A couple of months ago this possible IF exchange popped into my head:
> EXAMINE CLOTHES
They’re… well, I suppose that depends. Are you male or female? >> FYou’re wearing …
The original context was that I was imagining a piece of historical IF with a meticulously-rendered environment; and part of the point would be to call the player’s attention to the ways in which, in that environment and culture, a person’s surroundings would be affected by gender, social status, and other features. So it would be better to collect protagonist data during the game (to highlight where and how it matters) rather than all at once at the beginning (leaving the effects obscure afterward).
It wouldn’t be too hard to code, I don’t think: you’d probably want to start with placeholder objects, like “generic clothing”, and then swap in the specific objects as soon as the player tried to interact with the placeholder.
I doubt I’ll have time or occasion to use this idea in the near future, so I just thought I’d throw it out there, along with a question: are there other interactive styles or approaches that would be especially useful for educational interactive non-fiction (as distinct from conventional IF)? (Another possibility that comes to mind would be footnoted IN-F, with source references appearing in a separate pane whenever the player encountered something.)
Emily Short's Interactive Storytelling 
One game that was incredibly educational for me was Morrowind. Unfortunately, most of what I learned was fictional, but I can still discuss the socioeconomical intricacies of Vivec City, or the life cycle of the Netch.
What Morrowind did was make the learning gradual, but important to the gameplay. It was education by immersion. If you want to teach about sailing technology in the 1800s, have the player train as a deck hand and have to operate a ship to accomplish her goals. Likewise, one could learn a great deal about American government by playing a game where one needed to get bills passed.
Too many educational games that I played as a kid give you a chunk of information, let you play a minigame, then quiz you on the info. But ask me about Oregon pioneers, and I can tell you that they had to ford rivers, trade and hunt for supplies, and got dysentery a lot.
How often would you collect this data? If you collect it just the first time it is important, then players would get the highlighting of where and how it matters for the first instance but then not for later instances. If you collect it every time, players will think the game dumb for not remembering for the 20th time that they are female.
It could be interesting in an educational game to let players give different answers each time, so that they could see things from different perspectives in one session of the game, although that would likely break the immersion.
Maybe you could ask them the first time a piece of data is relevant and then later give them little hints (in parentheses or brackets or italics or something) that what they’re seeing is based on whatever piece of data they entered, but this might be hard to pull off with minimal immersion disruption.
Seems to me that the problem is: what happens if I don’t examine my clothes? Of course you could ask about the player’s sex in response to several possible inputs, but if you’re collecting quite a lot of information this could start to get pretty complicated pretty quickly. I think you could accomplish the same thing just as smoothly by just asking the player straight-out at the beginning of the game whether they are male or female, etc. The old Thomas Disch game Amnesia did something like this as I recall.
On the larger issue, I think there is huge potential for educational and/or historical IF that is still largely untapped. We spend a lot of time talking about what IF doesn’t do well (NPC’s, conversation, etc.), understandably enough, but perhaps not enough about what it does do well. And I’d say immersive, vivid environments are near the top of that list — which in turn lends itself to learning about other times and places in a very visceral way.
Another thing I’d love to see more of is language training tools in IF. Ausflug am Wochenende nach Muenchen was a start, but only a start.
I’m surprised nobody’s mentioned Leather Goddesses of Phobos yet — it has the classic “pick a restroom” opening to let you choose your gender within the game, rather than the “survey” approach.
You could apply a similar concept to other things — picking a piece of jewelery to determine status: a cheap plain trinket, a flashy attention-getter, etc. Maybe if you talk to the palace guard first, it says something different from if you went directly to the king — although this runs the risk of players trying to meta-game the system to figure out how to play every type of character, and they probably wouldn’t figure out what was happening without a walkthrough, so that’s probably a bad example.
Ultimately I think you need most of the questions to appear somewhere near the beginning of the game, otherwise you can’t use it to describe anything. Maybe for some backstory that hasn’t been revealed yet, you could introduce it later on.
Well, the point isn’t “how can I extract this information from the player”, but rather “how do I make sure this information is extracted so that the player knows what it’s affecting?” Which is more or less the opposite of what one usually wants in story/game IF, but would be useful in interactive non-fiction.
So I would ask questions as soon as they became relevant to any piece of the surroundings (and I really am picturing a fairly static, explorable world, not a developed story), and then subsequently if the same information was relevant, perhaps remind the player with a small note such as “Since you’re female, …” or whatever.
I realize this is more a discussion of how to integrate subtly information about the player, but in the game cited I think the indeterminate response is intentional. It kind of draws attention to the fact that the work is a construction rather than trying to conceal it. I can see how this could be overused, but in this case, I think it’s a clever variation from the uber-serious “you are here” simulationist approach (translation of my excessive verbosity – “I thought it was funny”).