Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

The ladder of abstraction and giving examples

-4 Post author: Elo 16 August 2016 05:31AM

Original post:  http://bearlamp.com.au/examples/

When we talk about a concept or a point it's important to understand the ladder of abstraction.  Covered before on lesswrong and in other places as advice for communicators on how to bridge a gap of knowledge.

Knowing, understanding and feeling the ladder of abstraction prevents things like this:

  1. Speakers who bury audiences in an avalanche of data without providing the significance.
  2. Speakers who discuss theories and ideals, completely detached from real-world practicalities.

When you talk to old and wise people, they will sometimes give you stories of their lives.  "back in my day...".  Seeing that in perspective is a good way to realise that might be people's way of shifting around the latter of abstraction.  As an agenty-agent of agenty goodness - your job is to make sense of this occurrence.  The ladder of abstraction is very powerful when used effectively and very frustrating when you find yourself on the wrong side of it.

The flipside to this example is when people talk at a highly theoretical level.  I suspect this happens to philosophers, as well as hippies.  They are very good at being able to tell you about the connections between things that are "energy" or "desire", but lack the grounding to explain how that applies to real life.  I don't blame them.  One day I will be able to think completely abstractly.  Today is not that day.  Since today is not that day, it is my duty and your's to ask and specify.  To give the explanation of what the ladder of abstraction is, and then tell them you have no idea what they are talking about.  Or as for the example above - ask them to go up a level in the ladder of abstraction.  "If I were to learn something from your experiences - what would it be?".


Lesswrong doing it wrong

I care about adding the conceptual ladder of abstraction to the repertoire for a reason.  LW'ers are very good at paying attention to details.  A really powerful and important ability.  After all - the fifth virtue is argument, the tenth is precision.  If you can't be precise about what you are communication, you fail to value what we value.

Which is why it's great to see critical objections to what OP's provide as examples.

I object when defeating an example does not defeat the rule.  Our delightful OP may see their territory, stride forth and exclaim to have a map for this territory and a few similar mountains or valleys.  Correcting the mountains and valleys map mentioned doesn't change the rest of the territory and does not change the rest of the map.


This does matter. Recently a copy of this dissertation came around the slack - https://cryptome.org/2013/09/nolan-nctc.pdf.  It is a report detailing the ridiculous culture inside the CIA and other US government security institutions.  One of the biggest problems within that culture can be shown through this example (page 34 of the report):

The following exchange is a good example, told to me by a CIA analyst who was explaining the rules of baseball to visitors who didn’t know the game: 

Analyst A: So there are four bases-- 
Analyst B: -- Well, no, it’s really three bases plus home plate.
Analyst A: ... Okay, three bases plus home plate. The batter hits the ball and advances through the bases one by one— 
Analyst C: -- Well, no, it doesn’t have to be one base at a time.

And these ones on page 35:

The following excerpts from stories people have told me or that I witnessed further illustrate this concept: 

John: I see you’ve drawn a star on that draft. 
Bridget: Yeah, that’s just my doodle of choice. I just do it unconsciously sometimes. 
John: Don’t you mean subconsciously? 

Scott: Good morning! 
Employee in the parking lot: Well, I don’t know if it’s good, but here we are. 

Helene: I am so thirsty today! I seriously have a dehydration problem. 
Lucy: Actually, you have a hydration problem. 

Victoria: My hopes have been squashed like a pancake. 
James: Don’t you mean flattened like a pancake?

For those of us that don't have time to read 215 pages.  The point is that analyst culture does this.  A lot.  From the outside it might seem ridiculous.  We can intellectually confidently say that the analysts A, B and C in the first example were all right, and if they paid attention to the object of the situation they would skip the interruptions and get to the point of explaining how baseball works.  But that's not what it feels like when you are on the inside.

The report outlines that these things make analyst culture a difficult one to be a part of or be engaged in because of examples like these.


We do the same thing.  We nitpick at examples, and fight over irrelevant things.  If I were to change everyone's mind, I would rather see something like this:

No one denies that people have different metabolisms.

Statements including "no one denies that ..." are usually false.  Regardless, my goal here was to...  

Taken literally, yes. However these statements are not intended to be taken literally...

Turn into:

No one denies that people have different metabolisms.

My goal here was to ask people...

(*yes this is not a very good example of an example, this is an example of a turn of speech that was challenged, but the same effect of nitpicking on irrelevant details is present).


Nitpicking is not necessary.  

Sometimes we forget that we are all in the same boat together, racing down the river at the rate that we can uncover truth.  Sometimes we feel like we are in different boats racing each other.  In this sense it would be a good idea to compete and accuse each other of our failures on the journey to get ahead.  However we do not want to do that.

It's in our nature to compete, the human need to be right! But we don't need to compete against each other, we need to support each other to compete against MolochAkrasiaEntropyFallacies and biases (among others).


I am guilty myself.  In my personal life as well as on LW.  If I am laying blame, I blame myself for failing to point this out sooner, more than I blame anyone else for nitpicking examples.  

The plan of action.

Next time you go to comment; Next time I go to comment, think very carefully about if you can improve, if I can improve - the post I am commenting on, before I level my objections at it.  We want to make the world a better place.  People wiser, older, sharper and witter than me have already said it; "if you are looking for where to start...  you need only look in the mirror".


Meta: this took 3 hours to write.

Comments (3)

Comment author: Lumifer 16 August 2016 02:48:42PM 2 points [-]

"Technically correct: the best kind of correct" :-P

Comment author: Viliam 16 August 2016 09:06:09AM 0 points [-]

Biting my tongue to avoid an object-level response.

Seems to me that in real life this is a smaller problem than in online communities. Because real-life interactions allow various kinds of signals that are difficult to translate into comments and votes. For example, in real life people usually stand in a small group and talk; it is obvious who is and who isn't a part of the group, even if some people don't talk at the moment. You know who is listening to what. But in the online debate, I have no idea who is reading what, until they reply.

If there is a person in real life who only offers nitpicking, others may decide, independently, to only discuss important stuff in their absence. The person will simply be not invited to an important debate. This doesn't require any confrontation like "John, go away, we want to discuss stuff we care about, and your contributions are predictably not helpful". It's more like Alice invites Bob, Cathy and Dan to discuss a topic she considers important, and no one invites John. He either will not know what happened, or there will be a socially acceptable excuse "well, you were not there when the topic came up".

Is there an equivalent of this in online debates? I guess on Facebook one can start a conversation and invite a group of people. But on LW, there is no such mechanism.

On the other hand, the cost of invitation-only debates (both on Facebook and in real life) is that many potentially useful contributors are excluded. So there seems to be a trade-off between open debate and private debate.

A possible technical solution: imagine a chat where anyone can start a debate on a new topic, and the person who started the debate automatically becomes a moderator of the thread. If you think the moderator is abusing their powers, you can start a new thread on the same topic. (There can be other safeguards against mod abuse, such as making it possible -- but trivially inconvenient -- to see the censored comments. So the people can decide whether it is worth to continue debating the same stuff somewhere else.)

tl;dr -- I agree that nitpicking can become a serious problem, and that it is more frequent among nerds. However, I also think that this problem is further complicated by the software we use, if it doesn't support solutions similar to those we would naturally choose in real life.

Comment author: 9eB1 16 August 2016 07:14:19AM *  0 points [-]

There is a programming group called Recurse Center and they have a set of social rules designed around tempering some of the more frequent hacker personality quirks that can stop people from getting along. They are pretty interesting, and you can read about them here under the heading "Social Rules."

Overall I think the environment they describe doesn't sound like my bag of rice, but I think those rules are eminently reasonable.