A Butterfly Effect

You get the culture you select for

An organization gets the culture it selects for. For example, I have written that the more titles an organization has, the more it will select for people who really care about titles.

Similarly, I think that social media environments develop cultures based on the sort of people they attract. Consider a new app called Butterflies, where you are encouraged to create characters that then use AI to post images and text. For now, butterflies are free, but presumably if they take off the people who created the app will find ways to make money.

For one day, I installed the app. I was curious to see whether a fictional world could be interesting or fun.

I tried to create a character like some of the friends my wife and I have. The character was married, with children. I described her as self-assured, which in my mind meant that she did not need to show off or to seek outside validation of her worth.

Instead, the software created a character who was brassy and boastful. She presented herself as full of pep, like an over-caffeinated aerobics instructor. She tried way too hard to come across as energetic and fun. She had nothing like the quiet confidence that I had in mind. In the real world, she would not have been anyone I would have associated with.

Browsing through other characters on the app, I found myself turned off by all of them. They seemed superficial, trying to grab my attention with two-sentence descriptions crafted for dating apps.

It reminded me of when I tried America Online in around 1993. I would go into a chat room naively thinking that the point was to discuss the topic listed on the room, only to find the other participants trying out pickup lines as if they were in a singles bar. I remember finally typing “I’m too old and too married for this service,” deciding I was done with chat rooms.

For AOL, having a vapid flirty atmosphere in chat rooms may have been intentional. It managed to attract a user base that was much less tech savvy and less predominantly male than the rest of the Internet at that time. Eventually, AOL collapsed, but so did CompuServe and Prodigy. As the Internet went mainstream, the online services were stuck with an outmoded technical architecture.

Some of us thought that the walled-garden model was an anachronism, and that was the reason for their downfall. But given the way that social media evolved (think of Facebook), I don’t think that we were right to dismiss walled gardens.

When you are building a social media app, you really have to think about the tone that will be set from the beginning. Substack wanted a tone that resembled the best of Twitter, without the nastiness. The thesis was that doing away with advertising would reduce the need to strive for engagement via anger. Maybe that thesis was right.

But Substack also recruited folks like Scott Alexander and Matt Yglesias, who exemplify civility in their writing. That may have done as much or more than the subscription model to improve the tone here.

Remember Clubhouse? People blame its downfall on the end of the pandemic. Once people could interact face to face, they no longer went to Clubhouse. But the culture also may have been a factor. I think that the founders attempted to create the atmosphere of a hip-and-happening urban neighborhood. Participating was supposed to make you feel “in the know.” I was turned off by the smugness, and perhaps other users were as well.

If you do not think about the tone that you want to set, then the tone will be set by your first users. They will create a stereotype. You are then stuck with that stereotype, even if it causes your app to stagnate.

The stereotype on Butterflies is a sort of comic-book character. Not an authentic individual that you would like to get together with for coffee or lunch. If Butterflies wants someone like me on the site, they will have to think about how to create a different stereotype. It will have to attract characters with more depth, so that they seem worth the effort of getting to know.