• Add Review
  • Subscribe
  • Nominate
  • Submit Media
  • RSS

Design principles vol. 1: RPGs and strategy

  • Hasvers
  • 07/15/2014 09:02 PM
  • 15553 views
I tend to like thinking about game design. Articles such as Deltree's (Edit: and Craig Stern's) are the sort of things I read ravenously. But despite there being a lot of good writing on the topic*, not much applied to my case, so I figured I might as well start a series of discussions here.

* (among an infinite sea of meh writing; most game makers are not great at reflecting upon what they do)

Vol 1: Why is there no strategy in RPG battles?

This may seem tangential, but really, it's what prompted me into switching to a whole different design. I mean, I love the idea of RPGs, else I wouldn't be here. But RPG battles are satisfying through their simulation aspects, i.e. because of what they let you reenact; they are not satisfying at all in terms of strategic thinking.

The crucial symptom of everything that is wrong with RPG battles fits in three letters:

DPS

Damage Per Second in real-time games like MMORPGs, or Damage Per Turn, is the absolute scale on which you can measure the value of any strategy. Sure, your characters need to stay alive, but only so that they can deal more damage, so the time spent healing them should just be subtracted from the DPS. In the end, no matter how it's distributed over characters and skills and stats and turns and stuff, you're just trying to make a single number go down as fast as possible.

Why is this so bad? Because it means that

1) RPG battles are played on a single line. Multiple stats (like HP and MP) are in fact convertible into each other through skills, so they are just different ways of moving along the same one line.

2) Therefore, RPG battles have no memory: I hit you, you lose HP, you heal yourself, we're back at step one. (Maybe you lost an item in the process, but item consumption is part of the larger resource management gameplay; the battle itself could be replaced by a simple "you lost a Potion" message to the exact same effect)

3) Since there is no memory, for any instantaneous state of the couple {your hp, my hp}, there is always a single optimal move no matter how you got there, and it will work every time I encounter you. There may be some thinking involved in calculating that optimal move (with an arbitrary number of skills and elements and modifiers to complicate it), but once it's found it will always work. By that reasoning, there should never be more than one encounter with any given enemy.


To make this clearer, let's compare it with the case of chess, a thoroughly bidimensional game.

1) Every action is irreversible ; you never come back to the same configuration twice in the same battle (unless both you and your opponent are just derpily moving bishops back and forth).

2) This stems from the fact that, thanks to the second dimension, all your previous actions are "stored" on independent lines. They do not erase each other, they do not act only in aggregate (i.e. simply by summing their instantaneous effects). Thus, they can come back to haunt you later: the position in which you put that pawn in the opening can change everything 30 turns later, not even through an action of the pawn itself, but because it blocks the line of sight of a bishop at the other end of the board.


In slightly abstract terms, RPG battles are chess played on a single row. Characters are abstract bundles of actions, each action is a chess piece that has slightly different rules for how it moves on that line. The order in which you advance the pieces can be important depending on which pieces the enemy has, but you can never circumvent an enemy piece by moving to the side and thus putting the king in check unexpectedly; all you have to do is mow through these pieces until you get there. Some tactics, perhaps, but no strategy.



Now of course, you don't necessarily want a game to be extremely strategic. If every battle in FF7 required beating Kasparov, not many people would have got out of Midgard. But strategy does provide those aha moments that go far beyond the simple tactical calculus of maximal DPS, and they are definitely something I would like to capture in my games.


Next time: I will talk about the basic design principles that I consider especially interesting in the case of a discourse-themed game.

Posts

Pages: first prev 123 next last
I've briefly mentioned Radiant Historia, but it demonstrates juggling the bars around quite literally. Thought you might be interested.
The first one showing more grid management, the second one turn-juggling. You can either change the turn order to chain attacks (which makes you more vulnerable to attacks until you act) and with the help of the usual special bar you can either perform strong attacks/spells (character-specific) or delete an enemy turn.

http://www.youtube.com/watch?v=b6Tl8tWjUjE
http://www.youtube.com/watch?v=Ja8Vdjo0iKI

Just some food for thought.
^^I tried to check out the vids, but I can't watch them in a high enough quality to really understand what's going on. The sheer number of skills is off-putting to me, but, that may be because I'm just getting a peek in the middle of the game.

