One of my favorite early lessons in entrepreneurship was the idea of working “on” your business instead of merely working “in” your business.
To see the distinction, imagine running a restaurant. Here, working “in” the business is clear. Make delicious food. Offer great service to your customers. Keep the place clean and inviting. Being able to cook and host is often a motivation for many to start a restaurant.
Running a restaurant is a lot more than cooking and waiting tables. It’s business strategy, marketing, cost accounting and pricing. Working “on” the restaurant means thinking one layer above to examine what processes the business itself consists of and how you can improve them.
Many restaurateurs fail because they can’t think at that higher abstract level. They intimately understand the food and service dimensions. But they struggle because they can’t see the processes and systems that result in high-quality food, new customers and steady profits.
There is a pattern of thinking here, though, that’s a lot more general than just about business success. This is the idea of a basic level of understanding and a “meta” level, which takes as its objects the very elements of thinking in the basic level itself. I believe there’s reason to believe that much of what we deem “intelligence”, as opposed to mere calculation, involves this kind of “meta” leap in conceptual understanding.
“Aha!” Moments When Discovering the Meta
I can attribute one of the biggest changes in my own life to one of these moments of discovering a hidden meta layer. In this case, it was thinking about habits, goals and productivity systems instead of just the objects of those pursuits.
The “Aha!” moment for me was discovering that, instead of just trying to work on some project to achieve a particular goal, I could work on my habits directly to achieve that goal. Instead of blaming a failure on willpower or discipline, I could look at the habits that failed me and see how one could be redesigned in the future to avoid those problems.
After discovering this meta layer for myself, I became a little obsessed with it. I’m not alone. Many people I know who started blogs on personal development often do so with habits, goal setting or productivity systems as a first topic. The discovery of this “meta” layer to life can feel so profound that it’s hard to believe you didn’t see it sooner.
Meta-Metas or Turtles All the Way Up
There’s an old joke about an shaman and a scientist. The scientist asked the shaman what the origin of the world was and the shaman said that the world was resting on the back of a gigantic tortoise. The scientist responds smugly, “but what does the tortoise stand on?” The shaman responded casually, “Another tortoise.” “But what does that tortoise stand on?” the scientist asked again. The shaman replied, “It’s tortoises all the way down.”
Metas are a bit like tortoises, except in this analogy they stack up rather than stack downwards. Once you have a certain level of understanding of one layer, it’s always possible to reach out to a new higher layer and start seeing the “meta” of the layer you had previously discovered.
Consider our restaurant. The amateur restaurateur sees the business in terms of food and service. The smarter restaurateur sees it in terms of business processes that create the food and service. The smartest restaurateur, sees those business processes in terms of strategies that compel them. Metas on top of metas.
Or consider habits. The initial layer is to strive after things, and blame amorphous properties like willpower or motivation when you can’t reach them. The meta layer is to investigate the processes that guide willpower and motivation—habits, goals and systems. The meta-meta layer is to think about the ideas and philosophies that guide those meta-level objects. What kinds of goals should I have? What habits are meta-stable [1]? Should a system be thorough or sparse?
You Can’t Force the Meta
Clearly meta-understandings are incredibly valuable. Since a meta layer encapsulates the layers below it, you can always reason downwards, if that is more appropriate. The restaurateur who has reached the level of seeing business processes, for instance, doesn’t automatically forget about the food.
Given this idea, it might seem reasonable to ask whether we can generate these meta-level insights directly. The pattern is relatively clear—instead of reasoning about the objects directly, you reason about the higher abstractions that themselves reason about the objects. This might seem to form a “pump” so to speak, that would allow you to generate meta layers automatically, simply by thinking hard enough.
Unfortunately, however, I don’t think you can force it. I believe that this is because the meta comes from having detailed understanding of the layer below. If you don’t have that, the “meta” layer you generate has no power. You might be able to understand that it exists, but you can’t actually elevate your thinking towards it.
