I don't know exactly what it would cost to take over a third world country or topple the government of a first world country if someone was interested in doing precisely that, but my basic calculations and guestimations based on pursuing the most obvious lines of efficient attack suggest that the total cost should be well under a billion dollars. More importantly, we have some pretty compelling evidence to suggest that someone who wanted to do this could do it more efficiently and inexpensively if he was willing to be a ruthless, draconian despot than if he wanted to do it through more democratic methods. Putin's methods of control are very effective. Stalin made use of even stronger ruthlessness, so did Mao. The failed authoritarian regimes of the twentieth century tended to have a mixture of arbitrary enemies and political enemies that have been treated extremely poorly. Genocides have ended badly for their perpetrators on average, but for a genocide to take place, an authoritarian government has to use a considerable amount of its brute strength that it could be using to suppress political enemies and instead waste that energy killing and/or tormenting people for happening to be born the wrong race or for being raised in the wrong religion, or whatever. If you can get away with three hundred thousand murders (or whatever the actual limit is) before the international community and/or your own people start to decide that you are a legitimate threat to the world that must be eliminated, you are much better off using up your quota to squash your opposition than you are using up your quota trying to create racial purity or some other such nonsense. For anyone who can afford to do it, a basic strategy that ought to work for taking over third world countries is to buy some followers, equip them to fight your enemies, and have zero tolerance for dissent or resistance while avoiding war crimes other than ones that specifically have to do with destroying your enemies and eliminating resistance. It seems to me that tens of thousands of people or more have the ability to successfully implement this sort of conquest campaign, but to the best of my knowledge nobody's trying it. The world has a handful of rulers that use some of these policies to maintain control, but almost all of them worked their way up through an existing political or military institution in their country, and almost all of them were dealing with a country that already had similar policies in place. Frequently, these rulers are slightly less despotic than their predecessors. Nobody's used their existing success to secure extreme despotic power, and to the best of my knowledge nobody's even made the attempt. Every once in a while, we do see extremely militant draconian regimes rise up in this world, but to the best of my knowledge, they are all peasant regimes. Boko Haram and ISIS are rabble. Hitler was a convicted criminal who appealed to impoverished Germans. Mao was a teacher who had been disowned by his family who built up his popularity among peasants. These are examples of people who have successfully built a revolutionary movement that secured power by dealing harshly with its opponents, but ridiculously many un-noteworthy examples of failed movements formed by people in similar circumstances can also be thought of.
I've been thinking about this topic for quite a while, but I still struggle a little bit to piece it all together. But I think that a pattern we see in these sorts of revolution is a pattern that we also see many other places. What we call "evil" is often a failed attempt to achieve what we call "magnificence." I think humans have a basic need and desire to be significant, in particular, to be socially significant. I don't just want to be happy. You don't just want to be happy either. Choosing a course of life that leads to happiness is reasonably easy. Many people manage to be happy as peasants or hunter-gatherers. Most of them do so without being particularly ingenious or clever in their approach to life. It just doesn't require very much to be happy. However, most of those people are also in a position that makes it easy for them to be socially significant. Something about where they were born makes it easy for them to live a life that impacts other people the way they would hope to impact other people. Social significance has a lot to do with other people's happiness and a lot to do with having some ability to influence other people's happiness. Most of us have a desire to live in a way that causes our friends to benefit more than our enemies and for our enemies to be harmed more than our friends. People who manage to achieve this position have achieved social significance, and are no longer left craving it.
The obvious tie-back to success and revolutions is that successful people already have social significance. They can already afford to be magnificent. If I have enough money to take over a third world country, then I also have enough money to throw amazing parties, and to ensure that the people I like have wonderful jobs, etc. etc. Failing-to-be-my-friend would be punished relative to succeeding to be my friend, and I don't necessarily feel the need to distinguish much between my enemies and the people who are simply not my friend because I haven't interacted with them. People who I simply haven't interacted with yet should have a much easier chance of becoming my friends than anyone who has already made herself my enemy, so there is even a little bit of a distinction there that I can preserve without having to do anything that anyone else might call "evil."
People who haven't achieved magnificence, which I'm defining as the ability to reward one's friends to one's heart's content, have two options for pursuing social significance. They can attempt to become magnificent or they can attempt to become evil i.e. draconian punishers of all who oppose them. There isn't really much of a compromise position available because choosing to become "evil" taints you in the eyes of everyone who is not already your ally, and it also taints your allies and your friends in everyone but your allies' eyes. Your friends are not necessarily your allies. Most of us know and love at least one person who would be very hurt if we decided to turn "evil." If this is the case, we must become sufficiently evil to hurt the people we hate more than we've hurt the people we love if we are ever going to achieve social significance.
Some future AI may be able to more easily hide its future plans than we humans can, but it might have even a harder time doing so if people can examine its source code. However, if people cannot learn of its pre-commitment, it's hard to imagine why it would derive any benefit from having that pre-commitment... unless for some arbitrary reason that pre-commitment is a terminal value. Humans cannot make a pre-commitment a terminal value that we hide because our future values are shaped by our present actions. If I decided to be as magnificent as possible today and to postpone doing anything evil until I can get away with it, I make myself unlikely to choose to ever become evil because I will be continuously reshaping my personality and my values through my present choice to engage in magnificent actions. An AI is not required to share this property, so there is some difference between Roko's Basilisk as applied to the development of AI than there is when thinking of Roko's Basilisk as applied to the creation of a militant movement or when applied to someone's career trajectory or whatever. But in most other ways, the idea remains the same.
This is one of the interesting things about Roko's Basilisk, as distinct from Pascal's Wager, at least to me. Roko's Basilisk is a much better description of human motivation than Pascal's Wager. Being extremely powerful and telling people that they must believe you are extremely powerful before you demonstrate your power if they want you to show them mercy seems pointlessly arbitrary to me. In contrast, choosing to punish people who have opposed you is a fairly common feature of human movements; and it appears to be a successful tactic especially when people know you will use it.
But it is only a successful tactic in a certain context. Punitive, ruthless, draconian movements tend to be movements led by the disenfranchised poor, supported by the disenfranchised poor, and they only tend to succeed in places where the majority of the population is part of the disenfranchised poor. These are movements led by and supported by people who have no magnificent friends and who have very little hope of becoming magnificent themselves. When people have the option to join sides with someone on route to magnificence or someone on route to evil, most people choose to join and befriend the person becoming magnificent. It's the natural Schelling point because everyone can reach the simple conclusion that we are all better off if we all decide to side with people seeking magnificence than if we decide to side with people seeking to become evil. We all have finite resources to invest in causes and movements, so we can't infinitely hedge our bets.
No comments:
Post a Comment