[Season 4] Wow! There's no RNG on rolls. It is weighted!

[TaigaAisaka-1807]

Did we get any reason why they are pursuing this weighted rolls?
If they are trying to have us spend more time on their game then they are doing it wrong.

We are already half way this season and I have not seen any amulet with 2 rank skill rolls.

S1-S2, I have seen this a dozen times on amulets. But now? you’re even lucky to get 2 ranks skills in one. Probably because we have stats that are in general pool. (e.g. health)

Why not stick to the old groupings stats for gears. Like only weapons will have offensive stat (But to cope up with what stat we have, they just multiply it by 2 or 3)
another example: only torso,pants,helmet will get health that provides 2k base hp.

Inshort everythings got messed up on “ROLLS” because we have stats that are in general pool of affixes/stats.

[leusebi-1647]

@Hallfrom-2984

If this is a random number generator, then there should be an equal probability of getting any. numbers. From a scientific point of view, this is not even discussed.

So the numbers generated by rolling 2d6 are not random?

@Hallfrom-2984

What are we talking about besides the pattern that produces some parameters 90% more often?

We are talking about the design of the item system and whether it’s better to have a higher chance of seeing certain affixes or to always have every affix have equal weights.

You clearly prefer to insult people rather than actually read what they write, but being an idiot who left college after only a Master’s degree, I guess I’ll try again:

The random number generation is a step before the selection of affixes. Like an encounter table in D&D. Here are two possible tables:

Table 1, Random number generation: roll a d12
Affix selection:

1. Damage to anything that isn’t red
2. plus to Frozen Orb
3. plus to Incinerate
4. plus to Core Skills
5. Damage on Tuesdays
6. Damage
7. Max Life
8. Main Stat
9. Damage during a Full Moon
10. CDR
11. plus to Ice Shards
12. plus to Fireball

Table 2, Random number generation: roll 2d6
Affix selection:
2. plus to Frozen Orb
3. plus to Incinerate
4. plus to Core Skills
5. Roll again
6. Damage
7. Max Life
8. Main Stat
9. Roll again
10. CDR
11. plus to Ice Shards
12. plus to Fireball

Both of these use random numbers. In the first case, the weights are the same, in the second they are not. Some outcomes will occur more often than others, even though the random number generation is 100% fair. There is nothing unethical about this. The purpose can be twofold:

1. Shift probability away from narrowly-applicable affixes like + to a specific skill onto generally-applicable affixes like Max Life and Main Stat.
2. Replace bad affixes with better ones without also increasing the probability of the absolute best ones.

[Kubikluc-2412]

Enchanting costs gold and it can be extremely expensive. Drops are free. It makes sense for me to have different weighting for each. Obviously when enchanting we are not trying to get the affixes that are already available on most of the drops.

[Morfyus-1128]

I mean welcome to the real world.

[leusebi-1647]

@Kubikluc-2412

Enchanting costs gold and it can be extremely expensive. Drops are free.

Gold is also free. I agree that the amount of gold it takes to get a specific affix combo should represent less farming time than trying to get it to drop. But:

If an affix has a 1% chance, then the odds of getting it along with two affixes you want are 1/100 the odds of getting those two other affixes and a third affix you plan to reroll. Each time you use the enchanter, you get just shy of a 2% chance to see that affix. So on average you will find one about every 50 rerolls.

For non-GA items, that means that a decent average reroll cost would be twice the sell value of a legendary of equal item power. If that was the cost, it would be more efficient to reroll a drop with two good affixes, as long as you could find more efficient sources of gold than just selling legendaries (which you definitely can).

For 1 GA items, though, the cost needs to be much much higher. I’m not certain what the ratio is, but let’s say that 1/40 affixes roll as GA. That would mean that if one of the affixes you aren’t rerolling is GA, the cost per enchant needs to be about 20 times higher on average (because either of the two could have gotten the GA). You could discount it a bit, but a 20x discount would just make dropped items with 3 good stats pointless (unless they were 3 GA items).

And, if you balanced it precisely around 1 GA items, you’d be underpricing it by a factor of 80 for 2 GA items. Again, you can give a discount beyond the efficiency of gold farming, but that big a discount really distorts things.

