A Sociable Medium (Crosspost from Crunchysteve.dreamwidth.com)

A friend sent me a link to a Naomi Kritzer short story on Clarkesworld, today - Better Living Through Algorithms." It's a quite true-to-life and amusing tale about a group of friends and a new app. Not my usual style of sci-fi preference, but it had me hooked from the start. There was something in the tone that hooked me before my too-cool-for-school 'tude could get in the way, and I'm really pleased that happened.

Look, read it for yourself but the #tldr is this:- Linnea, June and Margo are friends having lunch and June explains her new social media app, Abelique, that "has changed her life." She's happier than she's ever been and feeling fulfilled, achieving day-to-day things easily and efficiently, thanks to the app.

Of course, Linnea and Margot are skeptical. Margot dismisses it with, "It sounds like a cult." And it actually does a bit. Linnea returns to the office after lunch and her boss insists she and her colleagues use the app because it will improve everybody's productivity. Her boss is one of those low-grade passive aggressives that would see her as less productive if she doesn't adopt the app, so she's forced to try it, even though she sides more with Margot than June on the cult thing.

Linnea discovers the app is a combination of encouraging, gently maternal, time management tools and a "phone tree" where users are assigned to call other users to add a personal connection element to the app's guidance. It is kind of "like a cult" except there's "...no belief structure," and it is just simply positive reinforcement of stuff that the app has determined the user likes to do. For instance, it makes clothing recommendations each morning, suggests meal plans, assigns users to make early morning wakeup calls and gives them scripts for how to be the right kind of friendly for the person they're assigned to call. It rounds up payments, putting the change into a kitty that is then given back to users as savings. Linnea is encouraged to take up drawing, an art she hasn't done since she was 10, she is assigned to a group of like artists and they meet on a forum daily, gathering weekly. It's quite the utopia.

As the story unfolds, Margo, a journalist, uncovers the truth behind the app, it's a research project: can AI be used make humans feel better? Read the story for the details, but the app shuts down up and the humans continue to try to maintain the connections the app helped them maintain in the first place. I want this app!

In fact, if you ever used OK Cupid in the days when it was still a research project, you know it was more than just a dating site. The original version of the site was a "datamine" for studying human emotional interactions in order to understand how we formed our best friendships and relationships. It was incredibly idealistic and uncapitalistic, a project where human understanding was the goal. The data was mined from our interactions in order to understand how we do, or don't, find our true friends and partners. It dripped fun and positivity. The quizzes had a cheeky vibe, loosely based on Myers-Briggs types, but not taking that now-debunked idea seriously, just that its popularity made it an easy tool to randomise our interactions.

Reading that short story about the fictional Abelique app made me think of OK Cupid almost immediately. Not for the friends-and-or-finding-of-friends, but the general positivity of living in a supportive community of encouragement.

Abelique, like OK cupid, would need to be run by a life positive group, a free-open-source collective or an academic community, genuinely seeking to better human existence. Capitalism would ruin it, like the ultimate sale of OK Cupid turned it into just another boring dating site. Abelique, as Ms Kritzer's depiction of it, would be a paid app, but for the purposes of helping its users be better, happier people, not to harvest data about how to push our buttons like "The Zuck" does to make a buck. The monthly subscription would need to be used to run the service and reward its users. The most important of those rewards would be the intrinsic community positivity, the uplift and acceptance it might provide, a place of safety... just like OK Cupid once was... well, mostly was.

Now, contrast this with what social media currently is. (And what Abelique becomes at the end of the story, no spoilers, you will see that coming.) X banning "leftists" yet claiming to support "absolute free speech" or Meta witholding our friends' posts until we've viewed enough ads, it being maybe 3 days, in some cases, before we see some posts, with the hope for Meta being that this might cause a fight between friends, to mine that data. Meanwhile, Meta also has threads, a Mastodon fork/instance that could be argued to be illegally harvesting reaction data from people who have never authorised Meta to use their data. Read the wikipedia article about the activitypub protocol that supports Mastodon.

The world needs an app like Abelique and we need massive legal strictures on apps like Facebook and X. The social dishesion, that these latter platforms trade off, is as possible a mortal threat to human existance as global warming, and I don't belittle global warming in that statement, either.

And on the subject of global warming, with AI being a major contribution to greenhouse gas emmissions, an Abelique app wouldn't actually need to be AI based. Even if it were, the AI power required for the kind of collating and sorting it would need to be able to do could be easily run on devices, not servers. AI like Ollama running an 4B language model. It's making choices based on local user preferences and their local to local interactions.

In reality, this is very low grade processor demand. My M1 Mac doesn't even hiccup when running massively large AI requests offline. My iPhone could probably run a local AI with enough grunt. All an Abelique app actually needs to do is run a peer-to-database, hosting the local connections of each device and make occasional survey requests on recent interactions, seeking feedback on interactions, one or two survey questions at a time, to a maximum of 5 times a day. The phone tree part already exists in a number of apps.

You can tell I want this app, right? Need! Can you also tell that the world this app would create is person-to-person rather than client to service. That is the point of Ms. Kritzer's story, OK. You're still with me on this, right?

Real world systems are imperfect. They leak. However, good design uses "leakage" to feed the garden, rather than undermine the foundations. Societies have never existed in cycberspace before the last few decades. We don't know how to design societies for cybercommunity yet. If you combine locally centered, lower level AI with peer-to-peer connectivity and a distributed personal domain name system, plus an automated phone tree for a peer-to-peer "call center" and paypal... That's Abelique.

Get coding peeps, but get your governance right. If you want to create community happiness, you need to think about dedication, not profit. You stain this beautiful thing with the profit motive and it will die. Abelique has to be built on sustainability. Abelique has to rely on interpersonal consent rather than a "shrinkwrap end-user licence agreement." Abelique will die if greed is any maintainer's motive. Pure of heart, pure of deed. Make a wage, make a community, make people happy, but the only profit you can make doing it is the wages for you and your team, or the golden goose will die. Profit capitalistically and you will be the one who kills it.

Now, unleash the nerds to code this thing.