Degreebling 2/4: Our World Model
How Darwin got degreebled
We talked about greebling as a property of systems. Now let's take a look at logical, mathematical, belief systems - are they any different?
I say no, they have similar greebling - degreebling dynamics. It is a special case, though. It is the system quite close to us, even influencing our sensory perceptions. 'Epistemic greebling' is a fitting name, opposed to the greebling of the world at large ("noumenal greebling", aka complexity of meatspace).
That is necessary to cover as a supplement to the previous post, as @pee_zombie talked about greebleness as about an objective complexity AND also our intuitive representation.
In this post I'll explore that kind of greebling, and then its relation to noumenal greebling. The concerns are similar to the questions about correspondence theory of truth.
1 A different kind of greebling
Imagine you're on a steamer from London to New World, circa 1840. You're a botanist with Linnaeus' Species plantarum - first edition, 1753 in your bag. Your goal is to document new species of flora.
The steamer arrives at the island (picrel) and you take a boat to the beach. The shore is rocky, and soon sparse vegetation appears. It's more of a meadow than a jungle. You can see the forest wall in the distance.
Your work begins. You look around, considering what to start with. There are many plants here, and plenty of insects.
You squat to take a look at the first plant specimen that grows high to your knees.
It is a new species to your knowledge. You call it A. It has two new (to you) characteristics: the shape of leaves and color of the flower. That gives object 'plant A' a volume of complexity.
You are unsure where to place it in the taxonomic tree, perhaps a new family will be needed? You need answers.
You're skimming through the book, you see two similar plants - E and F. Linnaeus put them in taxon 'Alicedae'. E and F share one of the two characteristic features - the leaf shape. You notice the similarity, and mentally add A to Alicedae.
Plant A has less distinctiveness on its own now. Part of its distinctiveness was offloaded onto a class it belongs to. So you add a note on the margin describing the new species. You also draw a sketch of the leaves on a separate piece of paper, that you put into the book. Nothing changed in Nature, only in the knowledge state of you - the scientist.
You continue your journey through the island. Jungle grows thicker. To your surprise you find another specimen of plant A. It shouldn't be here. The canopy is too shady!
You look closer, and it's not A. It's B. It's a similar one, a cousin species, adapted to shade. You look around, and you notice more similar plants around it. You call it C - a small variant of the same leaf pattern, with sprawling roots, and D - a specimen with a slightly different color.
You scratch your head. Now we know that A, B, C and D are very similar. Maybe they should be in a new family, different from E and F's Alicedae?
With a quick move, you write 'Bobidae' on a new sheet of paper. Now you put A-D into that space.
What you did was effectively reduce the complexity of A representation. Number of its unique features has decreased. Before, in Alicedae the unique features was leaf shape and flower color, now in the new class - all sharing the leaf shape - only the flower color makes it unique.
That is degreebling. We now have a better intuition to see it happening, but how to talk about this?
2 Graphic representation
@pee_zombie describes degreebling with yet another metaphor:
> just like in Tetris, if you rotate your ideas just right and assemble them in the right order, so many of them seem to just vanish in a row
That is quite similar to the botany example above. New ideas - new blocks - new information about the species, created a new row - new taxon - where you discover an underlying similarity.
Continuing this taxonomical example, greebling was the creation of new leafs on the tree. Degreebling was putting them into more high-level abstractions. Raw variance captured in a systematic structure.
Having A-D in a separate group from E-F is more degreebled.
Now we can generalize from the botany example. If we portray knowledge as a tree graph, with foundations at the bottom, greebling would be the creation of many new leafs. Greebling metric could be the number of leafs. Or the number of clusters.
Note that here we take epistemic status as a sort of hypothetical impersonal writing down of ideas, not the actual ease of recall. The botanists would talk easily about those plants they had more experience with, regardless of how they are in the taxonomic model.
Original @pee_zombie 's tweet was more formal, and related to Markov processes, rather than classifications.
Let's go back to the real world now.