So let’s say that most of the items being rerolled are 1 GA and some are 2 GA and very few if any are 0 GA. And let’s also say that people are more likely to try to maximize the 2 GA item vs the 1 GA ones, so about 1/10 of the total rerolls are happening on 2 GA items (instead of 1/80). Those ones should cost 3200x the sell price of a 925 legendary, on average, while the 1 GA ones should cost 40x. So the overall costs should be 356x, to make it completely equal to the balance on drops.

So somewhere between 15M and 35M depending on the slot, as an asymptotic cost. This would be a huge discount for rerolling 2 GAs but would only be worth rerolling a 1 GA while the cost is in the 200k-500k range, depending on the slot.

[Soluss-1870]

@leusebi-1647

I think if you want to see specific affixes more often, then the best answer is to have the drop level of an item shift the weights away from the more general but less useful affixes and towards the more narrow but more powerful ones. So an item that drops from a level 120 mob is more likely to have the best affixes and an item from a level 200 mob is even more likely to have them. Maybe it even goes beyond fully-equal weights at some level and makes it more likely to see a drop with, say, CDR, Crit Chance, and +3 to skills than to see Max Life, +Main Stat, and +X to a resist.

This is actually what I want and have been advocating for. Better loot drops from harder content. That would give a better sense of progression.

Current progression system has always felt off to me. Anything can drop anywhere. That’s bad to me because you can be in end game gear right after hitting wt4 if you are lucky enough and then spend the next I dunno 30-40 levels just grinding levels and paragon. Now all gear drops are boring. It makes the rest of the game feel like grind for the sake of grind. Yeah I guess now you can hit jackpot 3 GA gear but that’s not enough. Especially with the low drop rates. This is why gear doesn’t feel exciting. Cus you have the best you can reasonably expect from like level 70. The last 30 levels is almost just a waste of time.

Just my opinion.

[zfz-3619]

@leusebi-1647

The only really bad affixes are things like Life per second and Mana on kill, which are bad mainly because they are way undertuned.

Erm…they are undertuned, which makes them bad. I don’t know what you are trying to argue.

If the Maximum Life affixes were all limited to +20 Maximum Life at most, then Maximum Life would also be a bad affix.

If you are really that unwilling to use the term “bad”, let’s just call then “unwanted” or “undesired”.

[leusebi-1647]

@zfz-3619

If you are really that unwilling to use the term “bad”, let’s just call then “unwanted” or “undesired”

I guess my point was that there are affixes that are just inherently bad because the thing they do is never relevant and if it becomes relevant it’s really annoying. That’s what I’d call a bad affix. Only impairment reduction really fits in that category right now.

Then separately there are affixes that could be good if it was possible to build around them, but no classes currently have enough tools to make those builds work. I think life per second and life on hit fall into that category. There are reasons they could be used, it’s just that the whole build isn’t really worth it.

And then there are some that could be used on current builds if they were more powerful. Mana on kill might fall into this category, especially if they introduce an armory where you can easily swap out when farming helltide vs content with low-density bosses.

[zfz-3619]

I think your two points are really the same point.

IMO any affix can be made to work and be good, and any affix can be made to work in a really annoying way. It’s all up to the final design as a whole, and affixes cannot be evaluated in a vacuum.

Impairment Reduction, for example, could be overhauled to present breaking points like Armor. At x% IR, your movement speed is no longer affected by slow/chill, and at y% IR you become immune to crowd controlling effects. Then it become something like armor and resistances where you aim to cap. Now put IR in paragon boards and suddenly it is just another resist to cap that provides extremely high return.

Overpower is an affix that works but IMO is annoying, because the tools that make it work isn’t as scalable as something like critical strike. (You can make every hit a crit, but you can’t make every hit an overpower.)

For on-kill effects, I think they will forever be irrelevant because you never want any on-kill effect for boss fights. And in any situation where on-kill effects would shine, it would also be irrelevant because it means you are killing very quickly anyways. So to make it work, the overall resource generation would have to be nerfed into the ground to force people to take resource gain on kill.

Because it all depends on the design of the game outside of the affixes themselves, I don’t think there is really a point in defining what is a bad affix. If it’s not wanted by anyone and not desired by anyone, then it’s bad/unwanted/undesired and it acts like a padding in enchanting.

It doesn’t matter that an affix might have potential to be good. It doesn’t matter. It is still a bad affix if it’s unwanted.

