Difficulty curve in RPGs

[Neanderthal]

One word sums it up for me: Attrition. This is something that hardly any RPG does, there is now no degrading equipment, no permanent wounds, no exhaustion, no limited spell casting, no long term strategic goals at all because the player demands to always be at peak effectiveness. Trawling through a mega dungeon, they will emerge in exactly the same state as when they entered, except with more loot and XP. A long voyage over harsh wilderness, they might have a little mud on their boots, which will never need repairing. For me attrition should be the watchword when a party sets out into the wild, they will have limited resources, they must marshall these and use them wisely as well as make preparations beforehand that increase their chances of success.

In most modern RPGs there is none of that, in effect the characters are superheroes, they do not thirst, hunger, sleep, sit down, despair, run away from superior forces (who to them are just sacks of XP waiting to be harvested,) get lost because of quest markers etcetera. In essence I think Darkest Dungeon has the rights of it here, and the strategic aspects and punishment of players is overlooked almost endemically.

I'll use a mediocre but popular game for an example: Dragon Age, Cousland origin, you're a young man who witnesses your family, home, friends and life stolen from you in a bloodbath. You are then forced into joining an obsolete order of suicidal tainted madmen, who are fated to die in battle. Shortly after this your king, who promised you revenge is killed, your countries soldiery massacred, the Darkspawn almost kill you and your new order is blamed for all of this tragedy. Now you have to trawl though horrors in the Deep Roads, experience cutthroat Dwarf politics, weird magics warping the Elven forests, a demonic infestation of the Mage tower that leads through the metaphysical realm of the Fade, a township besieged by the walking dead and ruled over by a possessed child whom you have to deal with, and find a holy relic that seems to grant miracles. After this you have to kill an Archdemon, possibly at the cost of your own life.

The cost to the player, the damage to their psyche from all the shit they've seen, the wounds they have endured and that now plague them, the changes that they have need to make to themselves to survive this repeated trauma: Nothing. They're stronger, faster, more powerful than ever before and still spouting the cutesy, childish one liners that pass for humour in Bioware games. Think about what they should be after having gone through all of that shit, either a gibbering wreck, a double hard bastard who craps lightning and pisses thunder or a total fucking sociopath.

 

[Lhynn]

So , if I understand you correctly , you suggest to make the beggining tougher , but make make character's abilities easier to get and player's strength shouldn't strongly depend on level , so the player eventually will choose tools he feels more comfortable with and maximizes what he feels like .
Alright , this might sound good to me , but the main question here is how to handle difficulty in later game .

Nope, ramp it up without fear. Theoretically the player should have the tools to deal with said challenges as they come, provided he masters the mechanics and has some (if the system isnt deterministic) luck. Consumables can be used to bridge the gap forcing you to use more of them, in a smarter way. Of course there can theoretically be "builds" to topple all your challenges easily, but that goes deep into metagaming territory and i dont really care for that.

 

[worldsmith]

The fear of alienating people because it was "too hard" was, up until recently, something companies avoiding like the fucking plague.

I shipped a game in the early '90s that, while I wasn't afraid of alienating people, I was concerned it might be too hard. And the reason I was concerned was that when the game shipped neither I, nor my publisher, nor anyone else had actually played the game all the way through.

I had played all of the (several dozen) individual pieces, but never the whole thing to see how feasible it was to survive the attrition suffered along the way. Keep in mind this game was shipped on floppy, this was before internet access had become common, and providing patches was pretty much a non-option. After the game shipped, and after I had taken a few days rest away from the game, I started playing it to see if it could be beaten. I had to know.

As I played I managed to figure out some new strategies and even new ways to think about the gameplay, and some of my friends eventually came over and watched as I sunk hours into this thing, and they also came up with strategies for me to use. On my eighth attempt I finally made it all the way to the end. The win screen came up. "Good," I thought, "the win screen actually works."

And that was that. I never tried playing the game to the end again. I don't know if anyone other than me ever beat it. But I did determine one thing for certain - the game wasn't "too hard".

 

[RK47]

People think that? My reward for playing is to continue to have fun.

Yep. Look at how JA2 was.
The outlying towns are so easy to cap. Then you meet the greyshirts...and surprise. Mikey is here!
O shite. Time to bust out the cheap stuff. Mustard Gas, Tear Gas, Stun Grenades, Mortar, Tank busters.
Jagged Alliance 2 was well done even without mods.

The cost to the player, the damage to their psyche from all the shit they've seen, the wounds they have endured and that now plague them, the changes that they have need to make to themselves to survive this repeated trauma:

 

[laclongquan]

Yeah, JA2 is a good example of difficulty done right.

It's easy to manage a squad to hold a town. And early towns are mostly easy to capture. Shits are simple. You attack town A and you hold it.

But when you get to three towns, now that's hard. You have to juggle your personnels, as well as the weapon's durability. You can not just scavenge off the battlefield and fix it as it fail but to have dedicated mechanics to repair it, as well as defend the airport to import shits. The time come to when you helicopter a team in to replace a wound team and ferry them back to hospital. or ferry a large shipments of bought stuffs in and ferry a large shipments of loots out to fix-then-sell. SAM sites become important target to capture and hold.

The difficulty and complexity of the game expand as the game progress~ And you dont have to add artificial features to bloat up the difficulty.

 

[Siobhan]

I just don't buy the usual arguments that RPGs are such a special genre that they can't have the well-designed difficulty curves you see in other genres (RTS, turn-based tactics, action platformer). In particular because there are some RPGs with good difficulty curves.

Wizardry 7 was already mentioned, and Wizardry 8 is also fairly well-balanced. Those games do it via limited level scaling and bringing in new enemies with more attacks, magic spells, and immunities.

