Society is what Deleuze and Guattari would describe as a striated space, and social networks can’t live there.
A striated space is heterogeneous: to move a person or a message across it you have to continuously modify and adjust them, with every interaction being highly contextual, and every long-distance movement — as measured in social distance — requiring a multitude of very different steps (for a similar concept, consider Latour’s ANT). But a social network needs users to spend much time there so they can sell advertising, more time than your immediate social group can provide “content” for. And influencing, that other form of advertising, also requires their content to reach, unmodified, as many users as possible.
Both requirements come from the business model, yet they force social networks to present a smooth social space. This is an homogeneous space where there’s usually one form of connection, where opaque global rules determine who is shown what, where the unimpeded flow of unmodified messages is configured to maximize how much is shown to how many. It’s not surprising that Twitter or Facebook feel so emotionally damaging, or that they can have such negative social impacts: they are close to anti-societies, their topology designed to make everybody vulnerable to the periodic raids of bad-faith agents and relentless bland marketing.
Two takeaways from this. The first is that truly social software is by definition software for the handling of configurable heterogeneity. Software that doesn’t understand that I want to hear different things at different times in different ways from a coworker and my mom, and that my relation with my coworkers is different than yours, comes from an uncanny valley idea of society.
The second takeaway is that there’s no such thing as advertising/influencing/thought-leadership-driven social networks. They are media, mass media if they are successful. Apologies to Roberto Carlos, but nobody has a million friends.