you have to know that D3 was already under maintenance mode, all resources poured into D4 and D:I.
And the fact is even this method implement on D4, it won’t make sense because it’s simply increase the workload for each patch note. A good balance system shall be last long without much of changes while able to fit into any new system/item.
D3 already can’t balance all legendary items and you can’t expect they can balance/implement this type of method.
I’ve said this many times already. The claim that you need to kill trading to get rid of real money item buying is total nonsense.
There’s 2 ways to fix it:
Diablo Immortal is already doing the first idea. You can trade, but the buy-sell price is determined by the market, you can’t change it.
´´Our Free to Play Approach and Values — Diablo Immortal — Blizzard News
The Market connects anonymous buyers and sellers, with no ability to cash-out. There are limits to what can and cannot be put on the Market. While it will be a possible source of certain materials, supplementary items, and Legendary Gems, it is not a place to acquire gear.
“…not a place to acquire gear” part really makes me think. That means entire game is self found mainly. As long as sold items on the market have less variables to combine for, permutations allow you to keep a fixated price without any flow.
I don’t think there’s a system to determine a 6-7 random variable item’s true price unless we’re talking about learning artificial intelligence. Again, you can not prevent token items if you were to try.
Whatever you think, you’re not ahead of the system; stat range rolls and demanded stats will get ahead of you in the itemization. When you try to curb this down, you make a mess out of trash items and trivialize everything to affect difficulty scale and mechanic skills downwards.
Even if you make everything equally trash and only allow players to modify them later, some items will get ahead of others in demand by maximum stat roll thresholds according to the difficulty. If item-X is highly demanded because of affix-a and a possibility of having highly soared affix-b together, I’ll look for affix-a over 100 range to get a good deal out of it.
System can not learn this unless programmed which means extra resources in the server, neither you can force a price if you were to talk about something so whimsical as multiple variable combinations. There’s huge difference between a single affix charm or jewel to the multiple variable equipment. Please get that “6!” is bigger than “1!”.
Yes, I’m talking about that.
It applies to gems, but it can also be applied to gear.
And yes, the problem of working out the market price of items can be done statistically based on the prices of similar items and affixes.
Alternatively, you can force items to be turned back into commodities before they can be sold, like gems. For example, a Deathwish is sold as “Deathwish with high %Damage roll”, “Deathwish with high Attack Speed roll”, etc., with all its affixes rerolled.
I think we got over this… Who decides that and when can you curb the limit is still questionable in practice. I’m not sure if I ever get through you or you got the glimpse of this idea. If you try to force a price the flow will change, you can not keep it stable, you’re talking about keeping it stable in a vague manner. How would that even work?
Learning artificial intelligence still have to present for this to be any practical which means extra resources. Market will still go crazy for artificial intelligence to learn anything. Do you understand that?
Your imaginative closed testing of hundred players will NOT go ruly for millions out there. Again, you can not entirely dart a price for one item by looking at another item and ignoring stat roll ranges or combinations. Price can inflate unwarrantly if this were to apply.
People will be willing to pay more for max stat roll or anything that might benefit them more over a threshold. You’re ignoring the human interaction here. If people want a gg Windforce, they will pay for that but price has to flow so it wouldn’t be a forced grind for modicum amount of players.
https://www.investopedia.com/terms/r/regression.asp
Okay… I’m willing to learn. Let’s say you have unlimited supply (digital items) would those rules that go into a real world trade regression still would apply? What’s the “GDP growth” stand for a player in this situation or the item itself? Also I kinda can not see the relation. Help me here. Open market will not work for anything anymore, digital tax laws changed long ago. Give me and example how that regression model applies or means to a player.
If you think you can force a cap, again, you’re not ahead of system; high stat classic items that rolled a godly combination sold for million dollars in RMAH. What are you trying to cap or regress here? Anything similar but with lower stats only went for hundreds of bucks. What can you predict from a wildly changing combination of variables?
That still demands, machine learning, and there must be a ratio between different items’ prices which can be unstable. If you eliminate all the wild cards, you still go back to square one where you’ll have a model very close to the D:I to the point of making gear trades meaningless.
Yes, it’s basically “machine learning”, the price of an item is determined by the number of factors like how much crit damage rolls for, how much crit chance rolls for, whether there’s a combination of crit damage, crit chance, attack speed affixes on it, etc, which can be worked out with a large sample size of prices, and then used for new items.