3 Correspondence theory
We notice a duality between a thing with some degree of greebling, and our knowledge of it as greebling.
These two could be entirely separate, uncorrelated. Yet the Kolmogorov equations attempt to describe the noumena, as Markov processes can have hidden states.
Given that, Noumenal Greebling is a metric we can assign to any world state, and it has a mathematical representation using Kolmogorov equations.
Now since our knowledge seems to be influenced in some way by the noumenal world, we can expect our epistemic greebling metric to be influenced by noumenal greebling.
That does not require a correspondence notion of truth. We can just use the go-to Elizer's cybernetic The Simple Truth.
We can also look at the problem from a biological perspective. Then it becomes about the Umwelt, the phenomenal qualities of organisms, and systems more broadly. Insects have Umwelt running at double the speed of ours.
What kind of abstractions do they use? Umwelt of highly optimized animals is smaller.
It's more degreebled, as the hardware is more optimized. Big animals with longer lifespans and lifecycle (hardware update speed) have a much higher chance of being forced into software updates.
To summarize, the relation of epistemic greebling - complexity of representations to noumenal greebling - complexity of meatspace - is complicated. Let's discuss some more cases.
4 Over and undergreebling
There are times where there is an explosion of new forms, adaptive radiation into new ecosystems. It is not limited to biology. The same dynamics work for browsers, NFT3 projects, or mammals after the K-T extinction. Adaptive radiation creates lots of noumenal greebling that epistemic greebling has trouble catching up with. It is a very peculiar state of the world, to be surrounded by new things you can't make sense of (yet). In the worst case, you'll get outcompeted.
We can call the theories during these times as undergreebled. They don't account for much of the complexity that's out there.
We can extrapolate this, and say that if your estimated future degreebling (sense making) will never cross the greebling that happens around you, you live in singularity.
And your possible future degreebling depends on your processing power. For modern humans that is neocortex + exocortex + hive minds.
Singularity is when you no longer have noumenal capacities to produce epistemic degreebling of the phenomenal data given to you by increased noumenal greebling. By this definition, all non-human animals on Earth have been living in singularity.
Comparing with Machine Learning, overgreebling is overfitting. You then need to make your model more abstract, more degreebled.
So is it better to out-degreeble with your memes - that is epistemically the greebleness of your Umwelt?
That's just Ockham's razor, right? There is a preference for simpler theories in science - Occham's razor is a pro-degreebling tool.
But have we not learnt from the times of Lovecraft, Einstein and Heisenberg? Does Reality comply to our efforts to understand it?
Ockham's razor is not a path to wisdom, but to folly. It is a COPE, a heuristics used by those that can't keep track of all the hypotheses and apply them instrumentally.
The very notion of 'scientific paradigm' is Longhouse-coded. The need for stability is alienating novel perspectives. The world of tomorrow - the world of AGI will be free from these chains.
Now what is over-degreebling, or overly abstract words? Lots of midwit, non-reproducible advice is like that. What happens to scientific theories when they are too abstract? What about religion?
Exoteric religion is degreebled, folk rituals are greebled. Reconstructionists are greebling, perennialists are degreebling. Superstitions are very specific, and often don't designate the force behind the effect.
Here we can distinguish that good scientific theories are more greebled than other knowledges. Their predictions are lower-level, more like superstition than religion. Religion is a behemoth with theology, soteriology - greebledness in the memes, but degreebled in the user-interface. Religions embracing tradition more, such as Catholicism compared with Protestantism, retain more of the low-level greebled user features.
This comparing of distinctions could be very long. We don't have much time. Maybe another day.
Conclusion
What we know now is how the degree of abstraction works for human knowledges, from religion to science. But that is a very dry, technical aspect of the matter.
What about our feelings internally? What about our feeling of meaning in life?
Does it prefer greebling or degreebling? That's what the next post is about.
Disclaimer: It is good to be aware of the difference between the two types of greebling, but later I do not meticulously stick to that convention, as they are correlated well enough in most cases.