How Specificity Works

by Liron6 min read3rd Sep 201948 comments


Philosophy of LanguageConversation (topic)

This is Part II of the Specificity Sequence

You saw what mayhem we brought forth when we activated the first power of specificity, the power to demolish bad arguments, and hopefully your curiosity is piqued to see what’ll happen when we activate all the other powers.

But first, let’s pause here to ask: How does specificity work?

Consider this dialogue between Steve and one of his pals:

Steve: Information should be free!
Steve's Pal: Whoa, this is thought-provoking stuff. Ok, so, would you say you're advocating for digital socialism, or more like digital libertarianism?

Oh jeez. Not only is Steve's pal not pushing for Steve to be more specific, the pal is an enabler who pushes Steve to be less specific. They're climbing the ladder of abstraction the wrong way.

The Ladder of Abstraction

When you want to nail down a claim, the operative word is “down”: you want to bring the discussion down the ladder of abstraction:

If Steve says, “Information should be free!” and I'm trying to understand what he means, here's what I'd say:

Liron: Ok, why do you think I shouldn’t have had to pay Amazon for my paperback copy of To Kill A Mockingbird?

This way, I'm nosediving all the way down to the bottom rung of the ladder of abstraction. Down here, the conversation becomes grounded in the concrete language of everyday experience, with substantive statements like “Amazon charged my credit card $6.99 and kicked back $1.15 to Harper Lee’s estate.”

And how about that free shipping, Steve?

Having this kind of grounded discussion is usually more productive than having a flingfest of the higher level ballpit-words “information”, “freedom”, “socialism”, and “libertarianism”.

As Steve and I are hanging out on the bottom rung of the ladder of abstraction, talking through specific examples of information getting exchanged freely vs. non-freely, we’ll be able to notice if certain features seem to remain constant across our chosen examples. For instance, we might discuss various hypothetical authors who yearn to write books for a living, and observe that all such authors can still have some plausible mechanism to earn money (running ads on their blogs?), even without directly charging for the privilege of reading their books.

After loading our brains with specific examples, we can then abstract over these examples, climb our way back up the ladder of abstraction, and put forth a generalized claim about whether “information should be free”. Since we've been careful to think about specific example scenarios that our claim applies to, we'll be putting forth a coherent and meaningful proposition - a claim that others can study under the magnifying lens of specificity, not an empty claim that gets demolished.

So when someone makes an abstract assertion like "information should be free", the best thing you can do is hold their hand and guide them down the ladder of abstraction. I know it’s tempting to skip that process and just attack or admire their original abstract claim. But show me two people discussing a topic in purely abstract terms, and I'll show you two people who are talking past each other.

How To Slide Downward

How do you take a concept and slide it down the ladder of abstraction to obtain a more specific concept? What mental operation must your brain perform?

In Replace the Symbol with the Substance, Eliezer explains how to do it with baseball terms:

You have to visualize. You have to make your mind’s eye see the details, as though looking for the first time.
Is that a “bat”? No, it’s a long, round, tapering, wooden rod, narrowing at one end so that a human can grasp and swing it.
Is that a “ball”? No, it’s a leather-covered spheroid with a symmetrical stitching pattern, hard but not metal-hard, which someone can grasp and throw, or strike with the wooden rod, or catch.
Are those “bases”? No, they’re fixed positions on a game field, that players try to run to as quickly as possible because of their safety within the game’s artificial rules.

Or let's say we're discussing our country's education system. Eliezer would slide it down the ladder of abstraction like this:

Why are you going to “school”? To get an “education” ending in a “degree”. Blank out the forbidden words and all their obvious synonyms, visualize the actual details, and you’re much more likely to notice that “school” currently seems to consist of sitting next to bored teenagers listening to material you already know, that a “degree” is a piece of paper with some writing on it, and that “education” is forgetting the material as soon as you’re tested on it.

Let’s try it ourselves with the concept of a school “lecture”. What do we get when we slide it down the ladder of abstraction?

"A stage presentation of publicly-available educational material, hand-produced and performed by a professor who works at your educational institution, which you watch by locating yourself in a set building at a set meeting time, and which proceeds in a fixed order and at a fixed rate like broadcast television pre-YouTube."

Wow. When you hear it that way, it raises a lot of questions:

  • How about using the best educational materials from the internet as a course’s official materials? Those are surely better than what any professor in your school has ever performed.
  • How about making it a standard expectation for students to consume lectures at their own pace, including taking advantage of the pause and speed up / slow down features?
  • How about not forcing students to show up to a lecture hall at a specific time?

If people had never heard the word “lecture”, if people were always forced to talk about lectures via a specific description of what a lecture is… well, then they would have killed off lectures by now.

I believe the college lecture is only alive today because the word “lecture” is a protective abstraction-bubble.