Or alternatively, you can turn items into commodities like gems with a small number of fixed listings by rerolling the affixes on it.
That would make it fairly useless though. Basically gambling. It is the affixes that will determine whether it is a good legendary or not.
This is popping up more and more these days.
How to balance the game? Just let an AI do it!
How to make the game easy to learn? Let an AI tell each player which gear they should equip!!
How to determine item value? Machine learning!
Feels like meaningless buzzwords.
Looking at what people are willing to trade an item for, is not an indication for the items value (which is certainly true in capital investments as well, though not really comparable). As said earlier in the thread, if the trading system requires you to always trade one item for another, people can just trade awesome item A for trash item B.
Your data would now show that trash item B apparently got the same value as awesome item A. Even though it doesn’t. The system can’t take into account external payment through RMT, nor people giving each other items for free. Or complex trades where you pay to buy one item, make 10 different trades Skewing your data.
Not to mention all the issues that will emerge each time the game balance changes, and you would have a time gap between the items real value changed, and the data catches up, adjusting the cost in the trading system.
As for gems, crafting mats, and similar (likely not runes though), yeah, sure, you could have a system with fixed prices for those based on their in-game “value”.
And yeah, I bet systems could be made, for making a better trade system for all the gear too. Just not anything that requires machine learning or whatever to guess item value.
How are they doing it exactly? Fixed item price based on what?
I just saw it’s for some common items only.
Market rules are at 5:00 min
Vendor trading like in PoE is better than this.
AI vendor trading is what D4 should have.
Any trading in Diablo will be an upgrade. I for one, did think the AH was diablo’s finest work, attempting to bring an MMO aspect back to Diablo.
I actually made a java-based program that was not too difficult to use that could estimate and respectably so – which stat points (including damage and crit values) were the cheapest per unit of dps to upgrade. Lo an behold it was crit damage that was the cheapest, then crit, then raw damage.
All in all, I think that customization is more important than other things.
But I really do miss trying to get my first shako with pgems and high rolled leveling uniques.
they were not too far off with the AH in vanilla D3, the problem was always the RMAH.
if they wanna make trading viable without the chance of some 3rd party sites appearing, is if we get an AH again and that can be a good thing!
Pretty simple, make all gear that drops only tradable with party members for a duration of 2 hours like we have now in d3, if you dont wanna trade it but instead sell the item you have the option to put it up for sale on the AH instead.
But NO trading outside of the party beside AH.
Keep it simple. If any trading is supported, make item account bound upon trade. Zero flipping.
And the number of items that can be traded per 24 hours must be limited too.
What’s wrong with AI helping to create a more balanced trade system? AI can be used for tons of amazing things. Sure it will eventually lead to skynet rising and taking over, but for now we can use AI to help create awesome software that can do billions of calculations in seconds and not have to hire any additional employees.
Also, i think a lot of the anti-trading people are just always generalizing and claiming that ALL pro-trading people want 100% free trading when almost no one wants that. We do need highly-restricted trading such as the ideas laid out here to make both sides happy:
These should keep potential botters from passing items to alternate characters or accounts as well as keep them from being able to make any real money from the lower level stuff. is that enough? What else does it need to prevent p2w and 3rd party sellers?
With a restricted system (which is planned already), you don’t even need it; people can easily watch the flow and things won’t go too crazy. Without restrictions (tough chance of that happening), even machine learning can never estimate anything properly.
Also, including A.I still means extra burden on resources for server. I rather have it catch botters, not regulate a free flow market. For machine to learn, you have to remove some restrictions on market or run a simulation and that won’t fly well as you can not dictate a limit about anything from small examples.
If something is more valuable than something else because affix combinations or desired stat roll thresholds, then it will be more valuable. You can’t exactly prevent this just by reducing every item to be equal, forget it. There are 100+ affixes in possibility pools in a standard ARPG with varying weights.
Forcing a system where everything reduced to rubble or trash on an equal level, impacts the difficulty scaling or combat pace.
Maybe you have a brilliant idea about allowing the trash item to be modified by the end owner to make it better, but same concerns will apply.
When you restrict the very high and very low grade item circulation, do you really need a machine to work it up for you?
Wont really help much imo. Sure, it helps against flippers, but that is about it.
Now this could help. Though I doubt we would agree what the limit should be. I’d say like 1 item per month. Or maybe tier-based: 1 unique/legendary per month, 3 rares, 10 runes/gems. That way you will only trade for the items that really matters. (and, only allowing trading within those tiers; unique/legendary for unique/legendary).