It doesn’t matter if I’m getting rolls of “+123 Life Per Second” or “+3 flowers spawned nearby”, they both serve to prevent the vast majority of people (if not everyone) from getting the affixes we want. So they are both bad affixes used to pad the RNG.

[Hallfrom-2984]

@leusebi-1647

So the numbers generated by rolling 2d6 are not random?

This is already a theory of probability, but in most cases it is not. But it takes too long to explain, it takes even longer to search for articles on this topic. There is no generation of numbers here, and there are a lot of factors.

@leusebi-1647

We are talking about the design of the item system and whether it’s better to have a higher chance of seeing certain affixes or to always have every affix have equal weights.

It’s just not worth talking about this anymore, stop being a white king and you won’t be torn to pieces by a crowd of peasants. Maybe you haven’t noticed, but the majority of people don’t like it, and no one cares a little that you (apparently from another account) are trading RMT.
In any case, we already have the very rarity of items. And either they will add another type of item where affixes of higher rarity will drop out and there will be no affixes of low rarity.

@leusebi-1647

You clearly prefer to insult people rather than actually read what they write, but being an idiot who left college after only a Master’s degree, I guess I’ll try again:

Yes, I don’t deny this, all people are idiots in some area and geniuses in some other, it’s just that most people don’t bother with it.

@leusebi-1647

The random number generation is a step before the selection of affixes. Like an encounter table in D&D. Here are two possible tables:

This is what I call fraudulent gaming policy. The answer here is simple, either they introduce a separate type of item for each rarity of the affix and separate them, or let them equalize them, otherwise it turns out that they are cheating and deliberately humiliating some players over others using hidden mechanisms. This is not only dishonest, it is intentionally misleading and condones discrimination. I thought the players were all equal, but it turns out that they have a double market and some get everything and some get nothing. I believe that these are precisely the classes of accounts.

@leusebi-1647

We are talking about the design of the item system and whether it’s better to have a higher chance of seeing certain affixes or to always have every affix have equal weights.

You clearly prefer to insult people rather than actually read what they write, but being an idiot who left college after only a Master’s degree, I guess I’ll try again:

The random number generation is a step before the selection of affixes. Like an encounter table in D&D. Here are two possible tables:

Table 1, Random…

1. Shift probability away from narrowly-applicable affixes like + to a specific skill onto generally-applicable affixes like Max Life and Main Stat.
2. Replace bad affixes with better ones without also increasing the probability of the absolute best ones.

Here you are showing me how you should not program a RNG generator for random numbers, since such a system will not work without serious failures and a heavy load. By the way, frequent failures occur due to the fact that actions with the interface are made by requesting the server, and if it is at the peak of loading, the packet is lost and you are stuck stupidly without receiving a response. You have to close the game and open it again.

[leusebi-1647]

@zfz-3619

Impairment Reduction, for example, could be overhauled to present breaking points like Armor. At x% IR, your movement speed is no longer affected by slow/chill, and at y% IR you become immune to crowd controlling effects.

Yeah, I think this would be a good way to go. Or maybe just give players a stagger bar like bosses and have IR act like a resist for the stagger (plus have some stagger drain away naturally every second). Still, if being crowd controlled is common without sufficient IR, I think that would become annoying.

@zfz-3619

For on-kill effects, I think they will forever be irrelevant because you never want any on-kill effect for boss fights.

Well, the Blood Maiden is an example where that isn’t true. So if they added more overworld boss fights like that then those would be potentially viable for on kill effects. It’s true those that these are always going to be hard to make relevant.

@zfz-3619

It doesn’t matter that an affix might have potential to be good. It doesn’t matter. It is still a bad affix if it’s unwanted.

It’s bad and unwanted but it isn’t padding if the game is trying to interact with it and the interactions just lead to a bad build. There are quite a few interactions with healing that could theoretically create a build or two. You could literally make janky builds right now that leaned on healing. I think that makes those affixes fundamentally interesting additions to the game even if none of the meta builds current want them. Overpower is an interesting addition whether there’s an S tier overpower build or not.

@zfz-3619

It doesn’t matter if I’m getting rolls of “+123 Life Per Second” or “+3 flowers spawned nearby”, they both serve to prevent the vast majority of people (if not everyone) from getting the affixes we want. So they are both bad affixes used to pad the RNG.

