I doubt it's the solution, but a silly program I want to build is something like this:
- Give users a modern Tamagotchi
- Give the digital pet a need to socialize.
- Strap a basic LLM to it so users can talk to their pet.
- Have the pet imprint on its owner through repeated socialization.
- Owner goes to bed, pet still has social needs, goes out into the digital world to find other pets.
- Pet talks to other pets while you're asleep, evaluates interactions, befriend those with good interactions.
- Owner wakes up the next morning, checks their pet, learns it befriended other pets based on shared interests, and is given an opportunity to connect with their pet's friends' owners. Ideally these connections have a better-than-random chance of succeeding since you're matched via shared interests.
I'm sure there's a ton of unsexy technical reasons this is hard to make work well in practice... but dang, I think it would be so cool if it worked well.
I realize this exacerbates the issue in some ways - promoting online-first interactions. But, I dunno. I'll take what I can get these days, lol.
A thought-provoking vision. There seem many many underexplored opportunities for catalyzing social connections - match making.
When I'm doing "broad and shallow" at a meetup, there are invariably "oh, you'll want to talk with X over there, they <overlap on some intent interest>". It can feel tragic to look out at a room of nifty people, in largely desultory conversations, knowing that there are some highly-valued conversations latent there, which won't occur, because our social and cultural and technical collaborative infrastructure still sucks so badly at all this.
In lectures augmented by peer-instruction, addressing the "if you think your lectures are working, your assessment also isn't" problem, one version has students clicker-committing to a question answer, then turning to discuss it with a neighbor, then clickering again. One variant (which you now many not be able to use commercially, because of the failed-startup-to-bigco patent pipeline), has the system chose who everyone turns to (your phone tells you "discuss with the person behind left"), attempting to maximize discussion fruitfulness, using its insight into who is confused about what. So perhaps imagine Tamagotchis as part social liaison - "hey, did you know the gal at the optometry shop here also enjoys heavy bluewater sailing?"
So on the topic question: Want to incentivize greater social contact...? Increase the payoffs.
In my mind, it's more like meeting new acquaintances at the dog park. Dogs start playing with each other and getting along and you end up chatting with the other dog's owner while watching the dogs play together. Trying to recreate those vibes with digital pets.
- Give users a modern Tamagotchi
- Give the digital pet a need to socialize.
- Strap a basic LLM to it so users can talk to their pet.
- Have the pet imprint on its owner through repeated socialization.
- Owner goes to bed, pet still has social needs, goes out into the digital world to find other pets.
- Pet talks to other pets while you're asleep, evaluates interactions, befriend those with good interactions.
- Owner wakes up the next morning, checks their pet, learns it befriended other pets based on shared interests, and is given an opportunity to connect with their pet's friends' owners. Ideally these connections have a better-than-random chance of succeeding since you're matched via shared interests.
I'm sure there's a ton of unsexy technical reasons this is hard to make work well in practice... but dang, I think it would be so cool if it worked well.
I realize this exacerbates the issue in some ways - promoting online-first interactions. But, I dunno. I'll take what I can get these days, lol.