Tuesday, August 2, 2016

"Why Charlie Munger has known no wise people who didn’t read all the time"

From Knowledge.SparkCapital, Dec. 3, 2015:

Cached Thoughts and Variations on a Theme
There has been much written on the benefits of reading. It strengthens memory, reduces stress, improves empathy, expands vocabulary, makes you wiser, smarter and probably more useful. Vitrix fortuna sapietia, goes the aphorism: wisdom conquers fortune. Doubtful panacea, you may say, yet I feel that conventional wisdom still underrates reading’s usefulness. “Reading will improve your mind” is about as banal a claim as “exercise is good for your abs,” and this total banality causes us to accept the claim as given, without meditating on its nuance, thereby missing out on the depth of its value. The importance of reading becomes more apparent if we consider ways the brain and mind may work, and if we seek to maximize the likelihood of our efficacy.
This essay will attempt to convince you that reading will make you better at life because of the way the wetwear in your head (i.e your brain) works.
I’ll use a pair of metaphors to frame the argument. Metaphor one is the notion of thoughts in the brain resembling the cache in a computer — what feels to us like real time thinking is probably just our brains retrieving stored memories in response to particular triggers. Metaphor two is a bit wordier — it imagines a concept as a dynamic, catalytic thing surrounded by a sphere of hypothetical variations of what that concept could become. Bear with me, it will (hopefully) make sense in context. The gist is that each “new” idea is a variation of preexisting concepts, which themselves are variations of other preexisting concepts.
If you make it through the (~16 minute) journey, I offer a reward: a widely accessible strategy for maximizing our reading effectiveness. It’s called Charlie Munger’s Latticework of Mental Models.
Metaphor One: Cached Thoughts
There is a concept in neurology called the “100-step rule” which postulates a constraint on the real time processing speed of the brain. A typical neuron can transmit an impulse to a neighboring neuron about once every five milliseconds, or around 200 times a second. If we assume that what feels to us like “real time” thinking happens in about half a second, then information entering your brain can only traverse a chain about 100 neurons long as you compute a real time solution/action/thought. From the moment the light enters your eye to the moment you recognize you are looking at Donald Trump strangling a cat with his bare hands, a chain no longer than 100 neurons could be involved. In other words, there cannot be more than 100 serial (i.e. one after the other) “steps”. For comparison, the Intel Core i7 chip in your MacBook can execute well over 100 billion serial instructions per second.
“Not to worry,” retorts some eccentric looking gentleman at the back of the party, “the brain is a parallel computer. While each neuron can only trigger a 100-neuron long chain in real time, billions of neural cells can simultaneously fire 100-neuron chains in parallel. This parallelism vastly multiplies the real time processing power of the brain.” He then points to your MacBook, which has multiple processing cores, and describes how it breaks a computational problem into discrete parts that can be solved concurrently (i.e. in parallel) by the different processors, each running billions of serial calculations per second. “Like a neural network!”, he shouts.
It’s here that our brain-as-a-computer analogy begins to break down. The brain can compute in 100 steps or fewer what would take a computer billions of steps to solve. Indeed, the largest conceivable parallel computer can’t do anything useful in 100 steps, no matter how many parallel processors you add. To understand why imagine you had to get 100 bohemian nonconformists a distance of five million steps from Times Square, New York to Burning Man, Nevada by pushing them one-by-one in a wheelbarrow (if you have seen pictures of Burning Man, this scenario might make more sense). You decide that this would take a long time (and no one deserves that much exposure to dogmatic conversations about “non-GMO cruelty-free vegan pumpkin spice squad goals”). One way to speed this up would be to hire 99 Uber wheelbarrow pushers to each take a passenger. Now the task goes 100 times faster. However, it still takes you a minimum of five million steps to cross the country. Hiring ten million more Uber wheelbarrow pushers would not provide any additional gain in speed since the problem cannot be solved in less time than it takes to walk the five million steps. So too in parallel computing: after a certain point, adding more processors doesn’t matter and no matter how many processors you add, a computer can’t calculate anything useful in 100 steps.
How then does our brain, that most miraculous three-pound grey blob, achieve in less than 100 steps what the fastest parallel computer imaginable cannot solve in a billion steps? Well if you had to write real time programs for billions of 100Hz (using Hertz here as a proxy for serial actions per second) parallel processors, one trick you’d use as heavily as possible is caching. That’s when you store the results of previous operations and look them up in memory next time you need them, rather than recomputing from scratch. “It’s a good guess that the actual majority of human cognition consists of cache lookups,” says artificial intelligence researcher Eliezer Yudkowsky. In other words, the brain does not “compute” answers to most problems; it retrieves answers that were stored in memory. When I throw you a ball and your hand moves to catch it, that is not your brain computing Newtonian Physics in real time. You are smart, but no one is that smart. Rather, what happens is that your brain has stored in memory, from years of repetitive practice, the muscle commands required to catch a ball and this temporal sequence is automatically recalled by sight of the ball.
