So, in the various places I read, there's been a recent font of essays on difficulty, restrictions, rails, and lack thereof. People seem to be tending primarily towards one idea: this idea. As usual, Darius can feel the pulse of the community.
Darius is wrong!
That's kind of strong, I suppose. Let me sum it up.
Darius backs the idea that the best games (and other works of art) simply give - they hold nothing back. Because they are such masterpieces, they can give and give without running out.
But that's not right! It's... tangential thinking. It's mistaking correlation for causation. Darius is saying (simplified and paraphrased): Because good games give, giving makes games good. There's a piece missing from the equation.
He gives the example of the gravity gun in Half-Life 2. It comes in early in the game. He gives this as an example of "not being afraid to give". Ngah!
The gravity gun is a tool to improve agency. What they gave was access to a new technique of play: the ability to throw things around. Following the gift of the gravity gun were a slew of directed challenges revolving around it. And I'm not talking about the tutorial section: I'm talking about every opportunity to hurl explosive canisters, or spinning saw blades, or play with boxes.
Similarly, with Braid: the designer gives and gives. Each section grants a fun new meta-ability related to time. He's not simply giving. He's giving a tool, and then forcing the player to use it in order to meet these directed challenges revolving around it.
Now, with Katamari Damacy, Darius argues that unlimited time should be an option from the beginning. I disagree strongly. Do you see the difference between the previous paragraphs and unlimited time in Katamari Damacy?
Yeah, giving the players unlimited time is doing the reverse of what we just talked about. It's about not giving the players a tool, and in fact weakening the directed challenge.
Some challenges need to be weakened - restarting at level one is, in fact, a pain in the ass. (As a side note, Bubble Bobble doesn't do this, at least not as I remember it from NES years. Not only could you continue forever, but I think you got passcodes every 10 levels or so.) But that has nothing to do with giving - it's actually taking, from a game design standpoint.
It's not that good games give. It's that good games give opportunities. The player pursues them, uses them, masters them with some level of difficulty. Then the game offers new opportunities. New challenges.
The gravity gun is only fun because of how it interacts with the game world. The game world is "bumpy" - full of interesting navigational dilemmas - and the gravity gun gives you a way to approach them. The same thing is quintuply true of Braid. The fun time mechanics are only fun because they are tools with which to approach the challenges and opportunities in the game world. (I find challenge and opportunity are, in many ways, the same concept in game design. Choppertunity!)
Moreover, all good games actually introduce new approaches to the game at a fairly slow pace. The more "choppertunities" a tool offers, the earlier in the game it should be granted, because the more play a player will get out of it. Obviously, many tools build off of other tools, and the earlier tools have to be introduced first... but however you do it, the player needs to be given a significant fraction of the challenges related to the tool before he can be considered to have "mastered" it and move on to the next tool. Before you can "give" again.
Something which just gives a statistical boost, such as a plasma cannon over a machine gun, doesn't have very many unique opportunities. It shares nearly all its "navigational capabilities" with the generic machine gun... so giving or taking it is mostly a reward situation rather than an opportunity situation. It doesn't really fall under this theory...
I feel like I'm being incoherent. But I hope I was clear enough.