Say you're making a turn based strategy game. Every turn a player can Attack or Defend. If they Attack, they do 2 points of damage. If they Defend, they do 1 point of damage but block 1. If we do the math, it seems like these two choices are basically equal - two evenly matched players will kill each other on the same turn no matter which strategy they take.
But you will probably find, in playtesting, that players prefer to attack than to defend. Why? Are they just bloodthirsty?
No: it's that they want to win the fight as soon as possible, so they can get to the next part of your game, so they can finish your game, so they can move on to the next entertainment. You'll probably find that you have to overweight your 'defend' option to make it attractive, because players would usually rather lose a few resources than spend extra time 'savoring' the game by playing it 'optimally.'
Andrew Rollings called this a 'shadow cost' - the time the player takes to defend being an additional resource. I personally get a lot of mileage if I don't think of it as a shadow cost, and think of it as the real cost: your player is spending a real-world resource, time, in order to win virtual in-game resources, to extract the value out of your game. When you think of it that way, it's the in-game resources that are shadow costs.
This is different from the way things used to be. Back in the days of arcade machines when you put a quarter in and wanted to get as much entertainment as possible out, things were the opposite. Most players wanted to extend their time playing. We would start on the easiest level and bottom-feed, because that was the way to make our quarter last the longest. The quarter being another real-world resource that we wanted to savor.
But in today's highly cluttered attention economy, most players want to play a given game as little as possible - that is, they want to extract the maximum amount of value from it (see all the cool content, learn all the cool strategies, get to where their skill plateaus, get the high spot on the leaderboard, get the badges, whatever it is they value) for the minimum time investment. I know that sounds backwards! If you've made a game I like, shouldn't I want to prolong and savor my enjoyment? But I don't - I want the enjoyment rarified and compressed, all the great stuff in your game injected into me intravenously and instantly, so I can get on to the next game, TV show, job, book, quality family time, and enjoy that.
With that in mind, those two seemingly equal game mechanics are not equal at all. One can end the combat in five turns; the other will take ten or more turns. It wastes my valuable time. Healing mechanics have the same problem - should I heal or finish off the enemy? Healing wastes time - I'll only do it if I have no other choice.
Here's the nonintuitive part: this means shorter games are a better fit for most of us time-and-attention starved people. Making a game arbitrarily more difficult, forcing your player to replay sections they've already beaten, game mechanics that encourage taking it slow*, and most kinds of grinding probably do not add value to your game. You might ask, "How can that be? Reviewers always complain that games are too short! RPGs love to shout in their marketing materials that they have fifty or a hundred hours of content!" And, well, I don't know: maybe it's that we're talking about two real-world resources there, gameplay per dollar, and because true gameplay value is so nebulous it's easier to talk about hours. But I am willing to bet that almost all of those reviewers, given a choice between ten hours of Portal and a hundred hours of a mediocre RPG would take Portal!
I've got lots more to say about this and will, soon...
* What about slow-motion or 'bullet' time? You might ask. Isn't that slowing the game down and therefore decreasing value / time? I think you'll find that most of the games that have a bullet-time mechanic take less real-world time to play if you use the mechanic than if you don't, because the bullet-time makes the game that much easier, you're repeating fewer sections.