On Knowledge

What constitutes Knowledge?

When I was in school, nature teacher told me that “Life” is that which “is born, grows, replicates, and dies”. Simplistic, and it stuck with me.

Now, what does this definition doesn’t do? It never mentions cells, carbon, or DNA. It defines life by its behavior, not by its substance. I want to try the same trick with knowledge, because the definitions I was given (“justified true belief” and its philosophical descendants) describe what knowledge is made of, and none of them help me with my actual, daily problem: I am exposed to more information than at any point in human history, and very little of it seems to become knowledge.

Data, information, knowledge

In some sort of rough semantic ladder, data is bits: a hard drive full of logs nobody reads. Information is data that surprises you (Shannon formalized this and deliberately refused to go further) declaring the semantic aspects of communication “irrelevant to the engineering problem”. Knowledge then, is information that has been metabolized: it changed your mental models, and therefore it changes what you do next.

The metabolic framing matters because metabolism takes time and energy. You can hold information the way you can hold food in your hands; neither nourishes you until it’s digested. And digestion has a rate limit. Which brings me to the actual subject of this essay: what happens when the supply of information vastly outpaces anyone’s ability to digest it (not because of mass media, but because of the structure of our relationships).

Densification of our social graph

Seeing Like a Network: An article in a blog named “Strange Loop Canon”, author is Rohit Krishnan, posted on Jun 19, 2024.

When modeled as one human -> one node; one relationship -> one edge; it’s quite evident that over the last century/decade, that the average degree of each node has increased significantly, causing the social graph to become more dense.

Some evidence: In 2007-2008, E. Horvitz and J. Leskovec find an average of 6.6 degrees of separation between 240M people after analyzing emails and instant messaging. In 2016, Facebook found that to be 3.57. In eight years, the measured distance between any two humans almost halved. Each of us went from being a node in a sparse village graph to a node with thousands of effective edges, each one a pipe that never stops flowing.

This creates a surplus of offer of information that one receives from their social network (not “those” social networks, your actual kin, peers, tribe).

There is a surge of anti-cultural, Thoreau-like efforts of escaping it.

“The push to create private spaces […] to truly express oneself […] is to help make the networks you’re in a bit sparser”

In software development, it is argued that sparse network structures facilitates the diffusion and generation of ideas among groups, while dense network structures affects the implementation of idea within each dense group. Sparse graphs are where new ideas are born; dense graphs are where they get executed and where they get homogenized.

The outcome of having a dense network is insidious but powerful. It means only the narratives which can go viral do go viral. The collective epistemic commons becomes filled with those narratives which outcompete the others and muscle their way to the top. It means that at a time of unprecedented low unemployment, high wages, high standard of living, GDP growth, high stock markets, strong dollar, people in the US still think they’re living in the worst of all possible times.

At some level of interconnectivity we all fall prey to the weaknesses of information deluge. Our attention is finite and so is our processing capacity for information. You can have the world do a denial of service attack on your cognition by overwhelming it with bits of information, so you’re stuck in place like a fly in amber. And it does this so easily that we haven’t even recognised when it happens let alone how to prevent it.

That too is why we have so many tools for thought, and ways to capture notes to search them afterwards, and tools for doing work about work, and endless lists and notes and contextual reminders and and and … It’s why we yearn for cultural islanding. It’s why there’s the neverending “current thing”. We’re all left tilting at the windmill of being a node in a dense network.

What density does to knowledge

If knowledge is metabolized information, then a dense graph attacks knowledge at both ends of the pipeline.

On the input side, it’s the denial-of-service Krishnan describes: attention is finite, the supply of information crossed the supply of attention decades ago, and every additional edge in your graph makes the ratio worse. The feed doesn’t grant digestion time; it punishes it. By the time you’ve thought carefully about one thing, forty new things have arrived, and the social reward for having an opinion about the newest one is higher than the reward for understanding the old one.

On the output side, density devalues what you manage to know. When everyone receives the same viral narratives at the same time, knowing what everyone else knows is worth almost nothing. That can’t surprise anyone, it can’t be traded, it doesn’t differentiate your actions from the crowd’s. The information that becomes knowledge in a dense network is, almost by definition, the information that arrived through a sparse path: the obscure paper, the conversation with the one friend who works in the field, the thing you figured out yourself because nobody had packaged it for virality yet.

Sparsity as a practice

I don’t think the answer is Thoreau’s cabin, although I understand the impulse. Sparsity, like exercise, seems to be something you now have to do on purpose, because the environment stopped providing it for free.

Some things that look like sparsification to me: reading books instead of threads (a book is a single, very thick edge to one mind, held open for twenty hours). Writing before reading reactions. Keeping a garden like this one, where notes accumulate and link to each other on my schedule instead of a feed’s. Choosing a small number of people whose judgment you actually trust, and letting them be your filter, what the village graph did for our ancestors without asking.

None of this is about receiving less information for its own sake. It’s about restoring the ratio between intake and digestion, so that some information has a chance to become knowledge before the next wave arrives.

That which is born, grows, replicates, and dies

So, what constitutes knowledge? Let me steal my teachers’ move.

Knowledge is that which is born, grows, replicates, and dies. It is born as surprise (a delta against your mental model, distinguishable from noise only by the meaning you attach to it). It grows by connecting to what you already know, which takes the slow, expensive digestion the feed denies you. It replicates when you teach it, write it, or build with it: knowledge that can’t be transmitted decays into private hunch. And it dies when it’s revised: real knowledge is falsifiable and eventually falsified, replaced by something that explains more.

By that definition, knowledge is alive. A dense network optimizes for the first step only (endless births, endless surprise) and starves the other three. The narratives that go viral are born a million times a day and grow in no one.

Something I’m trying to protect against, one sparse edge at a time.