Text is special. It can do things that voice acting can't; it can do things visuals can't. (How? Well, Kenneth Young explains this really well in a post about dialogue in Dragon Age)
I recently played an piece of interactive fiction that can't possibly be made outside of IF -- "Byzantine Perspective." It has a single really great puzzle in it, which it won an XYZZY award for.
The premise -- you're a burglar trying to steal a valuable chalice in a museum. You begin in the vault room and type "take chalice" -- and your hand passes through it. Something's not quite right.
(SPOILER ALERT! Make sure you've either played it or don't intend to play it before continuing...)
An important part of being a designer is being able to imagine the user's view and thought process on things; to recognize your own biases and try to find out what's really going on. So, here's how I imagine most players going through this:
1) realizing you're wearing goggles
2) realizing that the "location" you see isn't the actual location you're in
3) working out the relation between your apparent location and your actual location
4) dropping the piece of paper in a room and leaving so you can look at it
5) pushing the button on the goggles to rotate the view
6) seeing the vault code in the security office
7) inputing the vault code
8) grabbing the chalice
Most IF veterans (or even newbies) will know to "x self" or "i" immediately, so (1) is quickly reached. (2) quickly becomes apparent as you begin walking through walls and encounter invisible walls.
Then everyone gets stuck on (3) because it is bloody impossible without knowing -- (a) what "byzantine perspective" is, (b) that the actual museum layout is in a separate .PDF file that is never referenced, (c) that there is a "feel" verb, which isn't commonly used in most IF.
I would solve (b) with a "remember" verb that has the player character recall the actual museum map layout. As a cat burglar, of course the character would've studied the map the day before the heist. The biggest problem with not having the museum map accessible is that you would never know there is a "security office" to help you open the vault because all you would ever encounter in-game is a "locked door."
Currently, without (c), the player's only clue to orient themselves is the bathroom door, that they automatically "push" through the door when they're clearly not in the bathroom yet.
But (c) is also the reason why this game could only be done (well) through text. I thought about how I would adapt this to a first person shooter, as my first instinct would be that it would be really cool.
But there's no "feel" verb in video games. The closest analog would be running into walls and hitting collision models.
... Hmm, actually, that doesn't sound too bad as an experiment. Not terribly fun for a real developed game, since the main player verb here would be "run into walls and bump into things," but a good experiment. Maybe I'll make it after all.
The game starts: black. Then, a green NVG-styled flashlight turns on. The chalice is in front of you. You go to grab it and end up walking right through it. At first, most players will think it's a bug, a horrible gratuitous bug. The collisions are all screwed up! What the hell is wrong with Robert Yang!
But I think the main value of the adaptation would be playing with the FPS genre convention of the "invisible wall." There's also a lot of other stuff that I'd have to change:
- Every room would have to be very different in floorplan and layout. The FPS verb "walk" would have to serve as the equivalent of "feel" as you discover the collision model is very different from the visual model. This would solve problem (b) because it would be easier to build a mental map of the museum, or at worst you could just put up a map placard on a wall somewhere for the player to see.
- Design problem: the "put the paper on the ground and find out there's a button on the goggles" puzzle would be impossible to replicate, seeing as HL2 has no inventory (thank god) and also it would be impossible to have a "rotate button" on the goggles because hey, you're wearing them! I will need an alternate method of "rotating" the view on the goggles. (Solution: don't know yet)
- Technical problem: when rotating the collision model for the museum, the player might be in the way. The Source engine will explode or clip the player or kill the player or whatever. (Solution: keep the collision model static, but rotate the visual model of the environment instead. It'll be more visually impressive too. Definitely more work on my end, but it shouldn't be too bad. However, this will place additional strain on what gets rendered because Source's PVS checks will just decide to draw everything in front of the player no matter what, so I'll had to have a really short draw distance for everything, which shouldn't be too bad since the whole level will be pitch black anyway.)
- Feasibility: very feasible, and best of all, no lighting required. I can re-use the Morning Star assets from Philip K to make a museum-y setting. I think I'll package it with Flatlander Woman for Volume 2.
- The only problem: now it's spoiled for you. Oh well.