The RoA games are also fairly challenging throughout, due to several design decisions. First, the power increase per level is noticeable but not overwhelming --- a level 1 character with really good stats is not that far behind a mediocre level 6 character. The greatest benefit is the increase in hitpoints and mana, due to point 2: scarcity of potions. You don't find that many, the good ones require rare ingredients, and you can only carry so many of them. Combined with the fact that resting doesn't recover a lot of HP or mana and that the one(!) healing spell is both costly and underpowered you get the attrition effect Neanderthal describes (plus weapon breakage, etc). The dungeons are also varied and constantly pose new challenges: your equipment has been seized, you have a time limit, enemies can destroy your weapons, some enemies are immune to normal weapons, and there's good riddles, puzzles, and traps.

The real problem is that RPGs (and most of their audience, I presume) still buy into that old DnD ideal of going from a total whimp to not just a hero but a demi god. In a system where a single level 20 character can still be taken down by 3 goons, on the other hand, difficulty can be scaled fairly well. Just consider JA2: loot and stats increases make your mercs a lot more capable, but if one of them is careless and suddenly finds himself surrounded by three red shirts, that still means a lot of troube (2 shots + stun grenade + 3 shots the next round = bye bye Shadow). Basically, progression should mean that you stand a chance against stronger enemies, not that weaker enemies become trivial cannon fodder. Combine that with good dungeon design that incorporates increasingly difficult cerebral challenges (puzzles, navigation, item usage, avoiding traps, dealing with attrition) and you've got a game that scales well with both character and player skill.

But yeah, even if you narrow the range of player power variation, you can't use these tricks in a game where you don't do much but talk in towns and run through open planes and valleys fighting enemy mobs the designers placed in semi-random intervals to break up the monotony. So the real question is, why is that all most RPGs do? Why is it acceptable for RPGs to have zero level design?

 

[laclongquan]

Hammer & Sickle.

Even though there is skill progression that make players feel their efforts worth expending, the difficulty increase due entirely to player's choice.

Early game, you can only proceed in a single group to make it easier to concentrate your power in one target. The level limit the skills, as well as the equipments.

But in middle to late game, since your characters become skilled, you feel ambitious to attack in multiple prongs, make it so one or two chars each prong. The skill-up and better equipment afford you that, but the divide of men make you not stronger in each task. So you feel your efforts are well spent but the game not become noticably easier.

Instead one squad proceed in a linear plan, now you can have a squad move to the fence, the sniper take down the guard, then the other group move to the other side in a smooth action. Or instead, two fire teams pincer one squad of enemies. In exchange for one super strong team, now they can have two teams of better than average, or three teams of so so quality.

It is the player's choice to increase their range of operations that make the game harder as it goes, in keeping with the levelups.

 

[Telengard]

JA2 also has very little choice character-wise. There is very little going on on a character sheet but a few skills that help you kill things better, and you don't even muss about with those skills much. That simplicity is what is easy to balance. Not to mention, it's a degree of choice that would get people crucified in the RPG community.

A slow progression system with a lot of choice is, for example, Twilight 2000 - a crpg where Potato Speak is a skill. Such a game is impossible to balance directly, even with its low progression, because players are progressing in vastly DIFFERENT directions. Low progression or high progression actually doesn't matter for balance, both can be easily balanced, as long as the system itself is simple. It is, rather, complexity that is difficult to balance.

Unless, of course, you use level scaling, however limited. Once dev-cheating level scaling is involved, you can balance anything.

 

[Daemongar]

The real problem is that RPGs (and most of their audience, I presume) still buy into that old DnD ideal of going from a total whimp to not just a hero but a demi god.

This is incorrect, as the Zero to Hero did not come from D&D. In initial AD&D/D&D and earlier RPGs, going up a level took months and death was around every corner. Having a player survive to high level was very rare. Even earlier games, such as PoR and such, you would cap out at level 7, and leveling took a while.

Now that I think about it the decline may have been started by... the original Fallout. AD&D required a lot of XPs to go up each level. It was Fallout 1 that changed the dynamic to 1000 xps to go to 2nd level, 3,0000 for 3rd, 60000 for 4th 10000 for fifth, 15000 for sixth, etc. This led to a more rapid leveling, which people seemed to like. This was only adopted in 3rd Ed. D&D 3 years later, in 2000. By that time, most games were not only allowing for more rapid leveling, but those that retained 2nd Edition D&D leveling tables were starting to give out gigantic XP rewards for stupid quests.


Honestly, I believe it was CRPGs that started the "level every 10 minutes of game-time" approach to gaming, and AD&D and other tabletop RPG's responded in kind (to their detriment.) The problem is not with DnD but in marketing. RPGs compete with each other and are entertainment. Leveling is fun. You get all these powers you didn't have before. More levels = more fun!

 

[Lhynn]

Aye, getting to lvl 5 or so was an achievement in AD&D, getting there with a mage was a fucking ordeal. People dont know how piss easy they got it.

 

[Keldryn]

Yeah, the last year of high school (1991), I ran my longest running campaign to date. After a year of weekly (sometimes twice weekly), the PCs were just hitting 7th level. That was AD&D 2e. I'm replaying the Baldurs Gate games right now, and levelling feels way slower than most CRPGs.

I agree that the expectation of frequent levelling likely came from CRPGs and JRPGs.

 

[Daemongar]

I played

Yeah, the last year of high school (1991), I ran my longest running campaign to date. After a year of weekly (sometimes twice weekly), the PCs were just hitting 7th level. That was AD&D 2e. I'm replaying the Baldurs Gate games right now, and levelling feels way slower than most CRPGs.

I agree that the expectation of frequent levelling likely came from CRPGs and JRPGs.