I grabbed this out of Steve’s ball pit

If you crack open the protective shell of “lecture” and press your nose close enough, you breathe in the stinky innards: “a stage presentation of publicly-available educational material”.

If everyone involved with the university system — administrators, professors, parents, students — were themselves cracking open the shell and taking a whiff of “lecture”, they would have noticed when the expiration date passed (the day YouTube went mainstream) and taken out the trash.

Instead, we’ve landed in a weird place where the concept of a “stage presentation of publicly-available educational material” is ridiculous on its face, while our ears still tell us that the concept of a school “lecture” sounds pretty good.

Ground Your Terms

You probably know that to have a clear discussion (or just a clear thought), you need to define your terms. How do you define a term?

S. I. Hayakawa illustrates an attempt to define the term “red” by connecting it to concepts higher up the ladder of abstraction (h/t Eliezer):

“What is meant by the word red?”
“It’s a color.”
“What’s a color?”
“Why, it’s a quality things have.”
“What’s a quality?”

This approach of defining something by sliding it up the ladder of abstraction doesn’t feel productive. It might help define “red”, but it’s neither necessary nor sufficient to define “red”.

Similarly, when I asked Steve to define what “exploiting workers” means in regard to Uber, and he put forth “to use selfishly for one’s own ends”, we found ourselves no closer to understanding what the heck his point was supposed to be.

So how can we nail down “red”? How can we slide it down the ladder of abstraction? Hayakawa illustrates:

“What is meant by the word red?”
“Well, the next time you see some cars stopped at an intersection, look at the traffic light facing them. Also, you might go to the fire department and see how their trucks are painted.”

Now we’re getting somewhere. This is a good enough definition to satisfy someone who previously didn’t know what “red” means. Whenever we define a concept in this manner, by sliding it down the ladder of abstraction, we can call it grounding the concept.

Grounding is easy enough to do. Just follow Eliezer’s instructions from the previous section:

You have to visualize. You have to make your mind’s eye see the details, as though looking for the first time.

For example: What is fire?

If you know chemistry, you might define it as “rapid oxidation accompanied by heat and, usually, light”. But if you don’t know chemistry, what would you say? Most people would give up.

Don’t worry, just follow the instructions to ground it. Close your eyes and describe what you see:

"The bright orange heat and light that appears when I strike a match, and can sometimes be transferred to other things it touches, and keeps appearing as long as there is wood and air around it."

Concepts have many definitions, and not all of them are groundings. But in daily life, grounding a term is usually as good as precisely defining it, if not better. So when a term is confusing, just slide that sucker down the ladder of abstraction.

Effort and Risk Asymmetry

If you observe your own stream of consciousness in a discussion, you might feel it being gently buoyed up the ladder of abstraction. You might start a discussion with a few specific statements about firetrucks, but before you know it you're talking about redness and colors in general.

Why is that? If the most productive kind of discussion is grounded and concrete, then why do so many people seem to relish the experience of pontificating and arguing abstractly?

Eliezer says in The 5-Second Level:

Over-abstraction happens because it’s easy to be abstract. It’s easier to say “red is a color” than to pause your thoughts for long enough to come up with the example of a stop sign. Abstraction is a path of least resistance, a form of mental laziness.

It seems our brain stores each concept with an easily-navigable pointer upward to its category (like red→color), but doesn’t store easily-navigable pointers downward to specific examples (like red→firetruck). I’m not sure what larger aspect of the brain's architecture accounts for this, but here's one observation about why the two operations are asymmetrical:

When you slide a concept up, you remove information. When you slide it down, you add information, and that requires you to make an arbitrary choice with more degrees of freedom. I think there's a sense in which going up the ladder of abstraction is safer, while going down is riskier: your underconstrained choice of specifics may leak information about you that your elephant in the brain is keen to monitor and filter.

Eliezer's description of how he intentionally sticks his neck out in arguments has always stayed with me (source), and I suspect it's related to why we find abstraction appealing:

I stick my neck out so that it can be chopped off if I'm wrong, and when I stick my neck out it stays stuck out, and if I have to withdraw it I'll do so as a visible concession.  I may parry[...] but I at least endeavor not to dodge. Where I plant my standard, I have sent an invitation to capture that banner; and I'll stand by that invitation.

When you say something abstract, like "information should be free", the space of possible things you can mean is vast. You're not sticking your neck out, you're not affixing your neck to a precise location in claim-space, so you never have to fear that an opponent's sword might slash there.

Plus, as a bonus, vague statements make you sound smarter. You're signaling more intelligence and sophistication talking about "digital libertarianism" than about "buying a paperback on Amazon". That's why abstraction is appealing.

The upshot is that you'll have to make a sustained conscious effort to acquire the skill and habit of activating your specificity powers. But it'll be worth it.

Next post: The Power to Judge Startup Ideas