To give an example of this, about a year or so ago, I started learning to play chess. I had learned the rules of the game as a child, but I never had any skilled opponents and didn’t practice. Then, recently, I started playing again with a good friend who was quite skilled at the game.
Chess knowledge is easy to think in terms of layers. The most basic understanding of chess are the rules themselves. Bishops move diagonally. Rooks move horizontally and vertically. Pawns move only forward, except to capture, which has to be on a diagonal.
The first “higher” layer of chess is encapsulated in the patterns that are not part of the basic rules, but are inevitable consequences of them. One such pattern is a “fork” where your piece simultaneously attacks two of your opponents pieces, and can sometimes force them to sacrifice one if they aren’t properly protected. There is no rule for fork in chess. It’s a higher-layer that comes from understanding the basic rules well enough to see that this pattern exists above them.
But there’s more than just forks. Further layers of chess become increasingly abstract. Great players can often lose on lower-level principles of chess, such as material, sacrificing a pawn or piece, but gain on meta principles like activity or positional dominance.
The thing is, when I started learning about chess as an adult, I knew about these things. I heard about concepts like pawn structure, aggressiveness, sharpness in tactics, etc.. but I couldn’t *see* them. I knew those layers existed, but when a chess master pointed out that a certain setup was favorable to white because of one of these higher-level concepts, I was blind to it. They literally saw the chess board differently than I did because their mastery of the lower levels allowed a facility of “meta” understanding.
Side note: Sticklers may be rustled at my overly loose usage of “meta” here. “Meta” interpreted strictly means something is “about” itself. So meta-chess would be… chess about chess? Maybe that’s meaningless. Incidentally, this is a problem with “meta” not being a concept in and of itself, but a prefix which depends on which word is used as the base. Therefore, there are many things which exhibit “layer-hopping” in the same way as strictly “meta” ideas, but may not officially qualify because the correct noun is lacking to truly make it reflexive. This is a little bit of an unconventional extension of the idea of “meta”, so for those who want to limit it strictly, feel free to substitute my overuse of the word “meta” with the somewhat more general (and in my opinion less illustrative) idea of “abstraction.”
Meta and Chunks
These days, the popular account of such understanding is that of “chunks [2]”. Human working memory is famously limited to just a few objects. Our mental powers comes from being able, through exposure, practice and insight, to bind atoms of understanding together into larger and larger chunks.
Experts have expertise because their repertoire of lower-level chunks allows for increasingly abstract patterns to be deftly employed when they’re needed. Physics experts see physics problems differently than novices [3]. They see them in terms of deep principles rather than surface features of a problem. An expert might look at a problem and say, “oh this is a conservation of energy issue,” whereas a novice might say, “hmmm this is one of the ones that has a string and a pulley… which formula do you use for those again?”
This view of chunks also implies that meta-layers, although I’ve conceptualized them as existing discretely on top of earlier understandings, aren’t really discrete. While the idea of working “on” one’s business as opposed to “in” it seems fairly clear, there’s still a lot of crossover. Does opting for smaller portion sizes represent a basic-level strategy of presenting fancy-seeming dishes? Or is it a higher level strategy of market positioning? It’s often not clearly separate, and the reasoning compelling such choices can blur boundaries.
What I think is useful to the meta understanding, rather than thinking about chunks alone, is that the meta understanding implies that often what you’re doing when building expertise is making understandings that directly manipulate the upstream causes of your previous reasoning.
Meta is Math
There’s a powerful analogy here between the meta-climbing of increasingly abstract conceptual understanding and learning mathematics. In a way, mathematics is a kind of rarefied meta-level thinking.
The hallmark of math is this kind of conceptual climb. You start by counting on your fingers. That generalizes to numbers that go past ten. Then you get arithmetic—plus and minus, multiply and division. What if you multiplied the multiplier? That gets you to exponents and logarithms. What if you took partial numbers, extended the number line backwards or allowed it to rotate? That gives you the continuum, negative numbers and complex numbers, respectively.