Just from memory that I selected Fallout 1, but I'm wondering if there was an earlier game with "lighter" level tables. Off the cuff, I think of the Ultimas, but nobody ever cared what level they were in those games anyhow.

 

[Siobhan]

This is incorrect, as the Zero to Hero did not come from D&D. In initial AD&D/D&D and earlier RPGs, going up a level took months and death was around every corner. Having a player survive to high level was very rare. Even earlier games, such as PoR and such, you would cap out at level 7, and leveling took a while.

True, but my point wasn't that every character could easily reach a high level, or that leveling is a common and frequent event. It's that the power increase from levels is enormous, much larger than in other systems (e.g. DSA) or games (e.g. JA2). A high-level wizard in D&D is almost like a force of nature, a high-level wizard in DSA (3rd ed.) can still be overwhelmed by low-level encounters --- and levelling up also takes dozens of sessions in DSA.

You are right that D&D's power explosion (a better term than fast progression because actual leveling can still be slow, as you point out) is offset by the fact that players usually don't reach these high levels. But the power curve is nonetheless fairly steep, and that is the paradigm many RPGs follow nowadays, except that they drop the lethality that kept the system in check. Whether the power explosion in cRPGs is directly inspired by D&D is another and somewhat orthogonal matter; instant gratification culture and power fantasies definitely play a big part, too, but my hunch is that D&D set an important example in this respect, just like Diablo made enormous loot progression mainstream.

Just from memory that I selected Fallout 1, but I'm wondering if there was an earlier game with "lighter" level tables. Off the cuff, I think of the Ultimas, but nobody ever cared what level they were in those games anyhow.

If close ties to D&D is not a criterion: Might and Magic has very fast level-ups.

Low progression or high progression actually doesn't matter for balance, both can be easily balanced, as long as the system itself is simple. It is, rather, complexity that is difficult to balance.

The implicit assumption here is that difficulty must be directly tied to stats, and only stats, in the sense that your level 2 warrior cannot stand a chance against a rampaging ogre no matter what you do and a level 10 warrior will win no matter what you do. But that isn't really difficulty, that's number crunching.

laclongquan made a good point earlier: the real difficulty should stem from the need to combine the skills and resources of your party/squad into a viable strategy, and how you distribute skill points affects what kind of strategies you can use. The viability of these strategies should be tied to level design (high ground, choke points, shooting from cover, explosive barrels, swamp pits, traps, and so on). So difficulty isn't increased via higher enemy stats but by forcing the player to make use of an increasing arsenal of increasingly more complex strategies while designing the combat field in such a way that these strategies become harder and harder to employ.

In my book, JA2 does this perfectly. The traits and skills of your mercs decide the set of viable strategies (run-n-gun, turtling up behind cover, sniper, explosions, stealth, martial arts for disarming enemies, etc), the level design becomes increasingly more devilish (no good high ground or line of sight for sniper, little cover, mines, hallways fludded with gas) while the enemies stay largely the same but get access to new strategies (like dropping a mortar shell on a group of mercs on a rooftop).

When you approach things this way, difficulty only means how hard it is for the player to come up with a viable strategy while the game constantly presents them with challenges that require modifications to their old strategies. Balancing means that every skill/attribute point you spend must pay off via an increase in the number of viable strategies (no, that's not filthy Sawyerism because not all strategies have to be equally viable in all stituations). That's of course easier said than done, but it's not categorically different from the balancing requirements in other genres.

 

[Frozen]

Id depends on a game system. It could be balanced better but it will always be tradeoff between linear and freedom. And you should reward a plyer later on if he's doing things right with easier path if he prepared himself.
Main problem nowadays is that RPG is targeted to causals so they don't want a challenge later on. Its ok at the beginning to give them fake sense of achievement . Not to mention its too much work, and gameplay in most parts is always lacking in RPG.
The way I see it more important is to try and "mix"" things up, make it so that its not so linear progression system- you go from 1 to 30 and enemies only follow that pattern. Make it more fluid.
It terrifies me to say but one TES feature if altered could help- having skills that lose effectiveness if you don't use them to the extent that after some time they are completely gone as "forgotten".
Or having in game restrictions that balanced it out-ether incorporating brand new enemies (stronger better AI ) later on.
Or having story do the restrictions- game strips you parts of your build power late in game because of plot.
Or having a system more centered around your playing skill than level. Probably the most difficult because RPG still goes from stat based tabletop linear progression where if you don't se a clear advantage in leveling it becomes pointless (Oblivion).
However you do it there will always be easy for some people who min-max it or "abuse" gameplay in some way to make it easier for themselves.

 

[DraQ]

Hello . I would very much like to read people's opinion about this . I've noticed that a lot of RPG games have harder begging part , but become easier as the game progresses .

I think this is quite bad and difficulty shouldn't be like this , I think that such curve indicates a somewhat bad design and developers not willing to spend proper amount of time on game's balance .

Yes.
Basically the problem is:

I think it's partially a side effect of the "zero to hero" progression in many RPGs. When you have a steep power curve, a beginning character can be taken out by one or two lucky hits from even the weakest monsters.

If PCs went from "competent" to "very competent" over the course of the game, the difficulty level would probably be more even. Many players might find it underwhelming to have a narrower range of progression.

Ideally an RPG, even from zero to hero one should start out with some content being hard and the rest being outright impossible, while progression should lead to the formerly hard content becoming easy but impossible (and therefore off limits) one becoming very hard yielding net average increase in gameplay difficulty.
The problem is that it's hard to do such a thing mechanically. First and foremost it's hard to constrain power curve in such way that player can't just outgrind and then outlevel any threat. Ideally all gameplay challenges should only be surmountable with good player skills, with character progression only preventing automatic failure, but in most cases, after early game where player has to deal with immediate challenges without opportunity to progress on the power curve, the player can defeat every single challenge by becoming more powerful and then attacking it in the most straightforward manner. This is of course bad design, but it seems difficult to avert and would require unpopular decision to do so.

