Foundations of Genetic Programming, Terascale Knowledge Acquisition, Introduction to Evolutionary Computing, The Ghost Map, Granta, The Oxford Handbook of Computational Linguistics, What Predicts Divorce?, The Blind Side, The Text Mining Handbook, Anticipatory Learning Classifier Systems, Memory-Based Language Processing, Scalable Optimization via Probabilistic Modeling, Foundations of Statistical Natural Language Processing, Blink, Universal Principles of Design, Genetic Programming IV.
Years ago I made a decision that if I ever was curious about a book, any book, and thought I might one day read at least a few chapters, I'd buy it. No questions. Ideas and knowledge are so hard to come by, but sometimes nearly infinitely valuable. Surmounting the barriers of time, motivation, and effort are hard; I might as well tempt myself by collecting a store of good bait.
It's worked well for me over the years. Novels and historical books get read linearly. Tech books get a random access scatter pattern where I basically try to follow some thread of interest throughout a collection of sources. I've currently got more technical books in the set than normal, since I was curious about whether AI had made any recent progress with web-scale datasets.
The divorce book has been causing me problems though. I bought that as a deep-dive from Malcolm Gladwell's Blink. One of Blink's chapters describes a psychologist, John Gottman, who can predict whether a married couple will get divorced within 6 years with 90% accuracy, based on watching a 15 minute videotaped interview of the couple. "That's amazing," I thought. "How on earth does he do that?" The details in Blink were sketchy.
So I bought Gottman's 500-page tome detailing his research, as well as related material about observing interactions and the Facial Action Coding System (FACS). I've even ordered FACS training software so I can try to learn to recognize the thousands of facial micro-expressions and what they mean. It seems like this would just be handy to know, in negotiation, or acting, or business, or life. (Try the smile test to see a little demo about what FACS is about).
I haven't gotten back to Blink yet because I'm still down in this sub-thread it spawned. It's been incredibly interesting, and I intend to blog about the whole business at some point. But in the meantime I have this book about divorce laying around. People see that and they instantly think they know why I have it. My mother-in-law spotted it and that's lead to all sorts of sideways glances.
I also had a book about how to own & operate your own bar sitting around. People would come over and ask my wife if I was thinking of leaving the technology industry and opening a bar. Good lord no. Foodservice and retail are the last things on earth I would get into. But there was the book in the restaurant supply store, and it had chapters like "How to know when your bartender is stealing from you" and I just couldn't resist.
Eyes work using a page fault mechanism. They're so good at it that you don't even notice.
You can only see at a high-resolution in a fairly small area, and even that has a big fat blind spot right exactly in the middle, but you still walk around thinking you have a ultra-high resolution panoramic view of everything. Why? Because your eyes move really fast, and, under ordinary circumstances, they are happy to jump instantly to wherever you need them to jump to. And your mind provides this really complete abstraction, providing you with the illusion of complete vision when all you really have is a very small area of high res vision, a large area of extremely low-res vision, and the ability to page-fault-in anything you want to see -- so quickly that you walk around all day thinking you have the whole picture projected internally in a little theatre in your brain.
-- Joel Spolsky, the Big Picture
It's a stretch, but you can kind of look at knowledge acquisition through reading that way too. Your head is a fast index in RAM. You can page-fault in whatever you want to know about from the slower offline world of paper and ideas. But that's a slow, expensive integration process. It requires reading, understanding, even sleep.
It's not just about wanting to keep the reading machine from getting bogged down in useless drivel or dead-ends. It's about actively managing the reading-input queue, nearly at the paragraph level, just like an engineering product queue. I think, if I've got time to read 20 paragraphs of something now...should they all be from this book, or that book, or should I scan 5 and then focus a bit? Am I still getting useful yield out of this thread, or could the next 10 minutes be better spent skipping forward?
You might think that's kind of a scattershot approach, like maybe I just can't focus on anything long enough to pay significant attention to it or something. But I don't think that's it. I'm a coder, I can focus like a madman for hours. But I don't have infinite time. I just want to maximize my data input yield.
I have an uncle who is a university English professor, and I'm sure he'd spit his coffee out at my engineer's approach to reading. But I'm pretty passionate about learning stuff, and I get even more excited about making use of that learning to make new stuff. It works for me...