. So for a 15% roll chance? Well… instead of 20k we’ll suddenly have it become a multiplicative thing inside that. Another 15% chance inside that 5% chance. So suddenly it would need to cost at least 300k minimum.
For a 2 LP which sits around 0,5% it’s even worse by a lot, a minimum cost of 4 million favor points.
3 LP (around 0,01%) hence would be 200 million.
4LP (0,0000001%)… well, good luck farming up the 20 trillion (If I didn’t mess up) Favor.
It’s the main reason why it’s a bit of a problem overall.
You can’t just cut off the top-end of whole system easily in any way without automatically flooding the game with those items that aren’t meant to be acquired reliably in any way.
After all to get a 1 LP rare axe in our example already you would need to go for 20 unique drops of the category and to get it as a 1 LP you need to go for ~190 unique drops of the category… so anything beyond already becomes ridiculous to farm out.
LP is a pure chase-type mechanic, only possible to be reallocated in a big community and never in a system which is meant for solo-usage. That’s the necessity of a system which has chase-items… and otherwise you won’t get to chase anything since it’s too close to acquire at any time to be called ‘chase’ in the first place.
A age-long conundrum of how to handle item acquisition in general.
Which is why CoF for acquiring items is good on the lower end but can never ever even think to get to the ceiling of the game. That’s a dream which can’t be fulfilled and was never meant to happen in the first place.
So the only viable solution for the whole issue is to create ridiculously expensive prophecy nodes which cause a simple re-roll mechanic in them. A ‘lucky roll’ mechanic. Rather then rolling once you roll ‘x’ times on the LP value, which means giving them a cost multiplier. 1 Lucky re-roll? 1,4x price added… 2 times? 2,8x price added and so on.
Then we can think about adding a function to allow as many ‘lucky rolls’ as you desire. ‘Empowering’ prophecies could it be called as an example, practically increasing your chances in a linear fashion… but in reality solely causing the amplitude of the wave-function to even out through the rule of big numbers.