That's why I believe that vast majority of cRPGs would fare much better without character progression, BTW, just initial builds, or, at most strictly limited progression.

Having progression that's horizontal (new abilities expanding character's toolset for dealing with in-game situations) rather than vertical (more power - HPs, damage, AC, etc.) is also very helpful as it prevents power inflation.

 

[Telengard]

The implicit assumption here is that difficulty must be directly tied to stats, and only stats, in the sense that your level 2 warrior cannot stand a chance against a rampaging ogre no matter what you do and a level 10 warrior will win no matter what you do. But that isn't really difficulty, that's number crunching.

laclongquan made a good point earlier: the real difficulty should stem from the need to combine the skills and resources of your party/squad into a viable strategy, and how you distribute skill points affects what kind of strategies you can use. The viability of these strategies should be tied to level design (high ground, choke points, shooting from cover, explosive barrels, swamp pits, traps, and so on). So difficulty isn't increased via higher enemy stats but by forcing the player to make use of an increasing arsenal of increasingly more complex strategies while designing the combat field in such a way that these strategies become harder and harder to employ.

In my book, JA2 does this perfectly. The traits and skills of your mercs decide the set of viable strategies (run-n-gun, turtling up behind cover, sniper, explosions, stealth, martial arts for disarming enemies, etc), the level design becomes increasingly more devilish (no good high ground or line of sight for sniper, little cover, mines, hallways fludded with gas) while the enemies stay largely the same but get access to new strategies (like dropping a mortar shell on a group of mercs on a rooftop).

When you approach things this way, difficulty only means how hard it is for the player to come up with a viable strategy while the game constantly presents them with challenges that require modifications to their old strategies. Balancing means that every skill/attribute point you spend must pay off via an increase in the number of viable strategies (no, that's not filthy Sawyerism because not all strategies have to be equally viable in all stituations). That's of course easier said than done, but it's not categorically different from the balancing requirements in other genres.

To see how easy it is to destroy the balance in JA2, just see 1.13, where game balance gets destroyed by adding just a few extra choices. Scopes bork the AI to the point where it can't fight back. Changes to the targeting system remove one of the few advantages the AI had, making them even easier to roll over. New weaponry and ammo makes it even easier to kill them. More ammo and equipment availability destroys the economy and removes any need to shop outside of buying specialty ammo for your 66s while also decreasing the drive to nab mines.

Or take the even simpler progression of an FPS, which is almost as simple as it can get since it's only about weapon loadout, and every weapon is hand-placed by the designer. Even then, even at that simple and low progression, companies can still ruin game balance with choice. Such as by EA adding weapon micro-transactions, causing the designers to not know what the hell weapons players are going to have at any given time. The designers will still be able to ensure that everyone can finish the game, sure, but they can no longer tailor the combat to a specific challenge. And carefully building that challenge is the core of FPS enjoyability.

What's more, on computers it's all number crunching. Even every thing you listed is number crunching, and you can't hand-wave it away. All of those possibilities have to be translated into numbers so that they can be programmed in, and, in order to make challenge, the player and AI have to carefully be assigned a list of those numbers that create the situation that happens on the screen. Ignoring the numbers is to ignore the very essence of what creates the challenge. - What is the percentage advantage to having high ground? What is the percentage advantage to having cover. How much damage does an exploding barrel do and over what radius. What is the chance of an AI character knowing what an exploding barrel is and not going near it? Will the AI character make a risk assessment if an ally is in trouble but it will have to travel near an exploding barrel to help that ally? Is the intelligence stat going to govern that risk assessment, so beetles will run right up to the barrel while humans won't?

Or, look at this a different way. Take Wasteland 2, a game absolutely hammered here on the Codex for its lack of options in combat, and realize that the baseline combat options for Wasteland 2 are the same as that for Jagged Alliance 2. Fire, Aim & Fire, Rapid Fire, hide behind cover and to any of the above, sneak up behind the enemy and use skills on them without engaging combat. Jagged Alliance 2 adds stances and uses the much more AP system, but the core baseline combat options are the same. Any RPG has to cope with the fact that RPG players will demand a whole lot more, even it's just silly power-ups on cooldowns a la Shadowrun Returns. That's like bare minimum for an RPG not to get piled on.

 

[Siobhan]

I think we're talking past each other, so let's isolate what I take you to be saying. I never said that balancing is easy but rather that many genres are hard to balance yet games of comparable complexity in other genres seem to do a better job. So if you say 1.13 is unbalanced, alright, that only shows that the mod fucks up the balance. That doesn't really show much, just like I could point out that Mass Effect has few meaningful gameplay choices yet much worse balance than JA2.

Your central thesis is that the more choices a game offers, the harder it is to balance. But that's not quite what's happening. Rather, it is the amount of variance between player characters that makes balancing hard, and this measure is only indirectly related to the number of choices. That's pretty much what DraQ talks about in his last post:

Having progression that's horizontal (new abilities expanding character's toolset for dealing with in-game situations) rather than vertical (more power - HPs, damage, AC, etc.) is also very helpful as it prevents power inflation.

Not all choices are the same, and a character system with a lot of choices need not imply an unmanageable amount of variance between power levels. I already pointed out one such case: DSA offers more choices than D&D, but the power variance is much smaller. That's also why I harped on the importance of level design and challenging the player rather than the numbers on their character sheet: these things can be affected in a qualitative manner by stats rather than a quantitative one, and this variance is necessarily more restricted than what you can get quantitatively. Think number of approaches to solve a problem rather than how quickly your barb slices up the goblins.

