Make Your Own Rules

February 15, 2018 By Venkatesh Rao

We seem to be in the middle of a renaissance of rules for life. Not since Robert Fulghum’s All I Really Needed to Know, I Learned in Kindergarten (1987) and Stephen Covey’s 7 Habits (1989) has there been such a peak of interest in such rules. Then, as now, we were going through a period of deep global changes, and everybody was very anxious because nobody knew what the new rules for the new normal were.

The proximal trigger of this current wave is I think, Jordan Peterson’s 12 rules, as well as the late John Perry Barlow’s 25 principles, which have both been doing the rounds. But the root cause is growing market demand for anomie-busting.

Well of course if there’s a gold rush of this sort on, I have to sell pickaxes. And my pickaxe is a DIY template for making your own set of life rules. Here’s an in-progress snapshot of the pickaxe in action in my own notebook (cleaned-up version with readable annotations key further down, but I wanted to share the working version, which includes several technical mistakes). My model may be a bit hard to grok if you haven’t been reading me for a few years, but the good news is, it’s color-by-numbers easy to use. And all it takes is pen and paper.

I only have one actual imitable rule to offer in the marketplace of life rules: Make Your Own Rules. But I do think I have a good theory of life rules, and a meaningfully systematic procedure for generating them that I’m hoping to sell to the Deep Mind team for making well-behaved AIs.

In the short term, other people’s rules can get you through a rough patch. In the medium term, you have to at least adapt them to your own life. But in the long term, only making your own rules works.

Because, to snowclone what Eisenhower said about plans, rules are nothing, but rule-making is everything.

We humans like making up sets of rules and principles for life. The modern consumer market offerings, from Covey to Peterson, follow in the tradition of both ancient ones like the Ten Commandments of Christianity, the Five Pillars of Islam, and the Eightfold Path of Buddhism. These traditional, institutional ones are the ancestors of both the modern consumer-grade genre, and the related enterprise-grade genre of manifestos and mission statements.

Organizations are not immune to anomie-inducing environments, so it’s no surprise that we’re also going through a season of more-sincere-than-usual corporate soul-searching, manifesto-crafting, and mission-adopting. It’s not just individuals who need new rules for the new normal. Apparently Facebook does too, and is at least pretending to figure out new rules for the world of Fake News and election hacking. One of Facebook’s new rules is apparently to ensure that "time spent on Facebook is time well spent", a phrase they’ve taken from my don’t-hack-my-attention-dammit activist buddies Tristan Harris and Joe Edelman.

But I won’t get into that side of things in this post. We’ll stick to personal life rules here.

Thinking about this broad set of examples, traditional and new, institutional and individual, it struck me that though they are good as fodder for reflection, I’ve never actually been able to use any life rule set for its nominal purpose.

I’d be hard pressed to even enumerate any of the sets I’ve encountered in full.

The famous ones are more memorization challenges for trivia contests than operating systems that can be encoded in neural firmware as embodied virtues. They aren’t routinely helpful in navigating the ambiguities and uncertainties of life. Nor do they set meaningfully universal go/no-go boundaries.

They do not, in other words, constitute clean MECE (mutually exclusive, collectively exhaustive) coverage of the messiness of life, capable of bringing order to chaos in any sort of guaranteed way.

You could leave it that of course, and treat such sets of life rules as rhetorical devices, meant to do no more than trigger some helpful reflection, and perhaps inspire your own adaptations or original rules, but I think that’s not true. There’s more to life rule sets than meets the eye, and there’s ways to adapt or construct them so they actually compile properly into your life so to speak.

Life Rule Sets

Let’s call this genre (at the individual level) life rule sets, or LRS. An LRS is a set of between 5-25 statements (based on our sample) that apply to common life situations. The range seems to reflect an easy-to-memorize limit on the one end, and the size of a poster on the other. Less than 3 is probably too few to be generally useful in any way (though the Golden Rule is a good singleton exception, as is my own Make Your Own Rules rule), and more than 25 is really just laziness at theorizing a more compact structure.

What can we say about LRSes?

An LRS is not a manifesto. Manifestos are about beliefs rather than action. Even though an LRS may encode implicit normative or positive beliefs, that is not their primary purpose. This is because an LRS is almost always built around an idea of individual agency. Its subject matter is what you are personally responsible for and can act on.

An LRS is normative. Even if they are worded as abstract reflections, the elements of an LRS are actual rules, not random observations about the world. Some are very explicit ("thou shalt not covet thy neighbor’s wife") and some take a lot of thought to translate into imperative form ("Right Speech" in Buddhism, gee thanks Siddhartha).

An LRS is abstract. A good example is Covey’s "Be Proactive". When they are weirdly specific, such as Peterson’s "Do not bother children when they are skateboarding," the author is generally either attempting to inject a bit of whimsy into an otherwise serious exercise, or offering a mnemonic for an anecdotal illustration of a more general principle. So the seeming specificity is some mix of synecdoche, parable, and allegory.

