My point was its actually the opposite of that. In billions or trillions statistically speaking it looks more normal, even my meager tests in tens of thousands of rolls were not far off what models would predict from true random, in small subsets of data lest say somewhere between 3-8 samples, the streaks caused by the patterns that pseudo random is subject to can be painfully apparent.
1 Like
Pseudorandom doesnt mean its going to roll 6 on a six sided dice over and over it means the rolls can eventually be predicted based on the seed though this itself is usually very difficult to find. This would for example be really bad if you were playing poker at a website and had a program to detect this. But for the average user pseudorandom is perfectly fine and is not doing anything sinister in D4. The code of D4 is what is affecting D4 not pseudorandom. The only way pseudorandom could affect D4 is if you somehow could find whatever is making these numbers and then predict the next outcomes potentially letting you reset masterworks at 11/12 because u know it will miss.
Yeah I read the same MIT article.
A six-sided die would actually be closer to true random than most pc based pseudorandom.
I’m not saying there is anything insidious. what I am saying is that the patterns pseudo random is subject to (predicted or not) can easily explain 6 or even 8 rolls of the same stat in a row, and that while it is not true random it doesn’t necessarily need to be weighted to see this outcome.
I dont see this connection. All I see is that pseudorandom can eventually be predicted but it does not seem to mean it favors one number over another. How does prediction equate to not random?
The predetermined nature of pseudorandom makes it subject to streaks that would be far less likely in true random. This explains how 6 rolls in a row can be expected to happen more frequently with pseudorandom even in the absence of any intentional weighting. with lower quality algorithms patterns are more present.
Being predicted does not make it not true random it is not causal, but if you can predict it then it does Indicate that it is not true random.
How though? I am not finding this info. Just that pseudo random can be predicted but how does prediction equate to not random? How does pseudo random lead to more streaks?
predictability does not cause it to not be true random
True random cannot be predicted
THEREFORE:
If it can be predicted it is not true random
I cannot tell you the exact mechanism by which pseudo leads to more streaks as I have only used the functions in languages written by bigger brains than me. I can say I have done tests with very basic pseudorandom tools and observed a much higher frequency of streaks than one would expect from true random.
I will take a somewhat educated guess:
Psuedorandom is driven by predetermined code with a seed, at its most basic level the measure of success it to produce a result that will look like true random over a large set of data. algorithms generate patterns the less complex the algorithm the more pronounced the patterns. patterns can tend to produce higher frequency of repeated results.
Pseudorandom, Random, Weighted - can’t we (almost) all agree that the Tempering system just sucks and does not feel good?
In other games, this might be handled as Enchanting or Augments. Generally, these Enchants and Augments have different power levels with the better Enchants and Augments costing more in materials and being more scarce. This sort of system scales with the player, enabling you to take advantage of the system and save “the best materials” for “the best gear” but does not exclude you from improving items that could be upgrades or even side grades.
This system is straightforward. There is not any RNG in it. You want Rank 1 of Critical Hit Damage, you get Rank 1 of Critical Hit Damage and what that value is. You want Rank 5 of Critical Hit Damage, you get that value.
The Recipes for this should unlock at varying levels. They could even require challenges to unlock befitting the specific Recipe rather than being wholely RNG or something mundane as how Elixirs work where as soon as you hit X Level you click a button and Boom Potions are improved or something (lame).
Adding multiple layers to items making them customizable in different ways is a great idea. Grim Dawn Augments and Enhancements stack upon one another. Each can provide different Affixes or Powers. One set is RNG World found and craftable while the other is purchased via Faction vendors (time investment). Not saying this has to be replicated but basically it gives you your “2 Tempers” in a more fluid and Unique way with complete control over it.
As you acquire gear you are able to move things around more freely in order to min/max. I absolutely hate how in D4 I have an inventory full of potential upgrades but in order to equip them, I have to find multiple other pieces to move things around like some ghetto jigsaw puzzle so I do not:
- Dip below Elemental Resist Soft Cap for any particular Element
- Fall below the Armor Cap
- Lose vital break-points
- etc.
Most of this however does not matter when you have a singular item like Tyrael’s Masterworked 12/12 and a GA for All Resist. In short - when it matters most, itemization is inflexible making progression not feel very fluid in D4. The Developers clearly stated their intent was for itemization to feel like a Journey where you have consistent progression. Well. It isn’t that. You basically go from terrible to fantastic in a few (lucky) clicks of a button. The systems in place to allow for progression are mostly bypassed either altogether or mostly because of their steep entry costs which are more geared toward end-game (when it does not really matter a bunch when you are swimming in Gold and Mats).
TLDR - Tempering sucks.
2 Likes
Once again I have to agree with you. The nerdy computer scientist in me really wants to know why, but the intrepid adventurer doesn’t care, he just wants his rare items to not be destroyed lol
2 Likes
It is fairly amusing as these solutions are in essence rigging the system in a different way and its from 13 years ago.
1 Like
“The gist of it is using statistics to correct unlucky streaks. The way you can do this is to keep track of how many times the event has occurred and use that to bias the number generated by the PRNG.”
This is exactly how i ended up dealing with the rarity issue in my loot table. i track a certain number of results and use that data to influence rolls in the direction i want them to go.
I would love to get my hands on some AI tools in Unity to leverage that to smooth it out even more. As I mentioned somewhere in some other thread my ideal goal for game reward would be to have AI ensure players got enough reward to have smooth progression while not speeding ahead so quickly that they got bored and stopped playing
2 Likes
Haha ‘loopy’ IKR! But waiting for the animation has no affect because you can select the item in your inventory to see what you got while the animation is still playing. The ‘decision’ has already been made. 
1 Like
SERIOUSLY???!!! Sheesh, I learn something new everyday, smdh 
I don’t think I’ve ever done that…I just wait until it was done or just hit that SKIP button coz I was too darn impatient! So what is even the point of the animation when people can tell beforehand!!! What a waste! 
You know a mechanic is bad when ppl start to have cultist rituals about it…
4 Likes
I don’t know about the odds being exactly stacked. Instead, I think there is a glitch in their code that causes results to repeat. I suspect that is why people get the same bad tempering result many times in a row.
Of course! what was I thinking. It can’t be a flaw inherent to computer generated RNG it must be bad code!
1 Like
Some of the tempers are definitely weighted. It’s pretty obvious that that’s the case, they weight enchantments, you really think they wouldn’t weigh tempers? The thing is with the rerolls, you never get the good tempers multiple times in a row. The fact that you can roll the same thing easily 5 times in a row proves this. Roll a dice, I bet you can’t roll the same number 5 times in a row as easily as it is in tempering. My longest streak of the same roll in a row is 21, that should mathematically be almost impossible if it was proper rng. If it wasn’t weighted it would feel like masterwokring where it’s extremely hard to hit the same affix multiple times in a row. The fact that everyone complains about it, also is just the easiest proof.