That's also why these things are not number crunching: it is not necessarily the case that the player can crunch numbers to discover the set of viable strategies because they may not even be aware of the full set of candidate strategies. It takes smarts to figure out what moves you can or cannot make, and the difficulty of this problem increases as the number of strategies grows while their efficient deployment becomes harder and harder due to level design.

tl;dr: It's not the number of choices but the power variance those choices give rise to that makes balancing hard. It is possible to design fun systems with many meaningful choices that do not give rise to huge power variance, but you don't see them often in cRPGs for various reasons:

- satisfying the player's power fantasy
- shallow combat mechanics focus on DPS and HP as the main means of progression
- lackluster level and encounter design leave harder enemies as the only way of ramping up difficulty

 

[gestalt11]

This is incorrect, as the Zero to Hero did not come from D&D. In initial AD&D/D&D and earlier RPGs, going up a level took months and death was around every corner. Having a player survive to high level was very rare. Even earlier games, such as PoR and such, you would cap out at level 7, and leveling took a while.

Now that I think about it the decline may have been started by... the original Fallout. AD&D required a lot of XPs to go up each level. It was Fallout 1 that changed the dynamic to 1000 xps to go to 2nd level, 3,0000 for 3rd, 60000 for 4th 10000 for fifth, 15000 for sixth, etc. This led to a more rapid leveling, which people seemed to like. This was only adopted in 3rd Ed. D&D 3 years later, in 2000. By that time, most games were not only allowing for more rapid leveling, but those that retained 2nd Edition D&D leveling tables were starting to give out gigantic XP rewards for stupid quests.


Honestly, I believe it was CRPGs that started the "level every 10 minutes of game-time" approach to gaming, and AD&D and other tabletop RPG's responded in kind (to their detriment.) The problem is not with DnD but in marketing. RPGs compete with each other and are entertainment. Leveling is fun. You get all these powers you didn't have before. More levels = more fun!

Technically, in AD&D, most "normal" people were "level 0" and a level 1 fighter was considerably better than them and higher level fighters could just slaughter level 0 things in droves.

But this is not really analogous to what many cRPGs do where they start you as a murder hobo with no reall special skills or effects and then at high levels you have like 10 special abilities from level and 10 special effects from gears. ARPGs and roguelikes do this a lot.

 

[worldsmith]

For proper balance, every fight would have to be keyed for every single possible choice available

Every fight should not be "balanced". That's like saying every door in every dungeon should be locked, and made with a material that is just the right amount of challenging for the PC party to break down and/or has a lock which is just the right amount of challenging for the PC party to pick. Some doors are just unlocked, or open, or flimsy. And in any game world that does not want to be constantly reminding me that it's really nothing but "Lame-Ass Combat Engine 147.0", not every fight is going to be "just my size". The strength of opponents should be more about game world consistency (who are they? why are they there? what are they doing there? who hired them? etc.) than about performing every conceivable design contortion in some twisted effort at "balanced combat" all while trying to pretend really hard that that's not just some overly messed up, convoluted form of level scaling.

There are some who advocate having dungeons of fixed power levels, and letting the player choose where to go and when, skipping over dungeons that they exceed the power level of and thus finding their own challenges. But 1) that means none of the dungeons involved can be part of the game's plot, since the player will be randomly skipping any number of them, which would make it really hard to understand the plot if key points are randomly being skipped.

Now you are getting to the real problem with most RPGs - the pre-written plot through which all players are supposed to be railroaded. All of that choice and divergence you talk about earlier in your post - that's what I want (though I want it in far more areas than just combat) - and that's exactly what you really don't get very much of at all with a pre-written plot. The pre-written plot is already a losing RPG design strategy, before combat is even considered. (Note that by "losing" I am referring to game quality, not to game sales.)

And 2) and more importantly, the skilled player will be spending an increasing amount of in-game time not playing the game at all, but instead wandering around looking for somewhere that he can start playing the game again. The farther he gets in the game, the larger the percentage of available dungeons he will need to skip, and thus the greater amount of time it will take him to find an appropriate challenge. And very few people have much patience for that sort of thing.

This is just silliness. Surely you can conceive of ways that a player might be able to become informed of where/how they might find proper challenges without having them wander all over the universe and popping their heads into each and every dungeon. How many RPGs do we need where you can go get rumors from a barkeep, or a person on the street, or be assigned quests by a guild officer before it becomes obvious that those very same mechanisms (and about 1,000 others) could be used to instantly provide the player with this information? Even one of the most lame "RPGs" in existence (Endless Battle) allows you to instantly select your level of challenge.

A related rant: Too many people think "exploration" should mean "aimlessly wandering around until you see something cool". And since they don't want to waste time (and lack imagination) they subsequently think stupid stuff like they should be able to wander around aimlessly for (on average) no more than X seconds without finding something cool (for some relatively small value of X). That's a real dumb way for it to work. These are games - there should be some challenge and fun gameplay in exploration - maps, rumors, clues, tracking skills, magic skills, interrogations, informants, fame (with NPCs subsequently pushing suitable requests to you) and a bunch of other elements that can come into play when finding your next challenge. And typically the PC/player shouldn't even really be looking for their next challenge, but for their next objective - something they want to obtain or accomplish - and the challenge that goes along with achieving that objective should (in general, due to consistent game world characteristics) tend to scale with the value of that objective.

setting the difficulty just below the average means the largest swath of people will be at or near the designed fun-point of the game all of the way to the end.

