Only implement quality/recycler randomness at change of recipe.

[reduke]

At the moment as far as I can see it, random walk is the biggest problem with RNG in factorio, and it also encourages save-scumming for high value items.

I would suggest that RNG is only run when the recipe for a machine changes.


Example for recyclers:

One recycler takes in 1 green chip - at present it will have a 25% chance of returning ingredients

I would change this to having it store that green chip until it receives another 3, and then it spits out the recycled parts of 1 chip - no RNG required.

If the recipe changes, then the RNG runs on whatever it has stored up to that point as per the current behaviour



Example for Quality (less important, possibly harder to implement):

If a machine has a 1 in x chance of a quality upgrade, then it will keep a counter of how many items it has produced, and when it passes each multiple of x, it returns an uncommon item. When it passes 10x, it returns a rare item, etc.

If the quality modules in the machine change or the recipe changes, then RNG will run based on how close to x it currently is and the counter will reset.


Thoughts?



ps - as an aside - has any thought been given to implementing specific quality percentages for every quality level? ie a 1 in 5 chance of producing uncommon would then also mean a 1 in 5 chance of that becoming rare, and then 1 in 5 epic, etc? Instead of 10% being fixed for later levels - I think this would help with balance.

[eugenekay]

Deterministic Recycler implements the “input 4 items to get 1 Output” concept.

next_probability can be changed to affect the chance of a “jump” occurring from each level.

Good Luck!

[mmmPI]

I think there are still some things that are not fully making sense to me after reading. I don't share the premises that random walk is a problem, i think it's fun and part the game too deal with it, but i tried to follow your reasonning about save scumming when or why would someone be tempted to do so now, and maybe i'm wrong but i feel your proposition can still lead to similar save-scumming "incentives".

In a situation where what i want is a "quality mech armor", i can be unlucky and have my first 4 "normal quality", then i can save the game, and put them all in the recycler at once, and reload the save game if the recycler does't give me quality material. ( 9 times out of ten ).

I missed the point ?


Edit : didn't expected a mod for this, nice, makes it easy to verify !

[reduke]

Re save scumming - if a quality upgrade happens 10% of the time - if you make it deterministic not random - then the first 9 items created will always be standard quality, and the 10th will always be uncommon (unless it's the 100th or 1,000th or 10,000th) - so save scumming won't work as you will still have to get 9 standard items first, no matter how many times you reload.

[mmmPI]

that make sense , i had missed some of the implications, it's the randomness both for the recycler and for the quality module that are concerned in your proposition when there seem to be a mod that demonstrate the effect on the recycler (only).

your explanation also makes sense to me for how it could work in practice when it would be combined.

My thought are then that it's a smart/good idea for a mod as i just don't like it for the game, i like the idea of having the tools with circuits and scale to turn what can be felt as random into predictible statistic, i think it create in a game a way for one to teach themselves those, as an illustration of how perspective shift when you are the casino versus a player, or when you farm a lot in MMO and you don't drop things by luck but with a rate you calcutated and monitor and compare location. I think it feels good and it would be a shame to miss on the opportunity to experiment what otherwise may be super costly in time or money.

I think it works well as an incentive to create machines that do not switch receipe "randomly" though, which otherwise only cost "productivity". I think it makes the incentive much stronger and i can understand how it would be "preferable" with your arguments as in "easier" / "less chaotic", that's why i think it could be good in mods, it can act as a reward, for a planet mod, that doesn't unbalance the game as to ruin the "next planet" by triviliazing some things, it can offer a balanced alternative that diversify the gameplay, making the player think which recycling method with their pros and cons 'should' be used after unlocking the "reward".