The article advocates "building a game that's meant to be played rather than just reacted to." That sounds right, but, then, it's sufficiently general that I'm not sure who would disagree with it.
The last paragraph of the article seems to equate "a game that's meant to be played" with "real sandbox simulation, autonomous agents and language-capable AI", and that seems like a narrow idea of what "playing" means, one which equates interactivity (which is the distinguishing feature of games) with choice or nonlinearity. Providing players with lasting choices is one way in which you can use interactivity to structure an experience, but it's not the only one. There's some interesting comments on this in a review by Emily Short of the IF game "Howling Dogs": http://emshort.wordpress.com/2012/10/10/if-comp-2012-howling...
"Howling Dogs" is something of a masterclass in the different ways games can use interaction, and I'd recommend people check it out as a supplement to the vision of gaming put forward in this article.
Your analysis of the last paragraph of the article seems pretty spot on. Kudos!
But I'm nitpicky; the distinguishing feature of games isn't interactivity, otherwise every fun activity with interactivity would be a game, and I'm not sure conversations are normally included when we speak of "games" in this way. Games have rules, and often some kind of goal, and other things. The Art of Game Design (a book by Jesse Schell) kind of roughly approximates it by describing games as a kind of activity involving problem-solving and fun. If I'm not taking actions specifically chosen towards achieving a result ("problem-solving"), then I don't feel like I'm playing a game, even if there's some interactivity.
So I'd say that the key towards quality game experiences lies more in the region of providing options to the player (actions they can take) that have an effect on the game world (resulting conversation, successful quicktime scene, dead goblin, shiny new sword, whatever!) where which option is taken lands the player in a different distance from various desirable goals ("solution" being what the player did to get there).
But there's still a lot of fuzzyness, and it's still very hard to even judge what parts of what games are quality experiences for who and when, and especially why. You have it completely right that we can't just say it's "real sandbox simulation, autonomous agents and language-capable AI", and that lasting choices aren't the One True Way.
The last paragraph of the article seems to equate "a game that's meant to be played" with "real sandbox simulation, autonomous agents and language-capable AI", and that seems like a narrow idea of what "playing" means, one which equates interactivity (which is the distinguishing feature of games) with choice or nonlinearity. Providing players with lasting choices is one way in which you can use interactivity to structure an experience, but it's not the only one. There's some interesting comments on this in a review by Emily Short of the IF game "Howling Dogs": http://emshort.wordpress.com/2012/10/10/if-comp-2012-howling...
"Howling Dogs" is something of a masterclass in the different ways games can use interaction, and I'd recommend people check it out as a supplement to the vision of gaming put forward in this article.