When you say 'add something truly new that could be done otherwise in terms...' are you saying something that does not contribute to the bar juggling or are you referring to something else?

In the third paragraph, am I right in guessing that you meant 'do away entirely with some aspects' instead of 'do entirely with?'

Assuming I have assumed correctly...

What would be an example of something that doesn't boil down to the bars? And, as the bars are ultimately the goal of combat, how would that fit? Adding such a thing would make for a more varied meta-analysis, but I don't see how it is necessary to create a great game.

I'm all for doing away with stuff (Like dungeons...I hate dungeons), but there must be a very compelling reason to dump turns or hp. Such things make RPGs accessible, much like RPG Maker makes making/playing amateur games more accessible.
Deltree
doesn't live here anymore
4556
Haha, I happened to see my name while skimming the front page and I jumped a little. Hi!

I like the mental image that accompanies the 1D chess analogy, though I do think it helps to recognize a conflict's "basic strategy" versus its "basic goal." In chess, the goal is to put the king into checkmate. The basic strategy varies: it can be a matter of capturing more pieces than the opponent, or sacrificing pieces to bait him into a foolish move, or even pulling out the old Scholar's Mate against an unaware opponent and ending the game in a few moves. RPG battles also have a basic goal: defeat all the enemies before they can do the same to me. It also has a variety of basic strategies (melee versus magic, elemental factors, special skills). Much like chess, the strategies' effectiveness varies with the opponent, though it's rarely an even match-up as you'll either be facing fodder enemies or a boss with an impossible-to-match amount of HP.

Chess, meanwhile, is entirely a spatial experience - the pieces are differentiated by how they behave in 2-dimensional space. Taking that away is roughly equivalent to taking away all commands except "Attack", meaning you end up with a pure but flat experience (like Half-Minute Hero, I guess!)

Really, you need a defined goal for battles for the player's benefit, and a focus on DPS is the natural result of that particular goal. To point a finger, it's technically the result of poor AI on the enemy's behalf, either accidental or deliberate. With chess, your opponent is on equal mental footing; outside of a tactical RPG, a battle seems to exist to pick away at your resources, so dispatching foes as efficiently as possible is the real goal. It's a design philosophy - not a particularly good one, in my opinion, but it's meant to test a different set of skills from the obvious goal. It's all down to implementation, too - chess is timeless because of its solid simplicity, while an RPG's balancing act of expending/straining resources is based in too many variables to easily contain.

(This turned into a ramble before I knew it, so I apologize for that.)

(Also, EO is pretty rad, though I keep ending up with the bad ending! Some day I will make good on my opening promise!)
Kylaila> Interesting, it's still a bit cluttered but there is indeed a very explicit element of manipulation that looks appealing. I've been recommended that game a couple of times now, I should try it someday.


Sviel> You assumed correctly, I edited that now thanks ;) I'll try to rephrase my thoughts more clearly. Imagine a battle system where all characters attack automatically every turn with the same power, and you have just one skill allowing you to delay or advance an attack. Then you would be able to recreate every other RPG battle ever just by changing the patterns of when and how you can use that skill - because that's all the skills of all RPGs ever do: move amounts of potential damage between now and some other time (you could kill a boss by spamming Attack if you had infinite HP, you just need to concentrate that potential damage into a shorter time, and dilute the enemies' into a longer time, to kill before you are killed).

Using this reasoning, whenever you create a new skill, you can evaluate whether it brings any new strategic possibility (instead of just being "Fire 2") by seeing whether it corresponds to a different pattern of "damage delaying" that cannot be recreated by any combination of previously existing skills.


