Anyway, on with the show...
Early game prototyping can be thought of like walking an insanely active dog; you want it to be happy and go about following it's desired course, but you need to keep it under some amount of restraint, otherwise it gets away from you.
While I have no concrete evidence of this, my personal experience is that good games (like most creative processes) develop "on their own" fairly naturally with only gentle shoves and yanks of "the leash" as guides. It's a careful balancing act, games left to run by themselves become a mess pretty quickly, and require quite a bit of leashing mid-way in the design. And games that are under the heavy hand of the designer leash, with no intentions of wavering away from the original plan. Well, they don't go any where much at all. Every game project that I've been involved with professionally was never, ever considered to be "done" by the creators; just "finished enough" given the tightrope walk between the various constraints of schedules and finances on one side, and creative design work on the other. They are never perfect, but close enough to almost be happy with the final the output, as there is always at least one more bit of polish to clean up, or one more feature to add that the game absolutely requires, but at some point, as a designer, you have to stop, and move on.
The development process usually works like this (of course, your mileage may differ). Early on, after settling in on one or two "gimmicks" that the game is going to based around, I let the game go, adding features, or doing whatever the game feels like it needs. Ultimately, the gimmicks are the leash; these are the things that will guide the decisions made for the rest of the design process. But I allow the game to go off and explore other possibilities. Until, of course, the game has now become so bloated, or confusing, or misguided, that now is the time to yank the leash, and get the game back on a course. Editing out extra rules, or components, or things that don't seem to affect the original gimmicks, etc.
In most cases, letting the game "roam free" doesn't happen and develop in a vacuum. They require the ability to bounce off walls, knocking into other people, and going through various filters that other people have, through your usual playtesting experiences. In fact, the first few playtesting sessions of a game probably should mostly focus on the main gimmicks, making sure that they even work, without much attention to the side effects generated by the game running it's own course. These additions (and subtractions) should naturally come and go as they please.
Recently, I've been playing around with the Shipwrecks game (the current version can be downloaded off in the "Things To Play" section in the sidebar, listed as "Dark Water Salvage"). There's been a nice progression of the main gimmick, along with various levels of fluctuating game-defining rules.
Luckily, I'm a member of the Board Gamer Designers Workshop, which is a small group of geekiness dedicated to just playing game prototypes of our own creation. A sort of a bi-weekly Protospiel. So, there's no need to torture real players who might not understand the concept of how prototypes work.
As is usual of my designs, the game went through a few revisions even before the first playtest. As I've noted in a previous post regarding the use of windowed cards for hidden data, the main gimmick of the game is hunting for shipwrecks that are "hidden" in a lake. This hidden information system was originally being developed as a haunted house game; but things "weren't meshing" well at that point. This system initially was testing for directions using simple binary coding (Y or N). The binary "bar code" worked with deducing items back when it was a haunted house game, but now the game involved searching for a location, and the game cried out for a modified version of the system that more closely relates to that action.
And so, while the gimmick was kept the same, the data was reversed. By testing from various locations on the game board, you can get directional information regarding where a location is in relationship to the testing location. In this case to the left, the location is Southwest of the city of Missaukee (when Missaukee is stacked on top of the location card). So, while "the leash" of the stacked cards revealing hidden information is kept, the game was still allowed to form naturally around it.
During the first playtest, one of the main focal points was watching to see if this mechanic would actually work (it did). Other elements of the game around it did not, but that's fine. As the core interest point of this game is based on the the searching and locating function of this mechanic, those are adjusted or removed, and tested again.
At this point, I'd like to point out that there's almost no sense of trying to balance scoring, or creating a more "interesting" map, or any kind of fine tuning. It's an early prototype, and the basic functions of the game are still being fleshed out. In fact, after finding a major rules break in the second playtest, it became obvious what some fixes were required, and what was still broken...there was no need to finish.
As an additional aside...
One of the games we played was very much of a loosey goosey party game, whose main goal was seemingly guessing random numbers. The only information that you had to go on was that the current number to guess is less than the previous number on a card. Basically, no strategy or skill, just wild guesses for the most part. And while the designer was happy with what he implemented, this IS a group dedicated to board game geekery, and simple random guessing games don't cut it.
The results that came out of the discussion at least sounds like an interesting game. At least playing the game mentally. It still keeps it's party game roots, but has some elements of a mind game. For the most part, it was an idea that was "let loose" to see where it would go. As is typical of a brainstorming session, no leashes are attached.
For the fun of it, I've dubbed the name of the "Vezzini," in honor of the greatest mental skills battle ever filmed, the Battle of Wits from "The Princess Bride."
I imagine that the game winds up being played in a "well, I picked this because I think you picked that because I figured you'll think I'll pick this" kind of manner. I also imagine that the game could be potentially be terrible, too. Of course, this is solely what I consider to be the best version discussed, I'm sure other would have different opinions.
A (potentially irritating) battle of wits for any group of players
Give every player a pad of paper and a pencil.
Someone will also need to keep track of a running point total for each player on a separate sheet of paper.
Agree to a Target Score that determines the winner.
The game is player in rounds. In each round, the players will first determine who the Host is. Once a Host is determined, all players then battle the Host.
DETERMINE THE HOST:
All players write down a number secretly on their pad of paper.
All player's reveal their number.
The player who wrote down the LOWEST value number, that is not tied with another player's selection, becomes the Host. All other players become Active Opponents.
The Host scores points equal to the number he wrote down.
BATTLE THE HOST:
A Battle consists of three rounds, in each round the host secretly writes down a number that MUST be lower than the number he previously wrote down. In the case of the first round of the battle, the number he writes down must be lower than the number he wrote down to become the Host.
All Active Opponents now write down their guess. Opponents can ask questions; the Host can lie or bluff, or refuse to answer.
Once finished, Active Opponents reveal their guesses, and the Host reveals his number.
If an Active Opponent guesses correctly, they stay Active; an incorrect guess renders them Inactive, and out of the remaining battle rounds.
The Battle ends when all Opponents have become Inactive, or after three rounds of Battle have been played. All Opponents that are still Active after three rounds (they have successfully guessed the Host's number three times), win points equal to 2 times the amount of the initial number the Host selected to become the Host.
THE GAME IS OVER:
...when a player has earned points equal to or greater than the target score.