I disagree because of the reasons above, but also it isn’t clear they even have the effect of padding the RNG, since the affix selection is already weighted. There’s no reason to think removing them would up the chance of anything other than life or main stat.

[leusebi-1647]

@Hallfrom-2984

no one cares a little that you (apparently from another account) are trading RMT.

@Hallfrom-2984

@leusebi-1647

The random number generation is a step before the selection of affixes. Like an encounter table in D&D. Here are two possible tables:

This is what I call fraudulent gaming policy.

@Hallfrom-2984

either they introduce a separate type of item for each rarity of the affix and separate them, or let them equalize them, otherwise it turns out that they are cheating and deliberately humiliating some players over others using hidden mechanisms

@Hallfrom-2984

such a system will not work without serious failures and a heavy load.

OMG what in the world are you talking about? Rolling dice and checking a table won’t work without serious failures and heavy load? Rolling 2d6 is humiliating certain players and discriminating against them?

Generating a random number and checking a table based on it is literally the only way to do this. It doesn’t affect performance at all if some affixes have multiple entries in the table and others don’t. It doesn’t affect one player any differently than any other.

[Hallfrom-2984]

You also cannot claim that we always have cubes on which all the values are present.
Yes, it can be argued that there are 2 RNG machines themselves and one has rare values, the other does not. In addition, characteristics are assigned not when an item is dropped, but when it is first viewed in the inventory. For some items, it takes more than a minute to assign parameters, although based on your statements it should take no more than a second, since the parameters are taken from a bloated table in which 90% of the 4 affixes are repeated.
On the other hand, the mechanism when one option has more values than the other is already called a dishonest sweepstakes. Either they need to separate items by rarity, or accept that the rarity should be the same. We already have bad items in the game (white, blue, gold), but one more point won’t change the system much, it’s just that players will waste less time sorting through them.

[Hallfrom-2984]

Let’s walk easier. Hardening. In theory, all 4 values have equal chances and cannot be determined based on a bloated table (this is impossible in principle) since the system initially knows what parameter you are looking for and gives you any other one or the one you need but with a minimum value.
Now prove that it is honest. I am saying that this system does not work as specified and there is strong evidence for this. The reason for RNG’s failure is bullying on the part of the developers, since no one else can interfere with this system. Anything to add?

[Vlad-28576]

W8 for new seasson. Its always better play seasson.

[VoodooVirus-1819]

Look I have been playing LE on and off since it was in alpha I was playing shortly after it ended its kickstarter. The devs certainly have passion, but they too have had their issues persisting that have been well documented for years which their ‘audience’ has been asking for. i.e. Forgeguard has been a meme build for its entire existence, where the only functional builds are those that don’t use Forgeguard skills.

Fortunately, we might be getting some of those issues addressed finally in this upcoming season (cycle).

[Beefhammer-1932]

They are not. We had a poster post tesults of like 3k tempers and they were pretty close to being completely balanced. You just haven’t seen enough if a sample size.

[Beefhammer-1932]

They need to do something about endgame. It is severely lacking and boring as hell to me.

[Legacy-2775]

Looking forward to PoE II.

To summarise Diablo 4:

Player = Donkey, Loot is the on the
You can play the whole season and not get a single good 3GA item.

[Kubikluc-2412]

Took me while to get a pair of gloves. +Pulverize affix. Tried several times at the Occultist but just wasted the gold I had. Finally managed to get the gloves from a drop. Would need +Envenom on my amulet. Have 3 amulets in my stash with this affix. They are the only ones I managed to find since the start of the season. I kept them but they are not really usable since the other affixes are bad. Again, spend 100s of millions trying to enchant the affix but failed. Amulets is still how I spend all the Obols I collect. I believe many players have similar experiences.

Don’t get me wrong, I am already satisfied with the build I have and the quality of its gear. I already managed to get all the achievements I was aiming for this season. The rest is just bonus (including on gear). I won’t be pushing the Pit anyway with my Bear.

I understand weighting and I agree that there must be weighting for the game to work properly. I just think it is currently not implemented correctly, at least for some of the game features. There is something wrong when the game makes it easier to get Uber uniques than the basics affixes a build needs to perform efficiently.