People are predictable in the mass quantity that you want to demonstrate.
Because you start noticing trends, and patterns, it doesn't mean everyone works in the exact same fashion
You are assuming everyone believes their actions are random, which most know their actions aren't all random. They may do a few "random" actions, but I don't know anyone who thinks most of their actions are random. Or who attempt such. So yes, if target is monitoring your purchasing habits and other data very closely, they may realize something about you that you didn't know.
If you however are closely monitoring yourself and trying to control your behavior in certain fashion its very likely Target couldn't find out you were pregnant.
The benfords law isn't great demonstration of this either, I read article, I am an accountant, never heard of it being used in a audit. Looked a little deeper found out its been used as a analytic tool a few times at least. Sometimes when used they found the company had weird expenses which caused the anomaly in the "law" but found these tied to legitimate expenses. It also found that if your company has unusual or irregular events this will also throw off that law. So, may be slightly useful along with other analytic for predetermining a possible fraud, but its definitely not great. Not useless. Don't know how this applies to "humans can't be random" kind of goes with the universe isn't random.
The random number generator, I think I heard once about this, but I am wondering the pool. Just because X is the most popular choice, doesn't mean its the only one chosen. How big of a pool and how widely was it chosen?. Like I said, outliers and other things. You can find patterns in behavior, but that doesn't mean when people genuinely try to play outside the box it won't.
I mean if someone told you, 17 is the most popular number picked between 1-20. Now can you pick a random number right now, that was picked the least! That gets you to think in a different way, and probably makes it harder for a computer to predict. (maybe not over a huge sample of people, but harder than the one with vague information). I mean relating this to mafia, say people know the common strategy is X, they will then thing to do maybe Y, Z or even W instead.
The rock-paper-scissors one is good, I have heard of it. Except if your opponent has heard of it and is aware you are attemping it, it won't work. This is part of the reason why information like this breaks the cycle. Once people know the trick, it starts becoming less of a good method.
The 93% of human behavior is predictable, fine. I think you translating this horribly. We are talking about a game of mafia. Were you are supposed to lie and figure out who is lying. This means that some will attempt to random actions, because it has been shown in game theory to be greatly viable strategy (especially against a better opponent). Yes, on a large scale everyone is probably predictable. In a game where your goal is to be not predictable, yeah I think it can be done.
I don't agree that players are extremely predictable. Trends form over time, but there will always be outliers. Something that won't fit into a model of complete predictability, I mean whats to stop someone from preplanning actions out using a chart like :
I Scenario A do one of the following: 1,2,3,4.
If Scenario B do one of the following: 1,2,3,4
Then using an actual random number generator to determine your action.
How random would you call this?