Deltree> Hi! That's fine, we're all old people rambling here ;) Also glad you liked the game, no worries if you don't finish it, there'll be a better one to play someday.

author=Deltree
RPG battles also have a basic goal: defeat all the enemies before they can do the same to me. It also has a variety of basic strategies (melee versus magic, elemental factors, special skills). Much like chess, the strategies' effectiveness varies with the opponent
In a sense, chess on a line has the same goal since you cannot get to the king without killing all the pawns, but I digress ;) Part of the problem is that the strategies you describe in chess are complex actions elaborated from the basic game rules (moving around+capturing), whereas the strategies that you describe for RPGs are implemented as the basic rules.

It's like having a menu to select between 6 different openings, and you need to know which one is good against Kasparov and which one against Vera Menchik. And it will always be good against Vera Menchik because she is of type Ghost/Insect and she has low HP, not just now because she has played this precise sequence of moves. And which moves you can or cannot do at any given time will depend only barely on what you and the opponent have done so far anyway, as long as you have enough MP.

Of course I am exaggerating a bit, but I guess you can see what I mean - this is the reason why it seems to me that RPG battles are closer to chess puzzles than to chess itself: their interest comes from being unique handcrafted tactical situations to solve, not from the possibilities of the game rules themselves.

But let's roll with it and say that every RPG battle is a tactical situation. Then to have something more like strategy, there should be rules determining which enemies you will encounter next depending on those you've beaten so far, so as to have a chain of tactical situations with different effective techniques every time that you can somehow plan for and influence ahead of time.
I'd play that game, actually.
But, with how far you've simplified it, what could possibly not be recreated by some other combination? Unless we account for enemy damage and factors affecting the player characters beyond just damage (like, say, threat rate), then we can't really escape from the 'everything is really a bar juggling game' outlook.

That said, can't there be legitimate strategy in a game about juggling enemy/allied bars? At least, provided we observe some rules like not having Fire and Fire 2.
There can certainly be legitimate strategy, but I think that to extract it we'll need to go back to first principles. The ultra-simple system I described above has all the strategic possibilities of any RPG battle system; it is in a sense the Turing machine of RPG battles (and I'd never have come up with it if not for your comments so thanks a lot ;) )

Everything except damage is just a way of making "damage delay patterns" available or not: out of mana, you can't use skills, dead or sleeping character, you can't use their skills, fully charged Limit gauge, you can access a skill, and so on. Aggro is a way of precipitating death on a specific character, i.e. removing a certain set of patterns in priority; but this is really the sort of tendencies that should emerge from the strategy instead of being imposed by some mechanical effect.

So here's my proposition: before thinking in terms of characters and enemies and status conditions and gauges and skills, plan everything as:
- a list of unique damage delaying/focusing patterns
- a set of basic rules enabling/disabling patterns as a function of which other patterns have been used so far (these rules should be predictable and generic, e.g. not "Pattern A activates Pattern B" but "any pattern with property x will temporarily activate all patterns with property z and forever disable patterns with y" with a short list of properties and intuitive relations explaining these rules)

How this system plays in abstract will convey the entire strategic possibilities of your game once you clothe everything in fluff (e.g. grouping patterns that can exist only together so as to make "characters" or "pieces of equipment").
If any one is interested and I have some spare time, I will try to come up with a demonstration.
I definitely agree that in order to create anything capable of any representation of strategy, we need to be aware of the first principles, so to speak.

But, taking everything to the patterns stage seems overzealous. On one hand, understanding the underlying principles can be very helpful (I've since forgiven my teachers for forcing me to learn processor commands), but (much like processor commands) making anything like a full scale game using those would require a gargantuan amount of time and effort. Once understood, it would be better to speak in more common-language terms.

So, a demonstration would be a very useful learning tool, but...I don't think it's practical for the games that people are trying to make today.

