Udo's Techblog

Dangers of Wrong Rule Sets


Last time I talk about creating a dynamic, thoroughly simulated MMORPG world that might fulfill the goal of personalized storytelling on an individual level - in contrast to the classic pre-defined games where every piece of content is planned and scripted. Though the idea of having a dynamic content model that is largely self-contained and self-hosting is very inviting, there are certain pitfalls with this model, as you can imagine:





The Desert Syndrome


We cannot simulate evolution from the start, so we do not really have the benefit of using a tested, stable set of rules like the ones governing our physical world. Instead, we need to make a lot of simplifications and guesses to implement our rules so they can work in a computing environment. Stable environmental systems have a lot of feedback loops that allow them to compensate for variations and even to recover from catastrophe. If we try to build such a system from scratch it might very well crash, making a lifeless desert or bizarre carricature out of our sim world. An ideal ruleset provides subtle but powerful feedback mechanisms that act as automatic lines of defense against this. Also, there is nothing inherently wrong with implementing hard-coded safeguards against disaster.





Bizarre World Syndrome


When executing the simulation, there will always be mechanisms that are unexpected and unaccounted for. The more complex the system, the more surprising its behaviour. And this is, after all, exactly why we do this whole thing. But the path that leads to a consistent and interesting game world may be very narrow. There will be very many paths that potentially lead to bizarre and unplayable worlds. There is really little that can be done to avoid this through planning, except maybe we should see to it that the rules stay simple. A rule that is complex can (and will) hide mistakes in it that can lead to very strange results and unforeseen interaction with other rules. Complex rules are hacks and a sure sign that something is wrong with our design.





Oversimulation/Undersimulation


We must resist the temptation to simulate every little detail, because computing power is a real constraint. Also, the more detailed our rules and mechanisms, the more difficult it will be to get the world up and running in the first place. Likewise a world with to little simulation is going to be boring, static, and too predictable.


Besides constraints in raw computing power and storage space, an over-simulated approach is going to be very costly in terms of development and maintenance.


There needs to be a fair compromise between complexity and simplification. A good way to find the balance may be to start with an overly simple model, and then add mechanisms and features over time until the simulation is sufficient.