That it seems pretty unlikely that AI could help creating a more balanced trade system. How would you do it? AI is not magic.
Perfect rolled as in highest numbers in the range (like 20 on a 15-20 strength affix)? Or as in rolling a combination of the right stats?
You can definitely make data analysis determine whether an item has rolled high numbers. Problem is, afaik, Blizzard is going away from D3s stupid affix ranges (and if not, they really should do that). So high or low rolls wont matter so much anymore.
Which means the most important rolls are the ones determining which combination of affixes you get on an item. The exact thing that is hard to figure out the value of.
An item with max rolled numbers (affix range) but all the wrong affixes, is still a trash item, even with those max rolls.
So limit the usefulness of an in-game currency, due to the harm trading will do. And people wonder why some of us do not like trading. I mean, I agree, if gold can be traded, then it should not be very useful in the game. It would seem better to not make it tradeable though, and make it more useful in the game.
I dont really care much about the PW2 and 3party sellers. They should be fought with the ban-hammer.
Trade restrictions should be about all the “legit” trading between players. To ensure the game does not turn into trading tycoon, and that finding your own items is the primary way to play. And to ensure everyone arent gearing up with D3 speed, ruining the gameplay.
wait, they won’t have affix ranges in D4? i didn’t see anything about that in the itemization update 
. So if an item rolls with strength and wisdom on it, it will always be “15” or whatever predetermined number based on the level or something?
and the “right” affix’s is where the AI can come in. It can determine if an item has the “best” possible combo of affix’s. but who cares, i was just saying AI can be used for stuff, they most likely won’t do that anyways.
Well maybe YOU don’t personally care about p2w and 3rd party sellers, but i hear a whole lot of whining from the anti-trading crowd about mostly those 2 things. They also don’t want it to turn into a game that requires trading, i get that. But their biggest complaint are the botters/cheaters.
and i agree about gold, make it very useful and just prevent trading it sounds better. If you HAVE to farm your own gold as well as your own end-game gear, i don’t see how illegal trading could flourish at all. Especially if they actually actively ban botters.
No, it will certainly still have ranges. I have just gotten the impression so far, that Blizzard understood the 250-500 kind of ranges from D3 was really bad, and that they are making the affix ranges smaller. Time will tell though.
How would an AI determine that? You would need to teach it somehow.
You could certainly collect data from leaderboards, on what items people are using, and what is being traded, and for how much.
But 1) Leaderboards are lemming effect based. Even they do not necessarily tell the story of an items usage value. They mostly tell you which good builds (and gear) have gotten the most attention on twitch, d3planner and whatever.
So throwing some data analysis on top of that will just emphasis what the market is already doing on its own; increase the cost of the most popular items, making the items that are 1% worse (or merely just percieved to be 1% worse) turn into vendor trash. The opposite of what was the goal of determining items value.
2) Looking at leaderboards and similar is kinda doing things in the wrong order. You are then trying to determine the value of items after people already traded those items. Defeating the whole point. It would always be playing catch up.
Leading to 3) Suddenly a new patch is released, and the meta changes. The items that were really valuable before, are less valuable, and new items become valuable, and the whole catch up starts all over again.
4) Players could feed bad data into the system. Like trade trash items back and forth between each other millions of times at high prices, to make them seem more valuable. The system could probably be coded to try and filter some of it, but it is still left open to data manipulation (which afaik is a very general problem with machine learning)
TDLR; I still dont see how AI can solve any kind of problem in Diablo 4. Neither in trading or any of the other areas where it is proposed.
Maybe AI can be used for something useful. But people should be specific imo. Right now it often seems to be thrown around as a meaningless word. I’m surprised we dont see more posts about how blockchain can solve Diablos issues too…
Some do. I see plenty of people complaining about trading as a much more fundamental issue though. P2W and 2rd parties is just taking it to the extreme, so they are the most obvious examples to use. Pretty much the same issues remain, even if you somehow got rid of RMT though.
AI does what a top player does when deciding between gear choices. That’s it. It gets its data about meta from top players and teaches itself about item comparisons Season after Season.
The first two weeks of each new Season AI trading/crafting is disabled. That’s the period the AI gets more data from how top players react on the new patch. After Season 10 this period will be one week since the AI will have enough data to adjust faster.
AI should also be used as a helping mechanic for new players teaching them basic stuff.