Definitely.
Balanced doesn’t mean you are equally good at everything. Neither in RPGs or RTS. Heck, many RTS’s are based on rock-paper-scissors, so it would be quite horrible to balance those.
Completely fine if one build is weak against fire enemies and strong against frost enemies. and singletarget vs. AoE etc. Those things should not be balanced.
But perfectly balanced as in having hundreds of builds being equally viable in the game overall, with their wildly different strengths and weaknesses, that would be a good thing.
It will always be so as long as players can copy/paste stuff - you’ll have illusion of choice and illusion of being clever. This can change only with a next gen randomization approach that makes all the variables unique for each player changing these each Season. Needless to mention the skill cap hits the sky then and you’ll have cry babies all over the place since they can’t copy/paste the meta, but you’ll be rewarded for being clever and finding the meta in your own context.
You have power tiers regarding builds with builds in each tier balanced among each other. The optimal efficiency play is to hit the top tier build for each activity (related to D4 - each Key Dungeon) as fast as possible. So, you have both balancing and a reason if you aim efficiency. If you aim fun, well yes, you play whatever you want - no need to change the default skill if you have fun with it.
Does the AI decide which build goes into which power tier?
And if it does, would that mean you could have a scenario where a build shows up; it does 10% too well vs. the highest power tier. Sadly, the highest tier already have too many builds in it, so the AI nerfs this build not 10%, but 80%, since tier 6 is the one that is too low on builds?
Well, hopefully respec costs kills that optimization strategy.
We are somewhat back to it no longer being a game, if the rules are constantly changing. I mean, sure, the rules are there, but each player has different rules.
Not to mention, the complete mess that would bring to multiplayer.
I agree but if we could get some kind of trading without incentivising RMT I think it would be nice not to alienate the D2 community.
I don’t know if the D4 team has evolved on this since this interview but this is what they said back then, at the 10m50s mark:
No Auction House.
Lower-tier items would be freely tradable - consumables, crafting mats, even some armor perhaps.
Some items could be traded once and then become BoA.
The highest tier of powerful items would be BoA. But like an accordion, all of the tiers in between are still being discussed. They haven’t decided on anything yet. There’s a big tier gap between the lower tier and the top tier.
That was kind of the motivation for the post - to see if something in between could be done that doesn’t either alienate free-trade D2 or alienate BoA D3.
No. The devs should set the margins, at least until they design the items. And it won’t matter how much a build is nerfed since the player should theoretically always be able to switch to other build easily.
You’d simply roll new characters then (no 1-40 tutorial each time please).
That’s fun, no? If a player can also set his Season margin himself as it should be (1-3 months for example), you’ll have much greater player participation since if you don’t find your current meta fun you’d do it (or a non-meta) only for a month, then you’ll start a new Season with new variables.
That’s the future, really! If Diablo 4 doesn’t do it, another game will do it.
I thought you defended more non-MMO/role multiplayer. That would be the case here with the extra fun factor of witnessing two same skills from different players having completely different damage/results, but not because one player is P800 and the other P8000.
Sure, but unless Blizzard is very incompetent, you hopefully cant just switch to different characters to do the different key dungeons. That would ruin the concept of strengths and weaknesses for a build. Just like free respecs do.
For example, I would very much make keys for key dungeons character-bound.
And no account-wide “paragon” would also mean that you cant lvl your WW barb, by doing a key dungeon on your Throw Barb. Etc.
I hope it is another game then
Choosing your own Season length would also be problematic for multiplayer. And not fit content patch schedules.
I dont think I have argued for that.
I have said that all builds in an A-RPG should focus on dmg, and any support should be self-support. No support/healer/zDPS roles.
Multiplayer is still a mess if you have no idea what skills, items etc. do for others. Makes the gameplay incomprehensible.
Well, if there is a ladder, you’d be required to do all KD with one character to rank (so respecs could help you if you farmed the resources for respecs and invested these), but I don’t see a problem a player to have X number of characters, each for a specific KD he aims to farm optimally.
You’ll still have restrictions to fit these. And it won’t be problematic for multi since you’ll have “waves of players” starting each Season at the same time.
You start your Season, play 2 months and stop, then start new. I play 1 month, start new for a month and again start at third month. We play together the launch of third month same patch before the new content drops next month when we should restart always since the cycle of this patch ends.