Fully realizing that my position has somewhat shifted over the course of this post, what say you about a contest in which people create a very short game (like 3-5 battles) in which they adhere strictly to these principles? Also, there's probably an article in here somewhere.
I tend to think you would save a lot on the time normally spent balancing your battles by playtesting again and again. It's a sort of investment: a little bit of algebra before adding that skill to your skill tree (or that skill tree to your RPG), instead of a lot of analysis later on.

But it would certainly require a more thorough understanding of the fundamentals than probably anyone has right now. I really like your idea of a contest actually, as a way to explore the full diversity of what people can come with. Let me come up with basic rules for that contest, and I will contact the higher authorities to see if that can be arranged at an official level ;)

Note that if it works significantly, I'm offering anyone who contributed to this topic to be my coauthor for our article in Nature. Or at least our panel at E3.
My, that sounds mighty fascinating. And you are very right.
I don't have any cents to add, but I support the idea of the system and contest.
What is Nature? Forgive me if I sound ignorant, but, I've been mostly offline for months so the only Nature I know consists of a very clean lake.

Even if built from the basic building blocks, iterative playtesting would still be time-consuming. I don't think that it'd be perfect(ish) right off.

Though, it sounds like you were not suggesting anything as arcane as what I was thinking. A 'little bit of algebra' is definitely a good investment; I was thinking more along the lines of a 'whole lot of discrete mathmatics.'
This has indeed been an interesting topic. Made even more interesting by the discussion in the comments :)
Good read!


Ps. Hasvers? In case you guys are still discussing this, I suggest you use the comments instead of chess as an analogy to strategy. This has been a freindly battle of wits, with each side aiming to nudge the other towards a single point. The point that they're proving.
Sviel> I realize now that it was a slightly obscure reference for anyone outside academia. Nature is perhaps the most influential scientific journal, it was a tad hyperbolic of me to hope to be published there ;)

Playtesting can never be removed entirely, but having some systematic bug testing would help dramatically, methinks. Well, I'll keep thinking about it for a couple of days, to design the calculus of damage rules in an accessible way.

Karins> Ha that's a great idea, although I think I will use it in my next article, about dialogue design; these comments neatly demonstrate something I'm planning to emphasize: contrary to chess, the endpoint of a conversation is not something any of us had planned in the beginning. It's a game of synthesis and learning as much as of persuasion - half chess and half collective jigsaw ;)
Ah, well if it helps, I tend not to get things anyway.

Going to be busy for a few days with work, probably, but I'll check back w/e you update. I hope.
I'd argue the biggest strategic gap between a (well-designed) JRPG and Chess would be the number of players. Take Pokemon. In the story mode, it is for the most part about choosing the best Pokemon and choosing the most effective damaging attack. In multiplayer, however, its a game of predicting your opponents move, choosing whether to attack to deal damage, use a status move to hamper your opponent, or to switch- and working out what your opponent is likely to, and playing around that.

Arguably this is a problem that could be overcome by an AI, but it's such a fine balance between an enemy who either acts predictably or totally at random, and one that plays TOO optimally, and makes the game frustrating and thwarts attempts at basic strategy by seeing through them.

One solution is to have battles be asymmetrical; the enemies have all the advantages and special powers, and are stronger than the player, while the player has the advantage of strategic thought, finding counters to each of the powerful forces on the AI's side. Some of the most satisfying enemies are ones who destroy you easily or prove a drain on your resources until you figure out a succinct way to defeat them. Personally, it's a real kick in the guts when the enemy then counters my solution, and can make a game quickly feel unsatisfying. But then again, there is definitely a market for deviously hard games.