Something similar likely happens with cognition. Somebody says “gun control” and your mind automatically dips into your memory cache to withdraw precomputed thoughts. Recognition, association, pattern completion. Kahneman terms this System 1 Thinking: fast, instinctive and emotional as compared to its slower, more deliberate, and more logical System 2 counterpart. Say we have a debate about politics or religion or some similar light topic. The discussion flows rapidly back and forth. We each offer arguments, evidence, thoughts, facts, counterarguments. To an observer, our mental volleying seems like an incredible amount of real time cognitive processing, especially given we could not have fully anticipated each other’s arguments. But it’s a good guess that most of this debate is a battle of cached thoughts pulled out in response to invariant trigger words and that very little new real time thinking occurs; that our effectiveness as interlocutors is largely determined by precomputed work. Combine cached thinking pattern completion with the cognitive limits imposed by the 100-Step Rule and it’s no wonder debates on contentious topics are so maddeningly ineffective. We change our minds less often than we think and repeat cached thoughts that we have accepted as truth without deriving them ourselves from first principles.
One cynical conclusion is that debates, particularly political ones, are hardly about convincing your opponent to change beliefs. Indeed this pursuit is often pointless since confirmation bias, commitment and consistency, hindsight bias, narrative fallacy, availability bias, scope insensitivity, anchoring, affect heuristics, and a host of other System 1 malfunctions will trump your 100-neuron chain attempt at seriously considering disconfirming evidence. The recent presidential debates have been an acute reminder of this futility: we mistake cleverness for content as candidates throw out evocative soundbites to elicit “applause light” reactions from the audience. It reminds me of that scene in Thank You For Smoking, where the protagonist (Nick) is teaching his kid (Joey) how to win debates:
Nick: Okay, let’s say that you’re defending chocolate and I’m defending vanilla. Now, if I were to say to you, “Vanilla’s the best flavor ice cream”, you’d say …?
Joey: “No, chocolate is.”
Nick: Exactly. But you can’t win that argument. So, I’ll ask you: So you think chocolate is the end-all and be-all of ice cream, do you?
Joey: It’s the best ice cream; I wouldn’t order any other.
Nick: Oh. So it’s all chocolate for you, is it?
Joey: Yes, chocolate is all I need.
Nick: Well, I need more than chocolate. And for that matter, I need more than vanilla. I believe that we need freedom and choice when it comes to our ice cream, and that, Joey Naylor, that is the definition of liberty.
Joey: But that’s not what we’re talking about.
Nick: Ah, but that’s what I’m talking about.
Joey: But … you didn’t prove that vanilla’s the best.
Nick: I didn’t have to. I proved that you’re wrong, and if you’re wrong, I’m right.
Joey: But you still didn’t convince me.
Nick: Because I’m not after you. I’m after them
“I’d never fall for a trick like that,” you may say, but unless you are trained to do otherwise — to consider disconfirming evidence in the tiny window where intelligence has a chance to act — you will likely rely on cached thoughts, and repeat fragments of other people’s beliefs without doing any real thinking yourself. Jonathan Heidt illustrates this point painfully and hilariously in his book The Righteous Mind, where he asks subjects bizarre questions like “Is it wrong to have sex with a dead chicken? How about your sister?” Most people agree these things are wrong when under interrogation in psychology experiments. But none can explain why. It’s like they penciled a conclusion to an exam question at the bottom of the page and, when pressed to justify this conclusion, they went to the top of the page and started scribbling down confirming cached thoughts. It’s kind of like that guy who has to go on TV and automatically justify any position taken by the president.

“Well, what the hell does this have to do with reading?” Glad you asked and thanks for the smooth segue. French microbiologist Louis Pasteur once opined that fortune favors the prepared mind. This is particularly true in a mind constrained by 100-Step limits where preparedness is largely a function of the breadth, depth, and intermingling of cached thoughts (this includes spontaneous bursts of creativity, which are discussed in greater detail in Metaphor Two). Given these limitations, one obvious strategy for mental preparedness is voracious reading and an accumulation of vicarious experience. The more you read, the more effective you become and, paradoxically, the more humbled you will be by how little you know. You will begin to see novel links and better understand the world around you. You will be quicker and more useful in real-time discussion. Your base of cached thoughts will build and these thoughts will intermingle and combine in novel ways, often resulting in serendipitous invention.

Which brings me to the second prong of my “reading thesis”: variations on a theme are the crux of creativity....MORE
HT: Jake Cahan