Flexible/custom determined Season length is the way to go.
Under the current way of doing things you’re still rewarded for being clever. There are plenty of people who don’t simply look up the best builds online, especially in non-competitive single-player or co-op games.
Also the “randomization approach” is just a roguelike with extra steps. It’s a great system for when playthroughs are relatively short and you’re constantly starting over.
It doesn’t work very well for a Diablo game where a player might want to play a character for 1000+ hours over the course of many sessions. The fact that people will make online guides isn’t even worth worrying about.
Builds already have power tiers without the AI.
The more you try to address the issues brought up with the AI the closer you get to designing an AI that just does the same job the devs would have done without wasting all the resources developing and maintaining the AI.
At which point you’ve wasted a whole lot of time and resources getting nothing practical done.
As we said in the other thread: It’s like you’ve found a solution and now you’re desperately searching for a problem because you think your solution is really cool.
I agree with you such randomization is good especially for Seasonal structure, but those characters moving to NS could still keep their defined variables. Yes, at some point NS would “break”, but the main focus in aRPGs would always be the Seasonal aspect.
PoE and D3 are TERRIBLY balanced aRPGs. AI would do x10000 times better. I find it cool, yes.
At least roguelikes doesnt change the rules. A fireball does what a fireball does in those. You might just not get a fireball spell in each run.
Ruins the basic concept of making builds imo.
If you can just ignore the weaknesses of your character, why bother having characters.
I mean, of course you should be able to have multiple characters optimized for different end-game content. The benefits from doing stuff on one character just shouldn’t spill over to the other characters. Well, outside of gear that is. The game should strongly encourage/force you to do varied content on the same characters.
Which btw is a very good reason for having a forced campaign as part of lvling!
It really shouldn’t. Seasons is a “new” concept pretty much only used by 2 A-RPGs, out of hundreds. And sure, it has its positives, but also so many negatives.
D3 is terribly balanced. PoE seems reasonably well balanced*. Obviously far from perfect.
However, D3 is not badly balanced because it is not doable without an AI. It is badly balanced because Blizzard is not trying to balance it.
Either because they dont want to spend the resources, or because they have a stupid idea that creating new flavor of the month builds each season is “fun” (there was seasons being a problem once again), or likely both.
*Okay, let me rephrase about PoE. PoE seems to be reasonably well-balanced in terms of having plenty of viable builds. It is horribly balanced, like D3, in the sense that the game is an AoE-fest.
When I have looked up info about PoE, it seems like it has quite a few viable builds for all of end-game?
Quite serious. Them struggling with balancing sets in PTR says everything about how few resources they are throwing at it. They are not trying much. Then after PTR is over, they get pushed a bit, and end up quickly “balancing” a set that they supposedly couldn’t get right during PTR (well, balancing is a bit nice, considering their balancing tends to be off by like 1000% or whatever)
Not having only 2 week PTRs would likely help too. Which is again about resources.
Both in the reddit post and in the SC2 DeepMind project, APIs are exposed in the game itself that lets AI, through scripting, control almost everything in the game. It allows programmatic control as if a player was using the game.
Part of the power of this is that computer simulation is so fast. For example, in GTAV, companies like Uber-subsidiary Otto used self-driving AI to control the cars in the game itself. It’s 1,000s of times faster in GTAV than in reality. They can iterate over and over so quickly with code changes and see the results. The physics and street design, stop lights, etc are sufficiently powerful in GTAV for their testing.
With D4 balancing and trade - they wouldn’t have to change things in real-time. They could do it per-week or per-season or w/e. But the capability would be there. Think of it like meta-rules where the meta-rules are at a higher level of abstraction than the rules themselves. Like in Chess a knight moves in an L shape - that’s a rule. a meta-rule would change how a knight moves from an L to a 2sq-then-2sq right-angle for example. Those meta-rules are at the higher scripting level that can be used or done whenever.
Right now, D4 team has said this on seasons: “We’re going to change the most powerful legendaries to force the meta game to shift each season”. That seems too simplistic to me.
Personally, I think musical chairs, just rotating power levels on gear to swing meta is not a great idea. I would much prefer to have a more living breathing itemization that allows for more theorycrafting, as mentioned in an earlier post.
The AI systems in the reddit post and SC2 project wouldn’t preclude that. I think it would just free up the developers from doing hands-on balancing to do other things in general. Completely agreed if they were to use the system in the wrong way.
