Why I play for High Score? The game outside the game.

A lot of games will, as a seeming after-thought, give you a score based on how you performed.

I guess it's supposed to give you some indication of your performance and how well you've performed. But a lot of these scores are meaningless to the average player. High scores, in a void, mean nothing. A lot of online mini-games will give you scores for your performance, but are those scores meaningful?

It's true that scores don't *need* to be meaningful, simply a measure of the player's arbitrary ability to play. But at the same time, a well designed score system can be very good for the game's meta-game. The meta-game is literally "the game outside the game." The game of competition to see, not who can beat the game, but who can score the highest. Getting to level 25 in Pac-Man is playing the game. Getting to level 25 in Pac-man collecting every single piece of fruit so that one can put your initials on the high scores list is playing the meta-game.

The meta-game can inspire people to do better, think harder, and come up with new and interesting strategies to play the game. The perfect example of this is Tower Defense, where simply beating the game might not be enough for players. One must beat it with the highest score or the best possible strategy. The meta-game thus extends the life-time of the game, by encouraging people to improve their game.

What properties do score tables need to have to ensure this?

1) Scores must be public.

There are far fewer rewards in having a score that no one ever sees. A little private assurance. However, scores are just numbers, and a number without context is essentially meaningless. Having a public high scores table also means there's the possibility that you could be a high-rank in those high score tables. That people will know about your accomplishments.

The desire to beat other people and the desire to achieve a rank on that high score table are very Spike goals. Without this kind of motivation, Spikes have little to gain from a simply score.

2) Scores must be a fairly accurate representation of performance.

This is fairly simple. If I kill 3 guys, and he kills 3 guys, we should have roughly the same score. If I kill 3 guys and he kills 3 guys and we have a wild score discrepency, the score becomes meaningless as it becomes less understandable to the general player.

A high score automatically implies certain things: A mastery of the game, or a long time to achieve the feat, or simply an efficient performance.

If high scores fluctuate wildly, then it becomes meaningless. If I said Player A had 5000 kills, and Player B had 4339 kills, that is a fairly good score representation. If I said Player A had 129387129387123 million points while Player B had 334934 points for the same amount of kills, you would have no idea what those points meant.

This also implies that there shouldn't be a point-farm where the only measure of success if to see how long a player can abuse a specific pattern to acquire a large amount of points with minimal effort to himself. (Like, hanging back and just shooting upward at the spawning enemies... forever.)

3) A high score must be achievable.

There is no point in keeping the top 10 scores of all time, as it only makes players feel that they can never reach that point. Why post the 45 millions and the 67 trillions when most players reach the thousands at most? We can't all play perfect games of Pac-man every single time.

The newest trend that seems to be working quite well, is to post the daily or weekly high scores. This way, you are ensured of competing with people of relevance. (Oh my god, I'm 15th out of all the people playing, right now?)

4) Scoring should not be based on some exponential system, scoring systems should strive to be linear.

If you make it to level 10, you should not get some exponentially bigger score than someone who made it to level 1. The key to this is again, understandability. If the first place has 100 times the score of the other top 10, that's bizarre. If he really 100 times better at the game? Human beings are terrible at understanding what an exponential increase really means. Thus, if at all possible, try to keep scores linear. If he gets to level 10, he should have roughly ten times the score of someone who got to only level 1, not hundreds. That way, you can easily tell from the score what kind of progress that person has made through the game.

Good: Wow! That guy has 10,000 points where I only have 5,000. He must have gone like, twice as far as I have.

Bad: That guy has... 100 times the score I do.... I have no idea what that means.

5) If a public high score isn't achievable, then smaller goals (or ranks) might be sufficient.

There is nothing better than being complimented by the game for doing well. By giving players ranks (like, silver star, gold star, A++.) You provide an additional incentive to do well. I mean, sure some people can be top 100, but if I can get a gold medal on this game, it's good enough for me. This sets the 'good enough' point, where the player might not be the best, but he's hit a goal that he can accomplish with a little effort.

(For example: If you survive 15 seconds longer, you qualify for a gold medal rank!)
(Especially, if said gold medal rank unlocks something)