Raph Koster's Home Page

Essays
These are full-blown essays, papers, and articles.

Presentations
Slideshows and presentation materials from conferences.

Interviews and Panels
Reprints of non-game-specific interviews, and transcripts of panels and roundtables.

Snippets
Excerpts from blog, newsgroup, and forum posts.

Laws
The "Laws of Online World Design" in various forms.

Timeline
A timeline of developments in online worlds.

A Theory of Fun for Game Design
My book on why games matter and what fun is.

Insubstantial Pageants
A book I started and never finished outlining the basics of online world design.

Links
Links to resources on online world design.

All contents of this site are
© Copyright 1998-2010
Raphael Koster.
All rights reserved.

The views expressed here are my own, and not necessarily endorsed by any former or current employer.

On dynamic virtual worlds

...My point was that even saving world state doesn't mean much if the state isn't very different. And I count item location as not very different. Now, if that item's significance had changed in some way (beyond what significance players attach to it). If say, it resulted in changing economic conditions, differing NPC behaviors, etc--actual change in that middle layer I mentioned earlier--then maybe I'd term that altering the environment.

[Martin] Keegan's paper in the newest Journal of MUD Research terms full-reset-based muds "Groundhog Day" muds. Cute term (though I think Ken Grimwood's novel Replay was ripped off by the movie.) but it serves to underscore something fundamental:

You make an item.

It either spawns back when destroyed, or isn't destroyed, depending on world-state model (repop versus save-state).

But when does it EVOLVE? Now, some may say that in mud architectures which permit dynamic attachment of scripted behaviors by players, stuff evolves. To get back to my example of TinyTIM, that clock or whatever it is at the entrance that by now is massively huge. But that isn't evolving by itself. That's just more functionality slapped on it by a builder of some sort. It is not dynamic; it is static save for outside intervention. It is therefore predictable.

For muds to evolve, they need to become unpredictable.

The "middle layer" referred to is the one of the three layers of significance in a virtual world: that which is the static database, that which is the play of data on said static database, and that which is the dataset of players themselves. Depending on the method of state-saving of the mud, they may save one, two, or all three of these things.

As an example, a classic Diku saves only the third layer: players. Ultima Online saves both players and the environment. A typical MUSH saves players, the environment, and also the actual map and setting which holds the environment, since MUSHes allow dynamic alteration of the static data. Some early muds did not save any of the three, and thus were not truly "persistent."