I think it would be overkill for a tiered trading system where only accurate item ranking is a challenge. But I think you’re right that it’s incredibly promising for many solutions, including balancing and playtesting where machine learning has more application. So kudos for the creativity imo.
Steps in the right direction I’d say. I was originally going to respond to the first part I quoted with “I agree, I just don’t see how it can be done”
Being able to pass along a little bit of gear just to get that first leg up/help out a friend, I completely understand. To me that’s passing an item or two, not a full 6 set with supporting best build items etc…
But just giving someone a decent item that can help them go on and then obtain the true items they want? That’s the ideal in my opinion. I fully support that.
Sure, I am just commenting on the reddit post where they talk about changing it on-the-fly, or each night.
Seems like a really bad idea to me. It sounds like an extension of their D3 strategy; making new flavors of the month each season.
Instead of making a good game where you have a wide variety of viable builds.
Sure, add new items over time, thus also adding new builds or new meta. But dont just buff and nerf existing items each season with the sole purpose of creating, as you accurately put it, “musical chairs”.
I just thought of an even worse one, but it would ‘work’ in a way.
Cap damage. No matter what, there’s a cap to how much damage per blow, and now fast you can strike, period. Make it a low threshold. So no matter what, all builds and gear cap out and no one can be best, no one can be faster, etc etc.
Horrible idea. But it would work. No one would play though.
I’m not even talking about seasonal. Roguelikes tend to have playthroughs that last no more than a few hours, after which you completely restart.
Re-randomizing things every season doesn’t give the more casual players time to really figure anything out on their own.
The builds are reasonably well balanced with each other in PoE. Enough so that gear and skill tend to be the main thing holding you back, not build unless your build is just badly designed.
Diablo 3’s terrible balance stems largely from 150 difficulty levels and insanely high percentile modifiers. Two things that Diablo 4 isn’t going to repeat.
Once they’re doing hands on balancing, you don’t really need the AI there.
All the AI can do is check the math, something any half way decent data analyst or software engineer can already do.
Yeah, I meant that the AI system could replace the need for humans to do balancing completely.
In the case of the SC2 API for example, the AI scripts control the character in the game - doing all the actions a player would do.
In the case of D4, they would expose a similar API that would allow AI scripts to playtest and balance the various classes and gear.
EX: Using D3 names for sake of example, they might change the DH UE set. Then AI scripting creates a Demon Hunter with UE set changes and runs through the game testing the powerlevels and balancing if needed.
Then it does GoD set, Marauder set and so on if there were changes to those sets - the AI scripting just uses the exposed D4 API to change the sets and gear on the DH. Then runs the DH through a greater rift of certain difficulties and so on.
So the actual people on staff wouldn’t have to do the playtesting and balancing. It would be automated. There’s a lot of combinations to test.
Edit: With Blizz staff, they could just set a goal for each class and set a build to only do GR140 and this other build should do GR135 or GR145 or whatever they want. The system would adjust the gearing power and do automated testing to match the GR cap they set in the system. Or something similar.
At this point we’re just ping ponging back and forth.
If you let the AI do the majority of the balance, then you end up in a place where players aren’t rewarded for figuring things out and choices ultimately don’t have any meaning. The only thing that changes from one build to the next is the spell effects and the order of the buttons you hit, and I’ve seen that style of balancing on WoW. It’s not interesting or enjoyable, and making the numbers better balanced wouldn’t help.
On the other hand if you have your developers do a bunch of stuff by hand or over-develop the AI to behave like a developer, then there really isn’t any need to have the AI over just having your developers do the balancing.
and regardless of if there is an AI, the developers should never think they don’t have to playtest their own game. They will always need to do that.
Right now with D3 they use PTR for a couple weeks to have players test changes out. They don’t have a bunch of iterations in PTR - they might do one or two patches during PTR. This is not a refined approach to balancing and leaves many people frustrated with overnerfs and overbuffs. This happens all the time. It’s a problem.
Presuming you agree with the above obvious situation, an automated system, as spec’d in either post we’ve been talking about would result in much better balancing and much less player frustration.
Edit: Also, AI doesn’t imply anything about being rewarded and so on. It’s just a tool - they could use it to cause the imbalance you seem to want where they spec a few rare combinations of gear to overperform when discovered.
