As our son reminded me the other day, the classical definition of knowledge, going as far back as Plato, are statements that are justified, true and believed. Each of those criteria though is problematic and lacks a clear meaning, despite the best efforts in epistemology. I have become enamored with a completely different approach that considers knowledge as the subset of information that is reproduced by humans over time. This interpretation is my reading of David Deutsch‘s views on knowledge as glimpsed from his books.

Before I explain in more detail let me give a couple of examples. Calculus is an example of knowledge under this approach. People are writing books, papers, websites etc. to explain calculus in a way that reproduces it for subsequent generations.  The music of Beethoven is another example. Again, it exists as externalized information such as recordings and sheet music that people have been reproducing over time.

What about examples that are not knowledge? The proverbial sound of a tree falling in the forest when no one is around to listen is definitely not knowledge. It is just a temporary sound wave that comes and goes. Neither are the orbits of the planets around the sun knowledge. They are information that is currently contained in nature. In both of these example the information exists entirely independently of whether we humans are around or not. It is also not reproduced — the sound of the tree falling exists for a very short time, the orbits of the planets for a fairly long one (but will still go away eventually).

So now let’s dig deeper into this definition of knowledge.

First, it is entirely external to any one human. It has nothing to do with what may or may not be in your brain or have been in anyone else’s. Instead it refers to information carried in a medium that can and is being reproduced over time by humanity as a whole. So that gets rid of whether or not anyone believes it. It also allows for something like music or art to be knowledge for which the concept of belief doesn’t apply in the first place.

Second, knowledge under this definition is distinctly human (at least here on Earth and for now, see below). No other species produces externalized information which is then reproduced by that species (*). That doesn’t mean though that other species aren’t reproducing some information. For instance, all species reproduce (sometimes with the help of others) the information contained in their DNA. Even the information contained in viruses is reproduced, albeit with the help of other organisms. So knowledge is the subset of externalized information reproduced by humans, which a subset of all information that gets reproduced, which is in turn a subset of all information.

Third, this definition of knowledge has an explicit role for time. Something may be knowledge at one point and then stop being knowledge as people stop reproducing it. This gets rid of the need to appeal to a concept of “truth” in defining knowledge. Instead, what is knowledge is what passes the test of time. Using this definition makes the Lindy effect a useful guide to the importance of knowledge (i.e. more important the longer it continues to be reproduced). To see how little “truth” matters in this definition just note that something “true” can stop being knowledge. For instance anything that didn’t have a copy outside the Library of Alexandria could be an example of that. The converse also holds. Something may not be “true” and yet be reproduced because of its relevance to other knowledge. For instance, we now are pretty sure that there is no aether in which light waves travel. Yet we reproduce that knowledge because it leads up to Einstein’s key insights into light. So the definition happily frees knowledge from the requirements of truth and justification also and instead institutes time as the crucial test.

Fourth, the definition accommodates varying levels of granularity. Revisiting the examples from above, for instance a specific performance of a Beethoven symphony is not knowledge in itself (just sound waves traveling around a room). But it might become knowledge if that specific performance turns out to have been (a) recorded and (b) important enough for that recording to be reproduced at least for some time.

Knowledge is extraordinarily powerful. For instance, it is entirely possible that we can use knowledge to get humanity past our solar system into outer space and eventually to new planets in other solar systems. There is nothing we have found so far that that makes this impossible per se. So knowledge of our solar system (and generated within our solar system) has the power to outlast the solar system!

On the downside of this power of knowledge (at least from humanity’s perspective), nothing precludes us from creating sentient machines that might have their own knowledge which they could continue to reproduce without us. Knowledge has already permitted us to create machines (e.g. nuclear bombs) that could end the existence of humanity. Incidentally, if humanity goes away before creating such indefinitely self replicating robots then all human knowledge will go back to being just information and slowly decay over time.

You may wonder why any of this matters. Well it matters to me, anyhow, because knowledge (in this definition) forms the underpinning of my personal answer to the question as to why we are here. More on that in future posts where I also promise to also revisit the relationship to free will.

* Gabriel Weinberg pointed out that some animals likely pass on tool use culturally (eg sea otters). I have not bolded the word “externalized” by which I mean through some kind of record that allows the knowledge to exist outside of the brain. In that I am pretty sure humans are unique.