I have doubts about how well that actually works, and that is because of something called sensitivity to initial conditions. E.g., even if the difference in skill between players is relatively small, if one of them tends to need a health potion or two every combat and the other doesn't, that means one of them is burning resources on health potions which then means they can't buy something else that would have helped them do better. And while the other person saved their money and bought that something and can therefore can do even better now, perhaps able to keep up with that difficulty curve so they are still able to complete most combats without need of health potions, the ever so slightly lesser player is now having to do without that extra help and is now burning 2 or 3 health potions per combat. (Of course, with RPGs that just auto-heal you after a battle, where time doesn't matter so outside of combat there's no rush to heal quickly with potions anyways, and with a bunch of other dumbing downs, maybe game designers manage to drive all of the chaos out of the system as well.)

Now, there are off-beat theories out there for getting around this issue.

I'm definitely interested in checking out "off-beat theories" that might provide (or simply hint at) improvements and/or alternatives. If this subject really comes up frequently (as Zetor mentions), it would be nice if there were a meta-thread that tracked them all (and a single link to that meta-thread from Project Monkey).

However, the vast majority of players are not accepting of off-beat solutions.

Maybe - but those are crap players. Who cares about them (other than devs wanting to make money)?

Anyways, my thinking is that is what config options are for. Design the game to be truly good, but also have config options that (even if they completely break the game in ways monocled players would never appreciate) turn it into more of a game for the masses. Have a small number of config pre-sets so a player can change from one game style to another without going through every single config option. One of those config pre-sets is the "true game", but the default pre-set is "dumbed down for the masses" version. This can include things like ridiculous hand-holding (e.g., quest markers not only for current quests, but also where new quests suitable for their current ability can be found, and even some kind of lame "main quest generator" that strings together some objectives for those whose minds are so addled they can't even decide what flavor of pudding they want without being told). By doing that it should be possible to not scare away the casuals.

No, it's the fact that computer tech was never the big wall that C&C runs up against.

That's not quite true - back when computers had 16KB or less and only cassette or cartridge for storage, it was most definitely a wall. At 16KB, even a text-based adventure/RPG is noticeably limited in total size. But for pre-written stories it is true that we have long-since left that wall behind (though art assets can always be made to fill any amount of drive space you wish).

The wall does still very much remain when it comes to simulation though. Even (tentatively) targeting a 64GB 8-core beast of a machine as the "recommended for full experience" system, I am not going to be able to have a world as large as I would like, mostly because land area needs enough NPCs to not feel empty (and more importantly, NPCs need a land to be constrained to prevent them from breeding out of control), and running that many "always on" (not just in combat) AIs quickly runs into some pretty crazy resource requirements if "you" (the game dev) don't get pretty darn creative in how you optimize it.

the end dungeon

There's the problem. Why is there an "end dungeon"? I mean, I know it's because of that pre-written plot thing, but the pre-written plot idea is already bad. Why use bad to justify more bad? How long before people realize that "pre-written plot" and "living world" (not to mention real player choice) are mutually exclusive?

the designers are doing their best to convince the rest of them too, otherwise all that time/money spent on the side content is wasted

Which is why all RPGs not using some sort of procedural generation will always be warped - very limited amounts of content because it costs too much to make more, and a strong push away from real choice because that leaves "too much" content not experienced by the player. Saying such games won't continue to be warped in the future is more hopelessly optimistic than saying pay-per-time MMOs won't be grindy. Unlike pre-scripted RPGs, procedural generation can benefit in a fairly direct way from tech improvements over time (i.e., more sensible and interesting generated material).

laclongquan, besides reminding me I need to try JA2, your post (and many strategy games like Warlords and AoW) also suggest an interesting option RPGs could have - splitting up your party to take on multiple quests at once. Heroes of a Broken Land is the only RPG I know of that lets you split/merge parties and have more than one active party. It's gameplay also points out how much "real time" (not as in real-time play, but as in "game world has a real sense of time") matters, because when you send out your strongest party to go clear some far away dungeon or tower, they may not be back to defend your home town for a long time.

First and foremost it's hard to constrain power curve in such way that player can't just outgrind and then outlevel any threat.

You seem to be assuming the standard "static world with pre-written plot" RPG. In a proper living world, NPCs can grind right up the same power curve the PCs do. (There's still save-scumming to deal with as that can be used to allow the PC access to more dangerous grinding than NPCs can get away with, but there are solutions for that as well.) Of course, in a proper living world there are other design issues which are "interesting".

Even if you assume a static world (fixed NPC levels), as long as you're willing to do without the pre-written plot the problem is still trivially solvable. Just one example:
1. Make the power curve be one of diminishing returns - don't let the player grind their way to infinity in anything, but only ever closer to some limit.
2. Make it so a party of N characters at max melee skill level will get their ass handed to them by a group of 4*N characters at half that skill.
3. Have threats that include (up to) 4*N (and more) such characters.
4. Only let the PC party consist of up to N characters.
That way there are always threats that will be essentially impossible for the PC party (at least without burning through a lot of consumables). Of course, greater threats have to tend to be in the way of more worthwhile objectives in order to entice the player to go after what they think they can handle rather than just killing tons of easy stuff (though some players here would apparently claim that such enticements are unnecessary for them because they only need the challenge).

And that's without getting into more interesting mechanics like deteriorating skills and actual sacrifice (including what Neanderthal was talking about).

What's more, on computers it's all number crunching. Even every thing you listed is number crunching, and you can't hand-wave it away. All of those possibilities have to be translated into numbers so that they can be programmed in, and, in order to make challenge, the player and AI have to carefully be assigned a list of those numbers that create the situation that happens on the screen.

This is too reductionist. To make it even more obviously absurd: In computers, it's all charged particles moving around. Everything in the game has to be translated into the motion of charged particles.

