Tuesday, November 13, 2012

Worse cases and Skills

Worst case scenarios are really important in game design whenever you have a struggle between people.  DnD 3.0 was a great example of this.  Fighters had lots of ways to increase their Armour Class like putting on plate, getting a natural armour necklace, wearing a deflection ring, putting on a shield, and getting magical and mithral versions of their armour and shield.  The problem was that if they did all this many monsters that were supposed to be a 'reasonable challenge' simply couldn't hit the fighter.  For the average case the Armour Class system worked fine, but for the worst case it was a disaster.

You can see the same sort of problem in WOW.  Everybody goes around hit capped such that they can never miss on an attack.  This isn't a problem.  The problem came when people got their avoidance so high that monsters actually couldn't land a blow on them.  There were plenty of funny videos around during Burning Crusade of people soloing raid bosses by having 100% chance to dodge.  Thankfully it didn't end up really wrecking any raid content but it required some heavy handed kludging by Blizzard to avoid that.

Because of this it is important to keep bonuses to Armour Class low and keep randomness high.  As long as everybody is rolling a d20 to hit and a reasonable amount of those numbers will connect everything is fine.  It might not feel realistic that a veteran soldier layered in magical protections can still be hurt by some dork but it keeps the degenerate case from happening.  I took this to heart in skyRPG and tightly controlled access to abilities that increased character defenses.  You can get tough, but you can never get yourself to the point where enemies are utterly unable to hurt you.  Problem is, I followed the same logic with Skills and that was a big mistake.

How often should a random dude like me be able to jump further than an olympic calibre jumper?  Never!  1 in a million when the olympian trips and falls halfway through their run maybe?  Unfortunately in DnD the 1d20 mechanic ensures that I beat the olympian 5% of the time.  Raw Strength checks are even sillier.  The strongest man in the world adds 5 to their 1d20 rolls and I add 0.  So if there is a heavy object that I might be able to lift the strongest man in the world still usually fails to lift it?  Preposterous.

The difficulty is that the system for success with skills has way too much randomness in it.  When any fool can succeed at a check with a couple tries it hardly makes any sense that someone who is a master of the trade might not make it.  What skyRPG needs is a skill system that places more importance on the skill of the character and less on the randomness of the die.  I figure the simplest way to fix this issue is to just shrink the die.  If I am attacking someone I roll 1d20 + Dexterity, but if I am balancing on a ledge I roll 1d8 + Dexterity + Acrobatics.  This way you end up with a system where the basic mechanic is still die roll + stat but combat stays safely random and non combat makes some semblance of sense.

Pics from:  http://www.jimnolt.com/firstencounters.htm and http://blogs.indiewire.com/shadowandact/jamie-nieto-olympic-high-jumper-has-acting-to-fall-back-on-or-jump-into


  1. Taking d20 combat as normative -- where you roll d20+mod against AC -- the impact of modifiers on each side is hyperbolic: Effectiveness ~ (10+x/(10-x)) (x is the +attack & +defence of one side over the other), which can be reasonably well modeled as being exponential (Effectiveness ~ 1.25^x, where x is the boost on your attack and AC).

    It doesn't seem nearly this strong, and part of game design may be to make massive power gradients obvious.

    So you keep these modifiers under heavy control, ensuring that to-hit and AC bonuses are in a narrow range, and everyone can reasonably hit each other.

    Power growth is moved away form to-hit and defence modifiers and into damage and other effects that "look" big when they get big. (Someone with 35 AC and +24 attack looks slightly stronger than someone with a 30 AC and +19 attack, but in fact is a factor of 3 more effective! -- line up 100 of the 35 AC dudes against 100 of the 30 AC dudes, 1 hit kill, simultaneous attacks, and have them fight one-by-one, and watch as only ~33 of the 35 AC dudes die. (also works if they fight in parallel, or they randomly attack first, but KISS)).

    With skills, if you keep things only as a pass-fail mechanic of (roll+modifier), you remain trapped by these exponential power curves. So your modifiers remain bounded (to keep things interesting), and chance dominates.

    In a combat situation, while the Goblin may be allowed to hit the Hero, the damage the individual Goblin does will be trivial. The Hero became more competent via doing more damage, and being able to soak more damage, rather than it occurring at the "do anything" phase.

    Skills and Skill checks could work similarly.

    A Str armwrestling contest might involve doing an opposed Str check. On success, you move the other side by 1d4+str bonus units. After 10 units you have pinned them.

    So the str 10 person needs a 15+ to advance on the str 20 person, and only advances 1d4 on a success.

    The str 20 person needs a 5+ to advance on the str 10 person, and advances 1d4+5 on success.

    Now each contest is interesting, but the outcome is pretty much preordained.

    Something like Climbing works well via this system as well -- the Climb check determines if you can advance, then you roll the amount you advance. Each round on the wall you maybe face a Hazard (where the "wall attacks you" and tries to make you fall). Ditto for trapped locks, or non-trapped locks with a time danger.

    Skills get split into (success modifiers) and (result modifiers). Someone trained in athletics might get a modest bonus to str on athletics checks, but a large bonus to str on the result of athletics checks.

  2. Your ideas for making Skill checks interesting are pretty cool. I considered adding additional complexity and rules to noncombat activities like climbing, intimidate, etc. but I ran into the problem of making the learning curve too high.

    People are generally willing (in a heroic fantasy game with a focus on combat!) to spend time learning the ins and outs of fighting. It is complex, but worth it. People are not willing to spend the same amount of time learning each Skill's mechanics. Keep in mind that in one Skill such as Athletics you would probably want a dozen (Climb, Jump, Swim, Throw, Run, Chase, Ski, Surf, etc.) different systems such as you describe and learning all of those, to say nothing of writing them, would be too much for the great majority of gamers.

    I think it would work great in a video game where you could have different minigames for various tasks but in a tabletop game I think it would be just too much. When I read games like GURPS that have systems for every crazy ass thing imaginable I find them fascinating and a source of cool ideas but when I think of trying to get new players into the game it just seems infeasible.