At least that’s the case with the modern ones. Occasionally, in older religious LRSes, you find genuinely specific and serious rules that seem arbitrary, like "don’t eat pork", and eventually acquire taboo status or ritual significance. But you can usually find a historicist explanation for the rule. I don’t know the reasoning behind the pork taboo, but one good explanation of the beef taboo in India is that cattle at some point got too valuable as draught and dairy animals to be eaten.

An LRS may imply a script. Some clearly imply a particular larger life script. The eightfold path makes most sense understood in the context of a life of spiritual discipline and pursuit of enlightenment. Covey’s is explicitly a Christianish life script (his context-setting visualization exercise is to imagine your own funeral). Four of the five pillars of Islam are about life habits, but the fifth is a script act (the hajj).

An LRS is usually either idealistic or tragic. The one big dividing line I’ve found in my quick survey is that LRSes seem to divide cleanly into two kinds: those that assume some sort of perfectibility metaphysics (humans can grow and evolve within and across lifetimes as a species) and those that assume some sort of irredeemably fallen condition, from which only the grace of some sort of supernatural agency can lift you. Of the ones I’ve mentioned, the Buddhist eightfold path seems to be the only idealistic one. I’ll characterize this boundary more clearly later.

An LRS usually encodes a virtue ethics. Though they may encode some elements of consequentialist and deontological ethical systems, an LRS usually lacks the systematicity required to be anything other than a virtue ethics.

In the best case, the rules in an LRS serve as a reference sample of behavioral cues from the life of someone who has lived a life both exemplary in its modeling of desirable virtues, and rich enough to have actually been tested across a sufficiently large range of human life experiences.

In other words, an LRS is usually a what-would-Jesus-do type artifact, abstracted away from the specifics of an individual life to greater or lesser extent.

Theorizing Life Rule Sets

Life rule sets vary in the context-free systematicity they offer, but basically no LRS I’ve seen achieves the analytical rigor of, say, Newton’s laws.

Some sets fit together independent of context better than others, but most of them seem, by and large, entirely arbitrary. A more or less poetic selection of behavioral Schelling points from somebody’s life. The bigger sets tend to the baroque, and are impossible to even parody.

I think this is because the underlying phenomenology that life rule sets cover has never been properly characterized. We’ve never been quite clear about a few basic questions of the sort a mathematician interested in logic and foundations would ask:

  1. What are life rule sets about? Where is the phenomenological scope of life rule sets? Where does that scope end and the scopes of (say) traffic rules, peanut grades, and tax laws begin? In other words, what is the domain of the LRS?
  2. Can we expect an LRS to uniquely classify every situation within its domain, and apply a unique rule to it, and expect as output a unique helpful cue about what to do, thereby narrowing the range of options and simplifying decision-making? In other words are the rules in an LRS mutually exclusive?
  3. What is the coverage of a set of rules within its overall domain? How often does the rule set apply when you feel a need for guidance, versus having nothing to say. In other words, are the rules in an LRS collectively exhaustive?
  4. If an LRS is not MECE (mutually exclusive, collectively exhaustive), can we at least expect that if more than one rule applies to a situation, they will both point in the same rough direction? In other words, is an LRS internally consistent?
  5. If it is not internally consistent, is it at least paradoxically generative? Does it comprise a set of yin-yang oppositions that help you think generatively and dynamically about a situation, in terms of transformations and trade-offs? In other words, is an LRS a dialectic as well?
  6. Are all the rules in an LRS citizens of the same order within the LRS, or are there lesser and greater ones? In other words, does an LRS admit a cardinal or ordinal ordering?
  7. Are all the rules in an LRS independently construed, or are some the consequences of others? If so, what are the independent ones? In other words, are the rules of an LRS independent or do they have an axiomatic structure?
  8. Are the rules legible or illegible in their mode of action? Do they only work if you understand a principle behind them, or do they work irrespective of your understanding of them? In other words, are they explicit or implicit belief systems?

These are not fussy, pedantic questions. They matter. You can’t expect to make random rules about grades of peanuts and expect to see a reduction in traffic accidents. Less flippantly, there’s no particular reason to expect a rule about when to wear ties to affect whether you get depressed. There might well be such a discoverable relationship, making a tie rule a good, indirect, life rule. But it might also just be an arbitrary rule that’s just behavioral noise.

Depending on the answers to these questions, we tend to develop an aesthetic, taxonomic sense of the universe of life rule sets.

Some strike us immediately as doctrinaire and serious, others as whimsical and subversive. Some are self-evidently too silly to even try to use. Others you have to test out for a while before the gaps, overlaps, paradoxes, orderings, and axiomatics become evident (ie there is a depth of learnability to a life rule set that can be either shallow or deep). Some are meant to box you in and confine your behaviors to safe zones, while others are intended to break you out of boxes and safe zones.

Any good theory of life rule sets must provide a reasonable account of these questions. Otherwise there is no there there.

Let me take a stab.

Task-Negative Cognition

Here’s the central dogma of my theory of life rule sets: Life rule sets are about governing task-negative cognition.

