The Deep Physics – Intro

There was a recent change to Planeshift’s crafting system. It received criticism from a portion the playerbase, largely those players felt that it added too much realism at the expense of fun. This is a common complaint and it is certainly understandable. Anyone who has ever grappled with the grappling mechanics in 3rd edition D&D understands this. Longtime MMO designers, such as Brian Green and Damian Schubert regard realism as a false god; one that gets in the way of gameplay.

What exactly is the problem with realism?

One of the delightful things about the old first edition AD&D was the smorgasboard of detailed, realistic special case rules for everything from loot in NPC pouches to detailed (and graphic) critical hit and critical miss tables that took seemingly everything into account. Rolling critical hits using the optional rules from Dragon Magazine was a mini-game in itself. Half of the Dungeon Master’s guide and almost all of the later rulebooks from the Fiend Folio onwards were an anthology of these special case rules. AD&D was in the tradition of the fantasy literature of Doyle and Tolkien. It had started as a game and had layer after layer of realism added and nurtured my simulationist, world building urges. It also had no sense of balance; that word that seems so important today. Viewed strictly through the game prism, AD&D was a horrible hodpodge. 2nd edition AD&D was the first attempt to simplify this multi-headed hydra, but it was not until the 3rd edition that D&D’s tyranny of special cases was substantially contained and even then – as with the grappling rules – this was not entirely so.

The problem with realism is the same problem that 1E had. In 1628, the Swedish navy’s newest warship set out on its maiden voyage. The Vasa was the king’s pride and one of the finest warships in the world; for about an hour. As it sailed out of port, it capsized and sank. It was recovered and restored during the 20th century and is now a museum exhibit in Stockholm. The reason it sank is that King Gustavus Adolphus insisted on putting more cannon on it than it was designed for, making it top heavy. Building lots of special case rules on top of a game is like building a top heavy ship.

Most modern persistent worlds are viewed as games rather than places. The exceptions are the social worlds such as Second Life; which follow in the footsteps of the MOOs of yore. A thematic adventure world that is a place first and a game second goes against the grain. This was not always so. Twenty years ago, most worlds billed themselves as simulations. In my lightweight rulesets post a few months ago, I suggested using physics and a lightweight ruleset instead of a heavyweight ruleset. The universe as we know it arises from a handful of rules. If you look at a ball game, such as tennis, something comes to light. The rules of the game define the boundaries of what good play is. The rules never define that a level 60 tennis player has a 95% chance of putting the ball on path X and that a level 55 player has a 80% chance of playing off of it. Newtonian mechanics defines how the ball behaves. The player simply chooses how he hits it and the designer simply defines the dimensions of the court and the scoring system. Unfortunately, most PW designers – as is most of the population in general – are woefully scientifically illiterate. Most people regards science – and physics in particular – as “complicated”; failing to understand that science – and physics in particular – is all about simplification. Because of this perception of complexity, when most designers do create realism rules, they invariably make them overcomplicated; and end up creating a Vasa.

The purpose of this post is to introduce a new series on building the physics which designers would then build the “rules” on top of. It is named in honor of the term CS Lewis used in the Narnia Series. Naturally, these will be lightweight rules and naturally, this series will have a simulationist rather than gamist bent. If you are a “game” designer first and foremost and enjoy the game design challenges that Dan Cook and Brian Green like to post on their blogs, then this series is probably not for you. If a framework where you can water down the wine for a little extra profit at a fool’s expense, tinker with the balance on your sword to bet match it to your swing style, smell the campfire, and feel the drizzle on your skin as you see the ruins through the mist sounds interesting to you, then this series is for you.

Lightweight and Heavyweight Rulesets

Eric Heimburg recently wrote a very nice piece on how 4th edition D&D rules simultaneously reduce the value of loot and make it a part of the mundane mechanics. Brian Green picked up on this and took the angle of how this is a case of MMOs affecting the design of pen and paper RPGs instead of the other way around. He went on to discuss how modern MMO (and some PnP) rulesets are a consequence of the constraints of software logic.

I can see the logic behind this, but I have to disagree that all of the options must be hardcoded. Let me try to explain my position a bit further.

First edition Dungeons and Dragons (1E) was what I would refer to as a “lightwieght” ruleset. Sure, by the time the transition to 2E, those rules had grown to thousands of pages of supplemental material with a potentially complex, ponderous and schizophrenic ruleset. The point was that all of these supplemental rules were optional and the DM could cheery pick from them. In fact, the at least half of the original DM Manual was a codex of tips for DMs to run campaigns, more than a rules manual. In 1E, DMs were expected to either make up rules on the fly, use house rules or both. You could think of it as the common-law legal system used in Anglo-Saxon and Commonwealth countries; where law is messy, murky and based on precedent.

