Designing Optimal Strategies

With the amount of competition that many games can produce (competitions can arise even out of games without a real built in way to compete), surely it's worthwhile to carefully consider which strategies you want to be the best.  In many cases, there will be no wrong choice, but sometimes, if you don't plan it out correctly, you can make something that many people will see as unfair.

My pad didn't register the last note and you beat me by 700,000!
I totally should've won!

Any time you have scoring in a game, it's a good idea to consider how it will change gameplay when playing for a high score.  Maybe the game is a platformer.  If you get a time bonus after finishing a level, is it large enough to be worthwhile?  Is it so large that it is what matters the most?  Is there a way to get points endlessly, making the scoring pointless?  Or maybe it's DDR.  If I remember correctly, in DDR Extreme (where the picture above is from) you could get say 299 perfects and 1 good, and you would only get an A rating because you have to get only perfects and greats for AA.  Then you could get 270 perfects and 30 greats and get the AA, and the points for AA were higher than the points for an A so you would score higher than someone who got 299 perfects and 1 good.  Which didn't seem fair to me, but...  Then there's Osu/Elite Beat Agent scoring, where you get 300, 100, or 50 (or 0) points per note, multiplied by your combo.  Which means, if you want a high score, you need a full combo pretty much.  If there are 101 notes and you miss the 51st note, even if you get perfect on every other note you still won't get half of the highest possible score.  Not that there's anything wrong with that, I just wasn't fond of scoring methods like that.
And then in that Pixel Purge game, as long as the lower levels were easier than the higher levels, the best strategy was to avoid the things that made you level up and stay on the lower levels as long as possible.  Which made the game longer and more tedious, but it was also an interesting idea.  It might have been intentional because it wasn't that easy to avoid leveling up.

Also, the more difficult things to do should be better than the simpler ones.  I believe that a game should be hard enough that you have to do the more difficult things in order to win.  I mean, if you can beat a game using some simple strategy over and over, it probably isn't that good of a game, right?  Similarly if you want to have an RPG with a lot of options or a fighting game with a lot of moves, you ought to make sure that every option or move has some situation in which it is the best, because otherwise there's no point in ever using that specific one.

And then there is game balance.

If it's a Mario Kart Wii world record,
it probably uses this combo.

There always seems to be that one strategy or character that's just a little bit better than the others, but that little difference matters so much at higher levels.  My feeling is that some things could be predicted beforehand (Funky's speed bonus helps throughout the entire course while other bonuses are only in specific situations, so it seems natural that speed bonus would be the best), but most things are hard to figure out.  I mean, the strategies and glitches are often only found out after hundreds of thousands of people play the game over and over.

Speaking of glitches, apparently everyone in this game can do this glitch, so it makes this game totally balanced, right?