Castles in the Air

Why am I not concrete in my thinking?

Why do I resist concrete practice activities? Why do I overwhelmingly seek abstract “research” over concrete practice?

Is it “because” of perfectionism? Why do I feel like I’m wasting my time if I choose a plan that is good but not optimal?

How many concrete projects have I really done in my life?

Learning and Reasoning

Concrete examples demand that you actually think, instead of hand-waving with vague concepts. And so they will lead to the best learning.

Plus, you have to actually build rich categories, instead of working with the deceptive fake-ideas in your mind.

Seriously, how can you say you know a category if you don’t have the exemplars that form it? Yes, perhaps you have its drawn-out logical definition (“a JPD is a table with probabilities for all possible configurations”), but that’s no use in practical thinking. You need to have example-based categories for fast reasoning (like a JPD for two variables and a JPD for 200 variables to know just how it looks in real life).

For example, you make plans for 10 years into the future based on the abilities you will have then. But, how can you know what it feels like to be 35 (if you’re 25 now)? You need to actually look at people who’re 35 and judge their physical fitness, their mental acuity, health, mood, etc. Study the empirical evidence on the general abilities of 35-year-olds as compared to 25-year-olds and, to make it stick in your mind, see it in the real life 35-year-olds around you. You probably won’t have as much energy or mental capacity or even the same drives as you have now. Account for that.

When looking for concrete examples of something, remember its broad sub-categories. For example, when trying to come up with examples of questions in information theory, remember that the three main kinds of information-processing are in learning, research, and problem-solving.

Robert Pirsig’s Seeing with Fresh Eyes

My explanation for why focussing on one brick unblocked the girl: she started working with feature-rich exemplars instead of barren labels. Seeing the red colour of the brick would have triggered other memories, looking at the shape of the wall would have reminded her of other things from her life, and so a train of thought would begin.

With the empty label though (“city of Bozeman”) she would have nothing to go on. She would keep asking herself - “Bozeman… Bozeman… Bozeman… come up with something” - but the only way it would trigger thoughts would be if she had studied the official answers in class, like in PG’s example: “The three main causes of the Civil War were…. Test: List the three main causes of the Civil War.” Otherwise, there’s nothing stored in her mind on the cue of “Bozeman”.

Lesson: You don’t really know an idea unless you can tie it to concrete examples. Otherwise, you’re just flying around in the universe of the formal model with nothing to tie you to reality (and thus nothing to contradict you). I think this is what Eliezer called the teacher’s password.

Updating on my New Understanding

I’ve discovered a major flaw in my thinking so far (lack of concrete thinking) but failed to update my plans on it. I ditched practicing of skills because I couldn’t make any progress at all. I didn’t know what to do or how to do it. It was all vague. Now I know exactly why I went wrong. Which means I can resume my practice. At full speed.

It wasn’t empiricism alone that held me back in getting good results in fitness or writing or thinking. It was that I was using logical thoughts - labels - instead of concrete examples, and thus my mind was stalling like an engine running out of fuel.

Where else was I being held back by my lack of concrete thinking?

Update on your new understanding of human cognition: look at “tree falls in a forest” (make your beliefs pay rent), cached thoughts, trying to try, etc. in a new light.

Never think Big Abstract Thoughts

Always work with concrete exemplars and variables. Hold the specific variables in your mind and move one step at a time. Otherwise, you’re just fooling yourself (“how do I use causality? how do I use causality?”).

The big abstract thoughts will follow naturally from the concrete ones. When you try to solve one simple puzzle (the ball weighing problem), you’ll understand the big idea of how entropy leads to efficient information-seeking.

Studying Academic Papers

Search on the keywords of “inductive learning” and “learning concepts and categories”.

Notes

Answer with concrete data, not abstract logic

As far as possible, when you’re trying to figure out some problem, talk in terms of concrete examples. Don’t keep things abstract. Like, if you need to figure out exactly what a causal model is, take three causal models and then theorize. Don’t just make guesses out of thin air. Use empirical evidence to guide your thinking.

Advice for Abstract Thinkers: Think depth-first, not breadth-first

Instead of flitting from topic to topic, pick one and go deep into it. Don’t move on until you’ve completely figured it out. This is how you’ve got your best rewards in the past.

However, you’re not going to feel like doing it. Be prepared for that. Push past the initial boredom. Do it even if it feels trivial; things will become challenging if you push deep enough.

A corollary is that there is no such thing as an un-challenging project. It depends on how deep you go into it. You can take the silliest work and push its limits till it practically merits a PhD thesis. (Hackers do this for everything they touch, I think, even if it looks silly and pointless from the outside.)

Remember the hacker deal: “you never have to work on boring projects, and in return, you’ll never allow yourself to do a half-assed job”. You get to work on cool, exciting stuff, but you have to hold yourself to a high standard.