AI code can make use of abstractions. Implementing an AI does not involve programming everything as numbers (nor as C++ or some other traditional procedural programming language -- for example, see Zillions of Games - it's not a great AI engine, but the wiki page does show an example of how a game is coded in it). There are generic algorithms that can be applied in a wide variety of situations. It's not a panacea - compute time (and space) and solution quality are pretty much always at odds, and the only way to improve one without hurting the other is to optimize/specialize the algorithm for the given situation. But "carefully be assigned a list of those numbers that create the situation that happens on the screen" is not even remotely an apt description for how such optimization is performed, so I'm not sure what you're really talking about there.

What is the chance of an AI character knowing what an exploding barrel is and not going near it?

If it has been programmed to know that (and the code isn't buggy), then about 100%. If it hasn't (and hasn't been provided some general learning facility by which it could learn it), then 0%.

Anyways, the points of your post seem to be:
1. crappy AI is bad
2. AI that isn't crappy is hard to create
3. gameplay options that are different enough from each other (e.g., more than just numerical differences) makes creating an AI that isn't crappy even harder

I think we can all agree on #1.

For #2, I think "not crappy" is a low enough bar that it's not that hard. It does kind of require that whoever is making the AI is actually shooting for "not crappy" -- if they shoot for "really good" but don't have the resources to pull it off, you can end up with a super crappy AI. E.g., a (very simplistic) "really good AI" could be made by just trying every possible move, looking N moves in advance, and picking whatever set of moves has the best result (best move-N situation). But if someone writes code like that they will find that N has to be large enough to get good results, small enough to not take too long, and that those two ranges for N don't even come close to intersecting. On the other hand, if you're just shooting for "not crappy" you can use (not necessarily carefully coded) heuristics to come up with a bunch of options (some horribly bad options in the set doesn't necessarily matter much, but it should try to include as many of the better options as it can), and then use monte carlo techniques to smoke-test them in the current situation. Unless you throw lots of CPU time at it (and always make the heuristics generate the best move as one of the options), that's not going to consistently generate optimal choices, but it will avoid making truly dumb moves most of the time. This might be considered cheating - but in the specific case of an RPG those heuristics could be made to include the player's "favorite" (commonly used) moves, greatly increasing its efficacy, or at the very least requiring the player to use a much more diverse and less predictable play style. The approach also makes it easy to allow the player to set a time limit for AI moves, so the player can decide how long to wait vs how smart the AI is. (This is all basically a simple variation on Monte Carlo tree search.)

For #3, it really depends on the algorithms used. Gameplay options that seem very different from a player's perspective can still map to the exact same optimized algorithm - sometimes. Sometimes they can't. Anytime a new option maps well enough to an algorithm that the AI already uses, it's not very hard to add it (though it can impact performance slightly - that again depends on the algorithm). Anytime a new option doesn't map well to any algorithm the AI already uses, then it's going to take a chunk of work to add yet another algorithm (or you reuse an existing algorithm and just let it handle that option poorly - aka inefficiently and/or inaccurately). In some cases, like the AI strategy described above, the same core search algorithm can be used for everything and it's only the heuristic "helper" algorithms (and some state management) that need be customized (and as mentioned above, those helper algorithms need not be perfect - if one of them suggests a bad choice, that will result in some wasted CPU time, but the core search algorithm will typically filter that bad choice out as long as any of the other heuristics has provided a better option).

(Note that the above is not a "sales pitch" for the described style of AI - it's just an example of something that could be used and wouldn't be that bad, either in terms of performance or in terms of implementation difficulty, while having the benefit that it can take advantage of tomorrow's hardware tech by turning more CPU cycles into better AI.)

many genres are hard to balance yet games of comparable complexity in other genres seem to do a better job

This is a bit of a tangent, but I see some parallels with another genre that I think are worth mentioning.
Typical RPGs have a lot in common with many 4x (and RTS) games - having more power lets you get more power - it's unstable. RPGs that let the player essentially become gods (eventually kill anyone/anything they want in the whole game world) effectively lack any "power cap" to keep that instability in check, just as most 4x games do. That is, after a certain point things become irrevocably imbalanced. In a good 4x game, the game somehow recognizes this point and ends the game (or scenario/map) - e.g., the AI admits defeat, or there's some "vote" where the winner is elected "ruler of the galaxy", or something like that. Unfortunately the RPGs that share that same imbalanced situation don't do anything so kind as that. Instead they have their fixed plot. And the player, if they want to "win the game", is stuck following that plot to the drawn-out bitter end, even though in terms of balance-of-power the game was already over long ago.
You could fix an unstable RPG just like how 4x games do it if you were willing to modify the win condition to include "gets a certain amount above the designed power curve", and then skip the rest of the fixed plot - maybe provide a summary saying all of the things the PC (party) did after that point (rather than making the player painstakingly enact it all). But that's never going to happen in a game with lots of expensive hand-made content where devs want to force-feed you a large fraction of that content to make sure you "enjoy the game" (and justify their own efforts/expense).
(The other thing 4x (and RTS) games do is have a difficulty setting. But the goal there isn't to make every battle balanced, but only to control the ultimate balance between "player eventually is destroyed" and "player eventually comes out on top".)

 

[Lhynn]

You seem to be assuming the standard "static world with pre-written plot" RPG. In a proper living world, NPCs can grind right up the same power curve the PCs do. (There's still save-scumming to deal with as that can be used to allow the PC access to more dangerous grinding than NPCs can get away with, but there are solutions for that as well.) Of course, in a proper living world there are other design issues which are "interesting".

