I love where he says it wasn't a public campaign where one message was sent out and everyone could judge it, but rather it was designed to whisper the right message privately to each individual, the message most effective at converting their opinion given their psychological and demographic profile, even if the messages themselves were contradictory across different individuals.
Hard to figure out what the future holds because we generally never get less sophisticated, but more sophisticated.
When Netflix can't even recommend movies to me that are more than 50% likely to be enjoyable for me, I really have to wonder how effective messages designed to change people's political opinions based on demographic and psychological profiles really were.
They don't really have to change people's political opinions. But they can know what makes people angry and give them more of it. They can know the lie they want to tell and make it more ubiquitous so it's harder to discern from truth. These tasks aren't anywhere near as abstract as predicting your taste in film. It seems like they knew angry people were showing up to vote for Trump, so just give people more of what makes them angry. And give people thinking about voting for Hillary more reasons to be skeptical about her weaknesses as a candidate.
Most voters in america are split on binary issues like gay rights, abortion, gun rights, etc. It's much easier to find what you're passionate about in this case, especially since location and party affiliation can give you a great prior.
Then there's A/B testing. A lot of people claim that it would be hard to sway an election on a million dollar budget but Facebook has made it incredibly easy to test what messages resonate with voters. For a few hundred bucks you can test your hypothesis and promote it to the few people who will help propagate the winning message.
Refresh your lists of logical fallacies a.k.a. vulnerability lists and make every effort to patch.