**** Cyborg Central ****

The source of Twitter's utility is its advertising algorithm, its ability to put words in front of people who are interested in things - but not interested enough to actively seek those things out, and who might not even know what to search for. It's easy to throw a few search terms into Google, but what if you don't even know you're interested in a topic? How can an advertiser (or a propagandist) find good targets, when those targets might not even know what they want themselves?

Twitter does this by spying on you. It works by reading not just what you post, but which links you click on, what sites you visit, tracking what you buy, interpreting the sentiment of your posts, and feeding it into machine learning that matches you up with messages you'll likely be interested in. Twitter also benefits from massive amounts of advertisement tracking purchased from other websites, and is quite good at correlating supposedly "anonymous" accounts with reality.

The reason Mastodon is "boring" or "less useful" is because the users of centralized sites have become accustomed to being fed information, and don't have the motivation to seek things out on their own, or the knowledge of how to seek these things in the first place. This is how social media becomes excellent propaganda - lull the users into a sense of trust, then start feeding them the ideas you want them to have. The propagandists can even track how well you've assimilated their ideas, using that same algorithm.

I wonder: how do we protect privacy while providing the same sense of convenience and trivialization of discovery? How can distributed services compete with centralized indexes, built off of massive tracking databases?

Perhaps this too is something a tool like stable-diffusion can lend itself to - as a user controlled AI, guiding you to information you didn't know you wanted. An AI for this purpose would need to be trained passively, and the resulting weights for a specific person's interests would be a very valuable target, so this data would need to be very carefully protected. If at some point you stopped trusting it to recommend things, you could reset it and start over - but there is a risk, of course, that as you are attacked with propaganda and misinformation, the AI also would become corrupt and make the situation worse.

Share your thoughts