ze's blog :: zefrank.com
Ads Via The Deck



Via BuzzFeed


« the third bubble's my fav | Main | sense of place »

March 17, 2006

something you could help with

i'm working on updating the memory games.

I'm trying to figure out the best way to score the game. Right now its just a percentage...wrong picks vs. correct picks, and even a good score is depressingly low. In this scenario only pure luck can get you 100% - you'd have to guess correctly without ever having seen the squares you flipped.

Arguably the player should be able to turn each tile over once before i start counting wrong picks, but this doesn't take into account lucky guesses, which could push the score over 100%.

I'd like time to factor into this, perhaps a bonus for faster play.

ideally the scoring would be a single number which was created using wrong picks, right picks (a fixed #), and time taken (am i missing something? - maybe the more times you turn over each individual tile, the worse your score gets...so returning to the same tile again and again would be a bad strategy).

i don't want it to be abstracted into an inflated points system (like pinball)...

any suggestions?


Bookmark and Share
Comments (5)

The easiest thing is to start with a score that depends on the complexity of the game (e.g. number of shapes*3). Each wrong guess decreases the score. A timer decreases the score.
Your game score is what's left when you complete the game.

Posted by: anon [TypeKey Profile Page] at March 17, 2006 9:53 AM

I agree with the notion of decreasing the score, but not with each wrong guess. It should only be decreased when you guess wrong with a square that you've already seen. That way guessing in the dark and hitting doens't help or hurt your score (except it helps you on time).

That way, if you ignore time for now, you could make a perfect score by remembering everything you saw. Time just runs in parallel to that.

I love those games, by the way... It is much more fun with the animations than with just tiles. Looking forward to the update.

Posted by: BarryInDC [TypeKey Profile Page] at March 18, 2006 8:26 AM

I agree with the notion of decreasing the score, but not with each wrong guess. It should only be decreased when you guess wrong with a square that you've already seen. That way guessing in the dark and hitting doens't help or hurt your score (except it helps you on time).

That way, if you ignore time for now, you could make a perfect score by remembering everything you saw. Time just runs in parallel to that.

I love those games, by the way... It is much more fun with the animations than with just tiles. Looking forward to the update.

Posted by: BarryInDC [TypeKey Profile Page] at March 18, 2006 8:26 AM

instead of starting with points and decreasing, you could start with no points and increase by giving points dependent on how challenging the match is.. it would depend on how many have been looked at, how many are not flipped over.. i suppose for those who enjoy flipping the same piece over and over without finding a match, this would not be in their favour. say 1 point for no match, 3 points for an easy match, 5 points for a hard match, and what about lucky guesses? more points or less points?

Posted by: sonya [TypeKey Profile Page] at March 18, 2006 10:49 AM

You could give each square a count. Each time a square is exposed after an initial viewing and not matched, the count for that square is reduced. The initial viewing of a square is free, i.e. it doesn't reduce the count for that square. When a match is made, the raw score for that match is the sum of the counts for both matching squares. Not decrementing on the initial viewing of a square helps to eliminate luck as a factor in the score.

For example, assume each square has an inital value of 10. If the player gets lucky and matches on the first attempt, they would add 20 to their raw score. If the pair didn't match, those two squares those square would be marked as viewed but still retain their count of 10. If the player matches squares they have only seen one time each, they still would get 20 added to their raw score. Subsequent views of those squares would reduce their counts by 1 each viewing until the count is zero. If a new square turned over was matched with one of the player has viewed twice, the player would add 19 (10 + 9) to their raw score. At the end of the game a percent score could be calculated by comparing the raw score to a perfect score of 360 (20 x 18 pairs), or Score = 100*(RAW_Score/360).

Adding scoring for time could be a just another scaler but if you wanted to be clever, you could define a time boundary of Best_time and Max_time. Best_time would be the time that the best player can complete a game in. Max_time would be the time that if exceeded yields a zero score. A 1 to 0 scaling value could be defined between the Best_time and Max_time time marks. To prevent greated than 100% score, if a player finished faster than the Best_time, the Best_time is used as their time. If they exceed the Max_time, the Max_time is used as their time.

The score with time factored in could be:
Score = 100*(RAW_Score/360) * (1 - (player_time - Best_time)/(Max_time - Best_time))

A player scoring 360 within the Best_time limit would get a 100% score. A player taking more than Max_time or having flipped squares over excessively before matching would get a score of 0% (and a bad mark on their permanent record :).

Posted by: Jeff S at March 24, 2006 5:19 PM

Post a comment




Remember Me?