Okay imma stop now. I just found the chess analogy to be short-sighted. I agree with your thoughts on random chance, though(mostly; managing random chance can be an extra layer of strategic depth).
author=Pokemaniac
One solution is to have battles be asymmetrical; the enemies have all the advantages and special powers, and are stronger than the player, while the player has the advantage of strategic thought, finding counters to each of the powerful forces on the AI's side. Some of the most satisfying enemies are ones who destroy you easily or prove a drain on your resources until you figure out a succinct way to defeat them. Personally, it's a real kick in the guts when the enemy then counters my solution, and can make a game quickly feel unsatisfying. But then again, there is definitely a market for deviously hard games.


I think you'd like Bonfire. I can't believe it took me this long to bring it up, actually, it's a good contributor to any discussion of strategic RPGs.
author=argh
author=Pokemaniac
One solution is to have battles be asymmetrical; the enemies have all the advantages and special powers, and are stronger than the player, while the player has the advantage of strategic thought, finding counters to each of the powerful forces on the AI's side. Some of the most satisfying enemies are ones who destroy you easily or prove a drain on your resources until you figure out a succinct way to defeat them. Personally, it's a real kick in the guts when the enemy then counters my solution, and can make a game quickly feel unsatisfying. But then again, there is definitely a market for deviously hard games.
I think you'd like Bonfire. I can't believe it took me this long to bring it up, actually, it's a good contributor to any discussion of strategic RPGs.
You weren't wrong! I love how they've refined every character down to their 3 most salient qualities. It's something I actually considered recently, and I'm glad it seems to work :P
Sorry,I can't believe it took me so long to reply! Hectic week.
I played Bonfire a long time ago so I don't remember the specifics but I had enjoyed it a lot, although its unforgiving rogue-like formula of permadeath made it painful for me to restart after a couple of failures.

Pokemaniac (that's an appropriate name for Pokemon comparisons), I agree that it is traditional for RPGs to have battle tactics centered on identifying the nature of your enemy and finding the appropriate response. But I don't like that principle so much: I would prefer to triumph because of what I and the enemy have done during the battle, not because he is of plant type and I used a fire spell. And really, decent AIs are done in a lot of other genres (take turn-based strategy like Heroes of Might and Magic or RTSes, they all have AI that does the job without being either idiotic or untouchable).

How about instead having enemies who have the same brute force as you, but are programmed to have an increasingly broad range of tactical possibilities? That way, every encounter would actually serve the overall gameplay by teaching you how to deal with them in more and more elaborate settings, instead of being its own beast that loses all purpose once mastered.

This sort of learning curve is what basically every genre does, except RPGs who are alone in being solvable by just spending more time on it.

Of course I don't want the endgame to be chess matches against Deep Blue, I don't play videogames to have headaches. But a simple difficulty setting in the menu can take care of that, really. And even better: if you really want to avoid blocking players from the story by extremely hard battles, what about a game where losing a single battle isn't such a big deal? Where for instance you need to lose X battles before dying, or being allowed to continue but missing the best ending?
In other news, the contest is live. Let me know if you guys are interested in judging! (or of course please participate to showcase all your ideas ;) )
I definitely want to help judge or participate, though, I think I'll wait a while to decide which. I won't be able to to do much RPG Maker wise until mid-august, if then, but I know I'll have full internet access back by the time judging rolls around.
The main goal behind a fight is to win it. You usually achieve that by reducing your opponent's HP to zero. Maximizing DPS is the most straight-forward, no nonsense way to do so and is therefore considered the best strategy to focus on. As I tried to say some time ago on the forums, to optimize numbers/values is the main task most RPG players will want to do at any given time (through DPS or stat customization).

If you feature battles in your game but want to be original about the way they are resolved, feature other desirable goals than the "bring enemy's HP to zero" one.

If you want the player not to focus on maximizing DPS all the time, offer rewards to actions that don't increase it. No reward, no incentive. Rewards don't always have to be means to increase DPS.

If you want to add depth to the situations and strategies, don't make a simple turn-based RPG. Add other dimensions inspired by time and space; ATB gauge, Skill cooldowns, Moving on the battlefield, Stances, etc.

Be creative.
Pages: first prev 123 next last