$NAME(7) "$AUTHOR" "Thoughts on The Social Dilemma" The doco was better than I expected, but I would have liked it to go further in a couple of ways... One thing is that I find people have difficulty conceptualising how machine learning works when it comes to personal data. Human beings think in terms of any adversary being similar makeup to themselves, i.e. if your adversary is looking at somewhere else then they're not looking at you, or that your adversary predicts your actions based on understanding your motivation. The dramatisations played into this by anthropomorphising the machine learning process. It's how people end up saying stupid stuff like "I've got nothing to hide" --- because they're thinking in human terms of information that a person would consider meaningful, or worse, they think that a machine could only know things about them *after* they've conceptualised those things. I think a better image would be like an ocean of points, where every point is some information about a particular person. The machine learning process only looks for patterns within these points and makes exponentially more connections than humans would never even consider. It's how they're able to know things like when a couple will consider having children months before either of the couple has even thought about it, because it's just spotting similar patterns based on billions of data points. But it's worse than that, the machines are more than predictive; they're not just spotting one existing decision pathway and watching in unfold again with someone else, but they know the entire decision tree, with the decision pathways of millions of other couples, and so know how to manipulate the couple down one pathway or another, and how effective that manipulation will be. And people may think "but why would Facebook manipulate my decision to have children?" but they've already been doing these things for years just because they can[1]. That's the other thing I think the doco could have been better, is going deeper than the "villain" narrative, i.e. that these tech companies are these machiavellian entities destroying lives to make money. Yeah that's totally true, but it's not the heart of it. It's like that scene where the guy who invented the YouTube recommended video feature explains that he never intended it to be evil. It's pretty much the way you might block a beetle's path with your hand, then block it again, and again. You don't hate the beetle, you're just doing it because you can. That is what makes these tech companies so evil. Back in the 60s Hannah Arendt wrote EICHMANN IN JERUSALEM[2], a report on the trial of a Nazis who was instrumental in the extermination camps. She coined the term "the banality of evil" because the details of Eichmann's crimes basically came down to that he never thought about what he was doing, he didn't hate Jews, his evil was remarkably boring. -- [1] https://www.pnas.org/content/pnas/111/24/8788.full.pdf [2] https://www.newyorker.com/magazine/1963/02/16/eichmann-in-jerusalem-i