Study: Social media probably can't be fixed

> They then tested six different intervention strategies...

None of these approaches offer what I want, and what I think a lot of people want, which is a social network primarily of people you know and give at least one shit about. But in reality, most of us don't have extended social networks that can provide enough content to consistently entertain us. So, even if we don't want 'outside' content (as if that was an option), we'll gravitate to it out of boredom and our feeds will gradually morph back into some version of the clusrterfucks we all deal with today.

My tech (I was founding engineer and CTO) company took over a co-working space and expanded it, we ran that portion of it at break even.

We intentionally set out to create a social club/co-working space. A lot goes into it. I'm a non-theist who comes from a multi generational group of theist church planters (like 100s of churches, just over and over), it's a multi factorial process with distinct transitions in space-time and community size, where each transition has to be handled so you don't alienate your communities founding members (who will be very different from later members) and still are able to grow.

People don't do it because they can't see the value while they are in the early mess of it. You have to like people to pull it off, you have to NOT be a high control person who can operate at high control at certain developmental stages. You have to have a moral compass everyone understands and you are consistent with, tech people like 0 trust. You have to create a maximum trust environment which means NOT extracting value from the community but understanding that the value is intrinsic in the community.

You have to design a space to facilitate work and play. It's not hard but you have to get everything right, community can't have mono culture and it must be enjoyable/uncomfortable, and you must design things so people can choose their level of engagement and grow into the community. It's easier once it has enough intertia that they understand they are building a thing with real benefits.

Even things like the flow of foot traffic within the space, chokepoints narrowing, these kinds of thing all effect how people interact.

It doesn’t surprise me if they found that the emergent behaviors didn’t change given their method. Modifying the simulation to make them behave differently would mean your rules have changed the model’s behavior to “jump tracks” into simulating a different sort of person who would generate different outputs. It’s not quite analogous to having the same Bob who likes fishing responding to different stimuli. Sort of like how Elon told Grok to be “unfathomably based” and stop caring about being PC” and suddenly it turned into a Neo-Nazi Chan-troll. Changing the inputs for an LLM isn’t taking a core identity and tweaking it, it’s completely altering the relationships between all the tokens it’s working with.

I would assume there is so much in the corpus based on behavior optimized for the actual existing social media we have that the behavior of the bots is not going to change because the bot isn’t responding to incentives like a person would it’s mimicking the behavior it’s been trained on and if there isn’t enough training data of behavior under the different inputs you’re trying to test you’re not actually applying the “treatment” you would think you are.

>I work in survey research and I'm rather appalled at how many people would rather survey a sample of AIs than a sample of people and claim they can come to some valid conclusion as a result.

A YC company just launched doing exactly that.

https://news.ycombinator.com/item?id=44755654