The hallmark of mathematics is to take one level and generalize it or extend it in some way. This goes on and on and on until eventually you have things like commutative rings in abstract algebra, which are completely opaque to anyone without advanced mathematics degrees.
Meta-levels have a similar flavor. There’s the layer you understand. There’s the layer above that you can see exists, but can’t really work with. Then there’s the layers above that which you don’t understand at all.
Depending on your feelings towards the topic, you may feel those meta layers are genuine, and represent a deficit in your understanding, or you may feel they represent and increasingly elaborate form of intellectual masturbation, with people coming up with increasingly esoteric descriptions of a fundamental lie.
The existence of a meta-generating process doesn’t say anything about its veracity, unfortunately. A genuinely useful process may get derailed if it introduces a falsehood at a lower layer. Then, you might get increasingly sophisticated elaborations of that falsehood. Alchemy and astrology had tons of smart people as adherents, but the sophisticated abstract understandings collapse since that’s not how stars or substances actually work.
Pathways to Finding the Meta
While I don’t think you can “force” a meta-level understanding where the object level is insufficient to support it, I do think the opposite is possible. That is, I think it’s possible to have a rich object-level and simply not notice the meta level sitting above it.
This explains why discovering the meta, for many, is an “Aha!” moment of insight. Discovering habits is such a personal development leap for many because they’ve trained themselves to take action to reach goals, but it didn’t occur to them to think about the processes that generate the behavior itself. If you’re ready, that insight can be digested relatively rapidly.
This suggests a path for finding the meta in many areas of life, whether it be self-development, chess, physics or art:
1. Acknowledge that there are levels beyond what you can currently perceive.
This is especially important if you have a tendency to dismiss them. “History is just one thing after another,” or, “All modern art is bullshit,” are meta-dismissing comments. Now it’s certainly possible that established views are wrong. It may be the case that historians are overconfident in their conceptual understanding of how events unfold, or that modern art’s philosophy is suspect.
However, if we accept that virtually everything has meta layers and meta-meta layers, turtling all the way up, then one needs to be careful dismissing the layers one can’t currently see as not actually existing. As I explained earlier, I simply couldn’t see the chess concepts the grandmasters were talking about. If chess had been a more subjective domain, I might have wanted to dismiss their concepts entirely simply because I lacked the ability to see them.
2. Probe your own meta-levels. You may be ripe for an “Aha!” moment in one of them.
Ask yourself what would be the meta level of the problem domain you’re trying to work in. What would it mean to “go up one level” in your business, career or philosophy? The answer may give you clues as to how to climb up there, if you’re ready.
3. Develop richer understandings one level below.
If you’re not ready, the answer will probably be vague or unintelligible, just as the advanced chess concepts were to me. The solution here isn’t to dismiss the meta or ignore it, but to work on enriching your understanding of the layer below. As that foundation ripens, it will be easier and easier to think of it abstractly until you’re ready to move up to the layer above.
Climbing Meta-Ladders
This idea, that meta-layers can only be reached with sufficient understanding of the layer below, I think prevents the biggest worry of this chain of reasoning, namely that if you obsess over the meta layer, where does it stop? Don’t you just go off into an increasingly heady realm, detached from the object-level concerns until you start asking those bizarre philosophical questions like what the meaning of the word “is” is?
This worry, of course, is perhaps a symptom of runaway abstracting without first trying to get a grip on the layer below. Properly construed, a higher meta layer should have even greater familiarity with the objects below it, so that they are enriched by that understanding, rather than forgotten.
That being said, perhaps there is a “meta” to this entire chain of meta-reasoning I’m presenting right now. Namely, when to think about the meta layer and when to focus on the ground-layer itself. If that’s the case, then perhaps I haven’t developed a sophisticated enough understanding to access that rung in that meta-meta. However, I’ll keep thinking about it, and perhaps I’ll discover something once I do.