The human brain has two major processing modes, task-positive and task-negative, which occur in the task-positive network (TPN) and default mode network (DMN) respectively. Brain activity in these two networks is anti-correlated. You’re in one mode or the other. The two modes relate to what I’ve previously written about as manipulative and appreciative cognition.

Task-positive cognition occurs when your brain has its attention occupied by a clearly defined and bounded external task. Generally one with very strong, and fast (ie with short time constants) closed-loop sensorimotor feedback loops connecting your behavior to the environment, via immediate consequences. Examples include driving or cooking.

If you’re a simple robot or insect, this is all you need. Task-positive cognition is insect cognition. I covered these in my post human-complete problems. Human cognition is more than the sum of insect-like task-positive domain level rules.

You don’t even really need rules because the domain itself serves as a perfect source and enforcer of correct behavior. You’re basically under the jurisdiction of strong and active laws of nature.

You drive on the wrong side of the road, you die. You don’t follow the "rules" of fire in the kitchen, you get burned. As a bug, if you don’t lay your eggs on the right kind of plant, you don’t make baby bugs.

You don’t need explicit principles of salience because they are generally obvious and self-enforcing. It is easy to stay focused on what matters and ignore what does not. The deer jumping in front of your car is highly salient. The billboard trying to sell you washing machines is highly not salient, but trying to be.

Task-negative cognition on the other hand, is what your brain does when there is nothing in particular it has to do; when the situation has no demanding urgencies to attend to. The DMN (which used to be called the task-negative network) becomes active.

The default mode network was a surprising discovery. Neuroscientists apparently thought the awake brain sort of just idled and did nothing if it had nothing externally anchored to do, like a car in park with the engine on. In other words, it was assumed that the opposite of a brain at work is a brain at rest.

Turns out, that’s not true. Our default waking mode (and also during dreaming phases of sleep) is not a brain at rest, but what is known as mind wandering. Mind wandering is the brain basically running potentially useful prospective thought processes. What computer scientists call speculative execution, the stuff behind the recently revealed Intel chip security flaws, while idly also paying non-vigilant attention to the environment in an open-ended way, without a strong salience filter.

When you are driving, your brain strongly filters for driving-related cues. When you are in mind-wandering mode, you are liable to notice anything vaguely salient to anything that concerns you, even if it isn’t currently the focus of active effort. But you aren’t vigilant the way you are in task-positive work (especially dangerous kinds).

Task-negative cognition is how you solve human-complete problems. Those open-ended, leaky kind of problems where any seemingly specific problem like "find a satisfying job" ends up being equivalent to the life-the-universe-and-everything problem of just living a good life.

Task-negative brain activity is not rest, but human-complete problem solving in the context of an entire life. Important-but-not-urgent stuff that takes every spare cycle it can find, in the gaps between task-positive processing.

It is your brain making sure it doesn’t get caught in ruts, miss important-but-not-urgent unexpected news, or obviously foreseeable contingencies. It’s your brain simultaneously figuring out the how and why of life in play mode.

In Boydian terms, the default mode network is all about reorientation; about swapping mental models around, reframing things, approaching things from new angles, considering the salience of new sensory inputs, and so forth.Task-positive work is about operating within an orientation. You know what you’re doing and what the environment is doing, and you’ve loaded up a bunch of habits at various stages of maturity and trying to be effective in a specific way: driving without killing yourself, cooking without burning yourself or the food.

Task-negative cognition is a natural and neuroscientifically meaningful definition of the scope of life rule sets. When you think about the part of human life where there are no necessary rules, no active and direct feedback mechanisms confining and correcting behaviors, you’re talking about stuff the default mode network gets up to.

So what does the DMN get up to? How does it actually run task-negative cognition to solve human-complete problems?

Where the Mind Wanders

In task-positive work, generally, if you don’t care, it doesn’t matter. Task positive work is inconsequentially under-determined.

You learn how to make tea, but you don’t have a preference for green or black. Well then, it doesn’t matter. You don’t need a rule. Do whatever. Green tea versus black tea is a distinction without a difference for you. Perhaps all you care about is a mild shot of caffeine in the form of a hot beverage.

Task-positive domains may be underdetermined, but they are not consequentially under-determined. You generally have enough information (or can discover enough) and there are enough constraining laws and dynamics, that you can compute a unique answer, solution, or decision that works. And you rarely have to change it. Insects can install the answer or solution in firmware, and humans can make it an unconscious habit.

Any leftover stuff is in the distinction-without-a-difference department where random and/or unconscious variable bindings and commitments will do.

This has a crucial effect: for task-positive work, means-ends reasoning is enough. You don’t need values, principles, rules, or ethical ruminations.

The DMN on the other hand, concerns itself with stuff that is consequentially underdetermined, but in a situationally non-urgent way. You do care, and it does matter, but you don’t have to make up your mind about what to do right now. The world isn’t conveniently narrowing down the options to one obvious right answer, but it also isn’t pushing you to bet your life on a coin toss.

More data may not help. If there are deeper physics type laws at work, you may not yet have discovered them, and might never do so.