How do I limit the characters' available story information? [a.k.a. "How did you read my mind?"]

[g1hanna]

Hey to anyone who looks at the GitHub for this project! I've been trying out Inner Self on an AI Dungeon scenario for a little bit, and I've got both a question and some examples to share, which may spark some discussion.

First of all, huge thank you to @LewdLeah for creating this masterpiece of a tool! When talking and interacting with the characters in this example, I found that this script added a good bit of depth to the story, and I was hooked by the reactions I got from the characters. I look forward to integrating Inner Self into a future scenario I'm putting together.

That said, I've noticed some funny results that someone could shed some light on. This isn't necessarily a bug and has even caused some intriguing (but sometimes realistic) events in the story. It could just be an odd quirk with AID, too. This also isn't a feature request, but it could spark some ideas for @LewdLeah's next iteration.

Main Question

After playing around with this for a while, it seems like Inner Self characters will pick up on whatever happens in the story, not just actions you do or comments you make when they are in the same space as you, but almost everything that happens in the story. My main question is...

• Is there a way to limit the information that certain I.S.-enabled actors have access to? (Mostly due to a character's unlikelihood of being in the same scene or having access to what the player is thinking — not noticeably doing or saying.)
  • Ideally, this would be based on the character's likelihood of being in the scene with the player or witnessing particular actions (e.g. sleight of hand checks if you are a rogue in a fantasy game).

Examples

Here are some examples of how I've seen this manifest in a scenario with this script (using the Unexpected Girlfriend scenario as an example [don't laugh at me, please]):

• Thoughts the player thinks internally

  • Example 1: My character [Gage] has a girlfriend [Hana]. Gage asks her out, she says yes, and the relationship is off to a good start, but Hana warns him about her dad being "intensely defensive." (Typical high school drama.) After school, Gage is alone in his house. Deciding to do some internal monologuing, I have Gage do this...

    You think to yourself, 'Why would Hana's dad make War and Peace a qualifier for potential suitors? He can't just have any reason for scaring guys off like that.'

  • When looking at Hana's brain card (under Story Cards), her next thought is:

    Gage's thought about War and Peace shows a quiet self-awareness that I find charming, like he's always thinking about what matters, even in the smallest moments.

  • Reaction: Pretty, but I was a little surprised to find this in the story card when it first cropped up, and Gage would probably be a bit creeped out if he knew too (lol). Funny that this person just happens to be able to read his mind. (Or the script. Though, most players probably don't do this at all.)

• Output from any Do or Say actions done outside of an I.S. character's vicinity

  • Example 2: Player sends a text to his best friend, who felt down earlier. Your character [Gage] goes on a walk, then AI dungeon writes...

    You observe a quiet valley on a cold night. It carries a feeling of serenity.

  • This triggers a thought in Gage's best friend:

    I wonder if I'll ever stop craving places like this, where the quiet feels like a secret only I know. It's dangerous, this need to hide, but for now, it's all I have.

    • One press of 'Continue' later, and there's the best friend in the valley. He says hi to Gage, still being somewhat guarded about what's got him down, but they have a heartfelt friendship moment.
  • Reaction: This was also surprising, but in a less game-breaking way. It's like AI Dungeon intentionally decided to put Gage's best friend in the scene to reveal a side of him we haven't seen yet.
  • Example 3: Sometimes, this can make it look like other characters are eavesdropping on conversations the player has. In this example, Gage and his girlfriend [Hana] meet in a library. Earlier in the story, it was revealed that Hana's best friend [Vanessa] sometimes works in the library. [Inner Self manages both Vanessa and Hana in the story.] As Gage and Hana are talking at a table, Hana fidgets with one of her books.
  • Vanessa thinks:

    I notice Hana is trying to be subtle about how she feels about her friends, but the way she fidgets with the book shows she's hiding something.

  • Reaction: Interesting, and it makes sense. I already knew Vanessa tended to protect Hana from getting hurt, so it's accurate that she would be watching her from a desk or somewhere to make sure her friend is okay.

Idea for Example 1: Thought-Dump Card (or Idle Brain Card) for Player

Edit (1/24): Thought of this idea after posting, but what if Inner Self had a Story card (similar to other brain cards) to record the player's thoughts whenever performing a 'think' action (could add synonyms like wonder, dream about, etc.)? If the player's "Brain" is active, then Inner Self could simply copy-paste these thoughts into the Idle Brain's Notes section, making the player the I.S. "Active Character" without generating reflective thoughts for the player. This could help prevent other characters from reading the player's thoughts.
...Or an easier approach would be to have Inner Self not generate thoughts when the player performs a Think action.

Again, don't want to sound like I'm nitpicking too much here. Using the script made for much richer characters than in normal scenarios, and the results have been better than expected!

Looking forward to whatever tips or thoughts you guys have to add, or maybe some of you just found this entertaining to read. Anyway, thanks!