Via the wise ramblings of the
Rampant Coyote, I came across this interesting discourse on
attributes in (C)RPGs.
This got me thinking about things. Sloooowly of course, because I'm dumb, but thinking regardless. One thing I'm keen to experiment with is making all attributes
useful in some fashion, and all combinations of skills and attributes, well, as far from stupid as possible.
So here's a long, rambling pile of waffle loosely related to attributes. Especially how they should work in a single player, single character, combat-heavy, turn based, randomly generated RPG. What can I say, I have specific needs.
Attributes: What the hell are they anyway?
Attributes in most CRPGs suffer from being a terribly mixed abstraction.
On the one hand, they represent basic inputs to the game mechanics. For the average test, most games seem to favor some variant of
skill+attribute(s)+dice roll. Sometimes the size of the dice depend on the attribute rather than it being explicitly present, sometimes the skill isn't really a skill, and in d20 you mangle the attribute pointlessly to generate some half-arsed bonus from it rather than simply using the damn thing as-is. But in any case, it has an often visible and direct influence on the result of tests.
But on the other hand, they often have an altogether more narrative and character-driven role. They're named for qualities we value, and can readily identify in the heroes we hope the character will join. Strength, dexterity, agility, intelligence, perception, wit, endurance... even pure jammy luck.
This tends a cause a big mangled mess, because these attributes are so obviously tied to feats in the real world that they end up assigned to associated tests in-game. Of course, these are rarely distributed evenly. The vast majority of important actions a character might take will end up clustered like freezing penguins around a scant few attributes, leaving the remainder in the cruel katabatic chill of 'dump stat' status.
Once the narrative and mechanical concerns get tangled, it's fairly tricky to sort out. Picking attributes based on game mechanics is transparent and easy (or at least
easier) to balance, but can feel clinical, dull, devoid of depth and flavour. Picking attributes that resonate with our heroic fantasies frequently creates a mish-mash of derived, secondary and tertiary attributes, and great swathes of illogical or poorly distributed actions. It becomes hard to reason about which choices are quantitatively better than others, both for the designer and the player. Strange caps and level dependencies are introduced in an attempt to duct-tape the whole thing into working order. WoW presents an excellent example of this kind of system gone completely bonkers.
So, are there any games or systems that got it right in this regard? Only a few come to mind.
- SPECIAL from the Fallout games seemed remarkably close. It was quite hard to find dump stats for any character, to me a good indication that the attributes have been well-used. Not bad, really. You were granted some leeway by the secondary characters you acquired, though.
- Guild Wars used purely mechanical attributes, and went further than most by making them profession specific. Add to this the ability to freely change attributes between combat areas, and the whole package worked rather well. Unfortunately all the campaigns hit the level cap fairly quickly, so beyond your initial choice of primary profession and the short ramp up to max level, there's really little in the way of permanent character development. To be fair, this wasn't and isn't the focus of the game, but it makes it hard to evaluate the effect of such an attribute system in the context of a more traditional RPG.
- Games which deviate strongly from human baseline can often use a more abstract, artificial system. Even though I hope they all explode from shame, a few White Wolf products make a fair job at this. Sometimes. The dot thingy still blows chunks, but anyway. The assumption here is that the character's can do pretty much anything a human could even remotely expect to accomplish, so the only attributes of interest are those involved with strongly superhuman qualities. This puts them safely outside the terrible reach of 'common sense', and therefore they're subject to being fiddled until they work within the game, without even giving the illusion of modeling the real world.
- If my brain worked, this would be a longer list.
The archetype that doesn't fly
There's an issue that rolls around in RPGs, and that's how infrequently the agile, fast guy is a wise choice compared to the big slow bruiser. This kind of contrast is a staple in fiction, but it rarely works out sensibly in games, at least in part because of the way attributes commonly work.
I did originally have an enormous spiel about my beliefs regarding why this so often doesn't work, but the upshot was mostly to do with timing so I'll save it for a later post.
With respect to attributes, this is probably just a very pointed example of how a narrative influence doesn't even necessarily support key clichés in the kind of stories they often want to tell.
Snut makes a considerable effort to get back on-topic
Pondering on the issues of chained actions. We think of some things in RPGs as being discrete events, such as a single attack. But what really happens often boils down to at least two tests. One to see if we hit the target, and another to see if we affect it. Sometimes this latter is more about magnitude than a simple all-or-nothing, but over time the result is much the same. Sometimes the first test influences the second, such as a critical hit in a d20-based system, but they're still rather distinct.
This presents an interesting knot, because often attributes will influence both tests. A large bonus to either end of the sequence is of less use than medium-sized bonuses (boni? Nah) at both testing points. Combat-based tests may have additional effects too. Does a poisoned blade do anything if you don't overcome the target's soak? What about a magical flame effect? If you miss the initial to-hit type roll, is there a chance of hitting another nearby target?
Going back to an earlier point, attributes which influence both ends of the equation frequently become overpowered. Strength in d20 is an incorrigible repeat offender in this regard.
As I've no desire to puzzle through the convolution of a two-attribute curve by some peculiar random number distribution, it seems to make sense to combine these two steps somehow. A wonderful game (The Secret of Zir'An) presents one option for doing that, although by itself it translates poorly to a roguelike implementation. In a nutshell, an attack is an opposed test as we're all used to, but every point the attacker beats the defender by is available to spend on various combat effects. At the most simple these allow for increased damage or armour penetration, but they scale to include all manner of fun options, disarming opponents or blinding them, trips, throws and generally being evil.
Zir'An still has a (flat) base damage for weapons, mind. Taken to the logical extreme, where damage is entirely dictated by the level of success, this kind of simplification might exacerbate some problems with combat munchkinism. In most current games even if your foe's basically guaranteed to hit you every time, you can still hope for minimal damage, and in fact the 'death by a thousand cuts' routine rather mandates accuracy and damage be distinct.
Implementation? Maybe?
A few things that have been decided about
this game (still unnamed, I need to fix that) which could help in designing an interesting and workable attribute and skill system:
- I'm drawing a clear and inviolate line between mental and physical attributes. This is a fairly obvious result of the nature of the player character. I'm also thinking that attribute increases through level advancement (or whatever) will alternate between these two subsets.
- It's a roguelike. Or roguelikelike. It's all about survival, combat, traps, treasure, random stuff and inevitable, bloody death. In addition to being an inhuman soul-sucking immaterial monster, there's no real reason to be good at making chit-chat with the meatlings anyway, as they're only there to be killed or point you to the next dungeon. And maybe sell you some overpriced tat. Hell, there may not even be non-hostiles for several game iterations! So social attributes and skills can bugger off, and I don't need to worry about trying to make them less dump-statish.
- There will be no humans. Even the creatures not inhabited by servitors of the uncaring gods won't need to be realistic in the slightest in their stat block, because the comparison with humans will be speculative at best.
Of course, there's also cloud lurking in this argent glow:
- Monsters, especially intelligent NPCs, must be easily auto-levelled by the game - including attribute increases. Random generation should also be possible. As far as I can tell at this stage, this will boil down to making attributes as evenly spread as possible. I place no great faith in the ability of my monster generation code, or the fairness of the RNG, so it's quite likely that really broken statblocks will come wobbling into view. They don't have to be equally dangerous, of course, but the more evenly spread everything is the less work the AI will have to do to at least make them interesting, if not effective.
