5 replies to this topic

[Decimatos]

I do NOT have a new MMO idea and will NOT revolutionize this or any other industry, genre etc. Questions and curiosity, on the other hand, abound here.

As I explore MMO games and think about how my old favorites from the pencil and paper rpgs might work or not work in an MMO, I am trying to grasp certain basic functions -- namely, skill tests and other checks.

Of course these are fundamentally the same in any rpg. The character has an ability to do something (punch, spot, dodge, sing opera, etc.) and has some amount of proficiency in that ability (whether by class level, skill level, etc.). Circumstances and/or an opponent increase/decrease the challenge for the character to complete the action with success.

In the pen and paper world, these factors distilled into a target that the character needed to roll (under or over, depending on the system) to satsify the challenge. Because the random component comes from a variety of dice combos, the variance range differs wildly among systems. Games built around a roll of one d20 offer a straight chance in 5% increments. Those built on rolling 3d6 use a weighted probabilty spread across 216 increments.

When I read about skill checks in the computer games that I play, they tend to appear in terms of percentage (i.e., 50% to hit). Behind the programming curtain, however, do any games import other classic rpg mechanics into their machinery? It seems to me that the only difference is the random element of the formula.

Do different rpg systems import into an MMORPG system? If not, why not?

Decimatos

[monjardin]

Tons of RPGs use the AD&D rule set. I'm not sure how they implement their particular RNGs. Most probably display a percentage because its makes more sense to most people at a glance. Which is easier to comprehend in real terms: 8+ on 1d20 roll or a 65% chance?
The big problem with chances to-hit and the like are the boundary cases. If you have a 90% chance to-hit, you can still miss 20 times in a row and get killed by some weak enemy. A lot of game developers fudge the numbers to increase the fun and decrease the frustration. I have a link around here somewhere... (runs to http://del.icio.us/monjardin/) ...and here it is: http://www.gamedev.n...article2206.asp

[eddie]

Good article monjardin!

I really should troll the GameDev articles more often.

[Axehilt]

Yes, excellent article monjardin. Thanks for the link!

I've noticed a few things in modern MMORPGs which are loose 'imports' from pen & paper gaming. For instance, your chance to score a critical hit with most classes in World of Warcraft is around 5% (ie, rolling 20 on a 1d20.)

Most of the time I'm wholly disappointed when a videogame constrains itself to the limitations of a Pen & Paper combat system. Pen & Paper combat systems were designed so that calculations and dice rolling were extremely simple, to keep combat moving without much hassle. In videogames, the computer can handle any amount of complex calculations that are needed, almost instantly - so you're working in a design environment with different constraints.

The only reason you see any AD&D videogames at all is the nostalgia/familiarity factor.

Aside from loose similarities (like the WoW 5% crit example above) all the MMORPGs I can remember playing seem to have developed their own distinct systems for behind-the-scenes combat mechanics.

[Decimatos]

I submit that many of the pencil & paper systems present fairly elegant solutions to potentially complex interactions, but I agree with your comment that more complexity is available to the game designer.

In CoH/CoV, however, their system seems to basically run on a linear percentile check. After reading the article noted above (very good read), I suspect that there may be some components that prevent long strings of misses, but otherwise the basic test is a base percentage to hit that can be modified by an accuracy percentage and by a defense percentage. Sort of lackluster in my view. Pencil & paper games that used that kind of system (e.g., Top Secret) faded away long ago.

I have been wondering whether a variation of one of the 3d6 systems (i.e., Hero or GURPS) with some changes to take advantage of available processing power. For instance, instead of deriving to-hit checks and/or skill tests from one attribute (GURPS) or from 2 attributes (Hero), an rpg with more processing power could use more complex formulae to factor in several attributes. For example, the basic to-hit formulae are as follows:

GURPS = using 3d6 roll less than your attribute (usually DX) applying any skill or situation modifiers on 3d6, then defender can attempt to save by rolling less than his/her defense.

Hero = using 3d6 roll less than 11 + attacker's offensive combat value (OCV) - defender's defensive combat value (DCV). OCV/DCV = DEX/3 and can be modified by skills/powers (e.g., martial arts) and equipment (e.g., +1 OCV for rifle scope).

I find these methods superior to the percentile check because the probability curve provides big increases at the mid range, but diminishing returns at the poles.

Rolling <11 = 50% chance, but rolling <12 = 62.5% chance. That +1 yields a 12.5% increase. Further up the scale, however, the change is smaller. Rolling <16 = 95.3% chance, but rolling <17 = 98.1% (usually 17/18 is an automatic miss). That +1 yields only a 2.8% increase.

This bell-curve makes a high-end character better than a lower character, but not overwhelmingly so. Further, by obtaining the bigger increases early, characters get to the good play quickly, but can "level" at an even pace. Under the percentile/d20 systems, the increase is fixed, so you have to obtain balance through postponement. Hence, the early levels come quickly, but later advances take much more XP, feel like grinding, etc.

If computing power can handle the change from checking against 1 RNG result to checking against the sum of 3 RNG for every skill check and to-hit roll (and it seems that it can), the 3d6 models might make for a better character experience.

Moreover, with the greater computing power, target numbers could be derived from much more than 1 or 2 attributes. One of the weaknesses in 3d6 systems is that certain attributes (usually DEX) turn up more than others, so players crank those up at the expense of other attributes. GURPS dealt with this by dumping all but 4 attributes. Hero charges more for DEX than other attributes. With all the math behind the scenes, skill checks might be derived from more complex formulae that account for more attributes.

For instance, assuming all attributes range from 0-18, a Ranged Combat Value could be based on a mix of Dexterity (need to be coordinated), Perception (need to see the best place to hit) and Strength (need to keep weapon steady or to deliver the thrown weapon). So instead of:

RCV = DEX + skill

or

RCV = DEX/3 (with a check measured by RCV-DCV + 11)

you would use something like this formula:

RCV = ((3xDEX)+(2xPER)+STR))/6 + skill

or

RCV = ((3xDEX)+(2xPER)+STR))/18 (with a check measured by RCV-DCV + 11)

Different weights could be given to different attributes to reduce min/max builds and to capture more subtle benefits and hazards of imbalanced stats.

Decimatos

Axehilt said:

Most of the time I'm wholly disappointed when a videogame constrains itself to the limitations of a Pen & Paper combat system. Pen & Paper combat systems were designed so that calculations and dice rolling were extremely simple, to keep combat moving without much hassle. In videogames, the computer can handle any amount of complex calculations that are needed, almost instantly - so you're working in a design environment with different constraints.

The only reason you see any AD&D videogames at all is the nostalgia/familiarity factor.

Aside from loose similarities (like the WoW 5% crit example above) all the MMORPGs I can remember playing seem to have developed their own distinct systems for behind-the-scenes combat mechanics.

[monjardin]

You can model a distribution using a uniform random number generator and a transform. In other words, if you had a curve which represents the output you want, you can use a function that operates on the output of your favorite RNG to simulate that distribution. Did that make sense?
Unless you are REALLY into the math, messing around with the multiple die roll as you suggest is probably your best bet for run-time performance.