You could try to teach pessimism in planning as a subgoal, e.g. by someone asking the player to predict their performance on the next task, as measured by a metric that they weren’t just measured on.
For an example, consider a puzzle game where you put together pieces to reach a goal. If well-done, plenty fun. Show the player some sort of level information—how many pieces will this level take you? How much time? How many times will you have to use the “delete a block” ability? How many times will you have to restart the level? What will your score be?
You can make the game score based. The final score will be inversely proportionate to the time it took to finish the puzzle (which will make the player want to solve it faster) AND proportionate to how closely they’ve predicted the time it will take them to solve the puzzle. (Solving under the predicted time will give the maximum score for the predicted time, since the player can just wait anyway.)
Example: I guess 10 minutes, solve the puzzle in 8 minutes → I get score for 10 minutes.
Example: I guess 7 minutes, solve the puzzle in 8 minutes → I get score for 8 minutes − 1 minute penalty.
If you notice, I accounted for that. The higher your guess is, the lower your score can be. If I guess 1 minute and finish in 1 minute, I get (let’s say) one million points. If I guess 9999 minutes, then I’ll get the score for 9999 minutes even if I finish sooner, for a score of (let’s say) 2 points.
You could try to teach pessimism in planning as a subgoal, e.g. by someone asking the player to predict their performance on the next task, as measured by a metric that they weren’t just measured on.
Can you give me an example of this in game-format? What sorts of tasks would be fun as well as being predictable?
For an example, consider a puzzle game where you put together pieces to reach a goal. If well-done, plenty fun. Show the player some sort of level information—how many pieces will this level take you? How much time? How many times will you have to use the “delete a block” ability? How many times will you have to restart the level? What will your score be?
Hmmm that could be interesting… and can be attached to existing game-formats.
Predictive sudoku?
You can make the game score based. The final score will be inversely proportionate to the time it took to finish the puzzle (which will make the player want to solve it faster) AND proportionate to how closely they’ve predicted the time it will take them to solve the puzzle. (Solving under the predicted time will give the maximum score for the predicted time, since the player can just wait anyway.)
Example: I guess 10 minutes, solve the puzzle in 8 minutes → I get score for 10 minutes. Example: I guess 7 minutes, solve the puzzle in 8 minutes → I get score for 8 minutes − 1 minute penalty.
That works—though there should be a penalty for guessing overtime too—otherwise I’d always just guess 9999 minutes :)
If you notice, I accounted for that. The higher your guess is, the lower your score can be. If I guess 1 minute and finish in 1 minute, I get (let’s say) one million points. If I guess 9999 minutes, then I’ll get the score for 9999 minutes even if I finish sooner, for a score of (let’s say) 2 points.
Ah ok—now I get it.