Even if you assume a static world (fixed NPC levels), as long as you're willing to do without the pre-written plot the problem is still trivially solvable. Just one example:
1. Make the power curve be one of diminishing returns - don't let the player grind their way to infinity in anything, but only ever closer to some limit.
2. Make it so a party of N characters at max melee skill level will get their ass handed to them by a group of 4*N characters at half that skill.
3. Have threats that include (up to) 4*N (and more) such characters.
4. Only let the PC party consist of up to N characters.
That way there are always threats that will be essentially impossible for the PC party (at least without burning through a lot of consumables). Of course, greater threats have to tend to be in the way of more worthwhile objectives in order to entice the player to go after what they think they can handle rather than just killing tons of easy stuff (though some players here would apparently claim that such enticements are unnecessary for them because they only need the challenge).

Decent post overall, but this i cant agree with. Especially putting a cap on character growth. But also coming up with statistically impossible to beat scenarios. Also i dont believe npcs should "grind it up", sounds awfully close to level scaling. It may work for just a few key NPCs, like a rival of the player thats on the same line of work. But remember, the party is the one tackling the biggest baddest obstacles in that place, they are the ones getting the shinnies and being in any actual risk of losing.

Bte, You do know you are basically describing Mount and blade across your entire post, dont you?

 

[Sjukob]

Also i dont believe npcs should "grind it up", sounds awfully close to level scaling.

In Space Rangers NPCs get better gear overtime , but only the gear that appears in shops . So it works like that , as time passes , better weapons , ships , engines etc becomes available in shops and NPCs get gear from there . They also follow different schemes: military ships will have more and better weapons and tougher ships , pirates will look for lighter and less durable ships but will have very powerful engines , peaceful ships have less weaponry and not so powerful engines , but can carry more stuff . Of course there are better things that cost more and NPCs have less chances of getting those . The player main source of equipment are those excapt shops , so NPCs and player are close to each other in terms power in a fair way . There are different ways to empower yourself suchs as drugs , upgrades , dominator's gear , but NPCs can use drugs and upgrade their gear too ( also depends on the schemes they are following ) , I don't want to go in details about dominators , but getting their gear is not easy at all . So what's bad about it ? NPCs just do what you do .

 

[Beastro]

Players want lots and lots of choices, but lots and lots of choices spreads the divergence far and wide. This has a number of effects, but the greatest of them all is the impossibility of directly balancing the game for all players. For proper balance, every fight would have to be keyed for every single possible choice available, so that at any point along the divergence, the fight would be precisely balanced. And that is an exponentially increasing number of balance points.

It's the same problem with freedom in general: Everybody loves freedom, just not the freedom to fail.

It's the rotten heart of Sawyer's mentality as seen in PoE. He claims it's not a "Everybody wins" one, but it's a delusion on his part. "No bad builds" is just his code for less choice and less choice to fail.

People think that? My reward for playing is to continue to have fun, that is, be challenged. Of course you don't want to feel like you're never getting more powerful, but that's why it's important for enemies to throw new curveballs at you, instead of just becoming HP obese.

The issue the power fantasy side of things. If you only seem to get weaker relative to encounters then there's no feeling of growth as a character. There are exceptions to that to play on other aspects of human nature like incentivizing loot and making people accept the difficulty curve to get it.

I know from experience in Everquest (yes, an MMO) the difficulty curve for most classes prevented all but a handful of them to solo or get the best exp rates. That combined with the raid scene and it being the only source of top end gear made people accept the challenge and work together for loot that might take them months to get.

Using an example again from EQ and MMOs in general, part of the fun after being a high level is going back to lower lv areas and cleaning house giving you solid proof of your progressive power gain even if the mobs your level you face would kill you in a second without a group or raid backing you up.

 

[laclongquan]

laclongquan, besides reminding me I need to try JA2, your post (and many strategy games like Warlords and AoW) also suggest an interesting option RPGs could have - splitting up your party to take on multiple quests at once. Heroes of a Broken Land is the only RPG I know of that lets you split/merge parties and have more than one active party. It's gameplay also points out how much "real time" (not as in real-time play, but as in "game world has a real sense of time") matters, because when you send out your strongest party to go clear some far away dungeon or tower, they may not be back to defend your home town for a long time.

Dont know why they didnt make it either.

Make a twin main quests with two heroes(ines). more may distract player~

Each quest received note who got it. Key quest location and items will require that noted hero(ine) present to be triggered as flag.

It would create several stories along side each other. From the standpoint of storyfags

 

[Beastro]

One word sums it up for me: Attrition. This is something that hardly any RPG does, there is now no degrading equipment, no permanent wounds, no exhaustion, no limited spell casting, no long term strategic goals at all because the player demands to always be at peak effectiveness. Trawling through a mega dungeon, they will emerge in exactly the same state as when they entered, except with more loot and XP. A long voyage over harsh wilderness, they might have a little mud on their boots, which will never need repairing. For me attrition should be the watchword when a party sets out into the wild, they will have limited resources, they must marshall these and use them wisely as well as make preparations beforehand that increase their chances of success.

In most modern RPGs there is none of that, in effect the characters are superheroes, they do not thirst, hunger, sleep, sit down, despair, run away from superior forces (who to them are just sacks of XP waiting to be harvested,) get lost because of quest markers etcetera. In essence I think Darkest Dungeon has the rights of it here, and the strategic aspects and punishment of players is overlooked almost endemically.

It's in our nature to do everything possible to gain an advantage. You put in permanent wounds and it'll just wind up with 95% of people save scumming to make sure they never get any.

 

[Aoyagi]

I must be the only one here who considers combat one of the least important aspects of an RPG.

But anyway, I think this is a result of developers catering to the players who rush through the game and are less leveled than those who do every single side-quest. It might also be catering to the power-fantasy feel.