Theme vs mechanics at the start of the design process

A question that seems to often get asked of designers is "what comes first for you, the theme of the mechanics?". It seems to be the boardgaming equivalent of the cliched music jorunalism question, "so what are your band's influences?".

This is an important question because most gamers are very interested in how theme and mechanics work together. Knizia is accused of pasting on themes, some American games are accused of having fiddly chrome-laden rulesets. These observations beg the question in the player's mind, how are games born? Do themes or mechanics usually set the design process in motion?

While I am no authority at all on game design, I thought I would share my thoughts on this issue about how the design process works out for me.

It seems to me that initial ideas for games must fall into one of three categories: When the theme is the first thing to arrive in the designer's mind, when a raw mechanic comes first, and when the process begins with some combination of the two.

I thought I would share my thoughts on how each of these three processes has worked for me, and something of what I have read from other (more famous) designers.

Theme comes first
Occasionally something will spark the desire to make a game with a very specific theme. For me this usually comes from watching a movie or something on television. A particular scene or situation might just spark my imagination, and I immediately get an idea for a game. Seriously, every viewing of Star Wars could yield ten game ideas: Escape from the Trash Compactor! Light Speed Hide & Seek! The Death Star Maze Game!
Usually this initial idea will suggest some basic things about gameplay, such as the aim of the game, or some foundational mechanics. Then it's time to start scribbling in the notepad, and try and flesh out the flow of play. Sometimes spare mechanics that are lying around will be plugged in to the design. Sometimes new and fresh mechanics will naturally arise out of the theme. This for me (if it happens) is one of the most exciting things in game design.
I often find that ideas which come to me theme-first also often fall flat pretty quickly. Sometimes there just doesn't seem to be any simple solution to a problem in the design, and it is very hard to properly squeeze mechanics into a theme if they don't naturally fit! If you start with a strong theme, you are in some sense 'bound' to reflecting it in the gameplay well, and if you can't, the design usually falls by the wayside. For me I feel I am selling the idea short if I just settle for a mechanic that will do the job without integrating with the theme.
A real world example of this would be Knizia's Lord of the Rings. Reiner was commissioned to create the game and he has said that after studying the book he realised he simply had to make the game co-operative. The theme came first and dictated the base mechanic. Now of course developing the smaller details to make the game flow how he wanted must have been very tough, and people debate how successful he was.

Combination of theme and mechanic together
Sometimes a game mechanic will arrive in my mind with some aspect of theme inseperably attached to it. These ideas usually come when I am actually thinking about games and game design. One recent example was a tile laying mechanism that popped into my head, but what I was imagining on the tiles seemed to only work if the game was set in a cave. Odd huh?
Now the cool thing about getting ideas this way is that there is some leeway with what you do with the development of the theme. Is it a pirate's cave? A tunnel through the earth? Ancient catacombs? Each possibility allows you to move the design forward in various ways. There is the opportunity to take the design down a number of different paths and see which one works best.
I often find these designs develop well and end up yielding nicely playable prototypes. The fact that the mechanic is automatically linked to a theme helps the gameplay feel somewhat natural. Of course, more arbitrary elements often slip in as a final theme is chosen and the game system develops.

Mechanic comes first
This is pretty rare for me, and these ideas come either when I am thinking about game design in pretty concrete terms, or when I am thinking about some mechanical process in real life. For example, watching the switches on train tracks as I take a train trip may suggest to me some sort of abstract strategy path-swapping mechanism.
I find that these ideas rarely develop straight into a game of their own. I usually just remember the mechanic or take note of it. Sometimes I later find a good situation in a different design were it makes sense to insert this mechanic. On some occasions these mechanics are interesting enough that they can become the comeplete basis and centre of a game. In hearing Ted Alspach speak about Seismic, it seems that for him the mechanic of laying tiles to build a path way was such an exciting one that the whole game design flowed out from it.
Amd again I can't help but think of Knizia. From my first play of Lost Cities I was sure he must have come up with the game idea while playing around with a standard deck of cards. But while the theme does feel pasted on, the real fun of the game comes from its exciting set-collection mechanism. I am sure Reiner realised that this mechanic was good enough on its own to simply 'be' the game.

What seems to work best
The games which for me have come the furthest in development are the ones where something like 'chain reactions' have occured from one stage of design to the next. What I mean is, that the design begins in one of the three ways listed above, but then at some point hits a wall, and one aspect of the design takes me down a totally different path. This may happen two or three times before the game really takes shape into something that I feel can be great fun to play.
The development of my game Archaeology is a good example of this. The game began with a mechanic, which was a modular board which created an island out of sqaure terrain peices. I am sure this was probably suggested by Settlers of Catan, but it also came from an old computer game I programmed in BASIC as a child. Anyway, after playing around with the mechanic I eventually decided this would be an island where explorers came to hunt for treasure. So I added treasure tokens which were placed face down on each terrain sqaure, and turned over until collected by being landed on by players. During playtesting the game developed into a fully fledged pirate epic, with sailing, weather conditions and all sorts of goodies. I eventually lost interest in the design, as I felt I could never come up with the correct mechanism for player movement, and the interaction between players seemed doomed to be very low due to some core design decision I made early on in the process.
What stuck in my mind though was the fun that there was in collecting a mystery treasure each turn, and the way the different types of treasures combined to be valuable. From there I started a new design that was largely based around this mechanic. The board was vastly simplified, and I soon realised the theme of an archaeological dig made much more sense. However I again had problems with player movement and late one night decided to toss the whole board in and basically make the game a card game!
Things took many more turns (and a board was later re-introduced!) but the finished product is so far removed from my island mechanic that I can scarcely call them the same game. I feel that Archaeology is a better game for going through this process, as I feel that it evolved organically. Because I threw away so many big chunks of gameplay I was always dealing with my favourite mechanics and theme ideas, and this I think made for a more creative headspace in which to design.
Something similar to these 'chain reactions' seems to have happened for Teuber with his Catan/Entdecker/Lowenherz design, and with Knizia and Ra. One design led to another, which led to another, the gameplay being distilled and refined with each iteration.