Dear Esther came out in February, but I don't have a Windows box (worth mentioning) so I skipped all the commentary and waited for the Mac port. That just showed up; huzzah! (Unironic cheer there. Three months is sharp porting. I'd love to see Fez three months from now, but I doesn't expect it.)
Because I skipped all the commentary, I won't try to do a full-on review. I'm sure it's mostly been said. Instead, you get assorted thoughts about interactivity.
Is Dear Esther a game? Sure it is. Two decades ago I was hedging and calling things like Gadget "interactive movies" rather than "games". But at this point, "semi-hallucinatory journey across a lonely landscape with background story" is an established game genre. It turns up regularly in the IFComp; I've committed it myself.
I might still hedge if the thing didn't come off as an interactive experience. But it does -- despite the absolute lack of any verbs besides "walk" and "look around". Dear Esther is the elusive zero-button game. (Okay, you can hold down a mouse button to zoom in a little, but this didn't add anything for me and gave a weirdly non-mimetic FOV-zoom effect, so I avoided it.)
So, given this interface, whence interactivity in Dear Esther? I say: from an understated but deadly-precise sense of attention design through spatial design.
You walk along the beach; a path goes up the bluff, another along the strand. You go one way or the other. There are no game-mechanics associated with the choice, and a plot-diagram analysis would call them "the same place" -- you can try either, back up, and go the other way. But this misses the point. Precisely because the game lacks keys, switches, stars, and 1ups, it has no implicit mandate to explore every inch of territory. Instead, you want to move forward. Backtracking is dull. Worse: given the game's sedate walking pace, it's slightly frustrating. (They left out the run button for a reason, see?) Moving into new territory is always the best-rewarded move, and therefore your choice of path is a choice. You will not (unless you thrash hard against the game's intentions) see everything in your first run-through.
Much of the game uses this pattern, and the choices are always distinctive. High or low, into a cave or down the beach, towards a building or up a slope. The paths join up again eventually, and you're always going to wind up in the next chapter, but this is de-emphasized; the game's sense is that you are exploring freely.
Smaller moments: you try to edge around a gap in the path, in order to continue up to the top of the hill. But the ledge is too narrow; you slip down to the beach and cannot climb back up. This is inevitable, it's just shaped terrain. But you have acted in the game and gotten a surprising result, and that's an interaction -- choice and outcome. (You could instead have turned around and descended the path, or scrambled deliberately down the slope.)
In a couple of places you advance into an area, explore it, find no way to continue, turn around, and see a second path by the entrance. You missed it on the way in because of the angle of view. (Or perhaps you saw it, but passed by because the area was more interesting. This pattern occurs at focal points, buildings or landmarks of detail -- not at mere dead ends.) Again, a map analysis would just call it a simple bend in the path, but the experience is of discovering a new choice and trying it.
None of this would work if the game were not beautiful, and of course it is beautiful. Dear Esther is the bar for next-generation immersive environmental gaming. Comparisons to Myst are easy (the Myst series set and reset that bar how many times? Four generations running?) but I would rather call survival horror onto the stage. Dear Esther draws heavily on the techniques of horror: thickly textured environments, dim light, distant glimpses, slow introduction of off-kilter elements. (If I relabelled its screenshots as from an unknown Silent Hill game, just before the blood and flayed nurses drop in, you would have no reason to doubt me.)
The aim is not a sense of dread, in this case, but the pace of survival horror: slow movement and careful focus on the details of the world. And it is the detail, the reward of each step and the greater reward of turning a corner, that allows the rest of the game elements to work. The spoken narrative is well-written, but it can't stand on its own. Its value is as an occasional addition to the visual world. It's the visual detail that you focus on; it's the beauty that draws you forward and keeps that sense of interactivity alight.
(Speaking as a text-game author, may I say how much I envy the sensory bimodality of the graphical genres. I too like to intermingle background, scenery, and story narration -- but I use words and words come in rows. Moving through a world while listening to prose is a wonderful experience.) (I know, it's been done forever now, but this post is my opportunity to envy it.)
(Although, may I also say, the bimodality in Dear Esther is pretty much ruined by the subtitles. For zog's sake, unless you have some hearing impairment, turn off the subtitles and play with a good sound system.)
The lesson, I suppose, is to be generous where your game design needs it (visual richness, in this case) and spare everywhere else. Dear Esther is admirably minimal in its storytelling -- a few motifs repeated in the narration, then incarnated in the world, and you make of them what you wish. The sense of surreality, so often inflated to overwhelming in those survival horror games, is here limited to a single hallucinatory scene -- plus the peripheral, negative-space impossibility of following a trail blazed by your own narrated self.
And, to be sure, the game itself is short. All hail the restraint of giving the player just enough of every locale, with a few vivid variations, and then moving on before the effect palls. (You may not roll your eyes at Portal 2 -- I approved, myself -- but we don't lack for other examples of games that pad out a good design sense until it's two and a half times ruined.)
You may wish to replay Dear Esther. I'm not sure I do. I re-ran chapter 3, out of game-designer's curiosity and the desire to write a sound blog post, but the game didn't call for it. This leaves the game's randomized text ordering rather in the lurch; it's interesting, but not something a player will notice in a single run-through, so it doesn't exactly fit into my account of the design. (Which is why it doesn't fit into this blog post either.) Perhaps the right approach is to play Dear Esther, let the experience settle, and then watch a friend play it. I'll let you know if I try.