Computer based rulesets (and 4E D&D, though 3E already had a tendency towards this) by contrast, traditionally assume that whatever is not explicitly in the ruleset is disallowed. These are “heavyweight” rulesets that try to form a complete, self contained game. You could think of these as Napoleonic Code based legal systems, where the law is based on theory and tries to be predictive of the future. The problem is that you can’t predict everything and emergent gameplay will happen. Due to their complexity, heavywieght rulesets are also prone to emergent gameplay; but this is often of the “unbalananced” or “exploit” variety. The fundamental difference between a lightweight ruleset and a heavyweight one is that the former embraces – even requires – emergent gameplay while it is often a QA issue for the latter. The freeform emergent gameplay of lightweight rulesets requires a quick thinking GM.

Except that it might not.

Think of all of the different kinds of judgment calls that a typical GM made in old school pen and paper RPG sessions. I vividly recall a session that took place in San Diego a quarter century ago. A cousin and I had a ranger and a fighter in a low level dungeon. Another cousin was DMing. We were fighting a pair of goblins. I rolled a 1 on a two hit roll and the cousin that was DMing declared that my sword broke. Suddenly the situation looked desperate. We had just picked up a loot bag with two hundred copper pieces in it a short while before. So I announced that I would grab the loot bag and swing it around, wielding IT as a weapon. The next round, my playing cousin scored a killing blow with his weapon and the DM asked me to roll to hit. As luck would have it, I got a natural 20. My DMing cousin rolled some dice and then announced that I had walloped the last goblin so hard that not only did I kill it, but that the bag ripped open showering the corridor with coins.

Consider the kinds of judgments that my cousin had to make as DM:

1. Was the bag wieldable as a weapon? His sense of naive physics told him that you could indeed wield a loot bag as a flail like object.
2. His other judgment was that the burlap bag was not strong enough to withstand the stresses associated with impact and would rip open.

In short, he was making judgments about physics. Most GM judgments involve either NPC psychology (the AI) or physics. Can that PC actually swing from the chandelier and knock the NPC over? AI is hard. We have 50 years of AI research and precious little to show for it in terms of creating artificial personalities. Physics, on the other hand, is comparatively simple. Yes, you are reading me right. Physics is comparatively simple. What is not simple is the usual approach to adding more and more physics to games. It mostly consists of adding more and more special rules. This introduces complexity in a geometric fashion. The fundamental problem here that that the traditional reflex action is that of a “game designer”. The game-designer twitch moves us towards heavyweight rulesets.

This is a real (albeit paraphrased) conversation I once had about encounters for a NWN2 server:

Me: Y’know, we have a situation here where it is highly likely that we want no encounter to occur, an encounter to occur some of the time and sometimes there are a bunch of encounters. Why don’t we solve this in an elegant way using a Poisson distribution. Let lambda be the most likely number of encounters for a given area and something settable by the dev. K (rounded down) will be the actual number of encounters.
Other Individual: *looks at equation* no way! Too Complicated!
Me: Huh? One value set by the builder per area, one calculation and a nice fluidly scalable…
Other Individual: Okay, if the PC is below 10th level, then we use this table here. We roll a D6 and add one to the roll. If it is below 4, no encounter, above 6, multiple encounters. Then we roll on this table over here to see how many encounters there are. If the person is above 10th level, we add 4 and use a different table. If there are low levels traveling with high levels, we’ll add d6 to the original roll instead and then flip a coin on the table to use.
Me: …

I am a software engineer by trade. By education, I’m a physicist. Physics is not about creating big, complex things, but rather simplifying and clarifying based on “first principles”. Depending on the level you detail you want to work from, there are perhaps two dozen fundamental rules that could use to describe the world around you. That world is the ultimate lightweight ruleset. It has no formal game rules. Everything is emergent gameplay from the physics. Chapter four is my far my favorite one in Bartle’s Designing Virtual Worlds.

It is possible to implement a platform where lightweight rulesets are viable on the computer. The first thing we need is a clean vocabulary for defining naive physics. As I said that the top of this post, this is near and dear to my heart. Memotica is all about making it possible to build the underlying physics of worlds in a clean way. The second and more important thing we need to do is get rid of the game designer paradigm and replace it with a “set designer” paradigm. A game designer creates a game. This is appropriate for someone creating a boardgame. A set designer is creating a themed space and the props to go with it where the kinds of gameplay that arise are not always explicitly created by the designer. As long as designers eschew realism in one breath and then add a hundred special cases to their heavyweight ruleset the next, MMOs will always have a “what not expressly allowed is forbidden” feel.