1.4 Two philosophical methods
1.4.1 Philosophy consists of updating the highest-level concepts of the mind. As discussed above, this ‘updating’ process can ultimately involve anything in the mind. That said, when undertaking philosophy as an intentional project, we can distinguish two methods. These are not opposed strategies, but complements; doing philosophy well may require both.
1.4.1.1 Firstly, we can try to draw out connections among existing concepts, showing, e.g., that they contradict each other, or that two seemingly different concepts can be unified. This is what is usually called ‘philosophy’.
1.4.1.2 Secondly, we can can learn things, or have experiences, which are especially likely to help with the updating of the highest-level concepts.
1.4.1.2.1 Learning new things can be useful for several reasons. They could refute an existing high-level concept; they could suggest a common pattern; or they could give a model for how to compress ones’ existing knowledge, even if this was already possible earlier.
1.4.2 If we are aiming to follow the second strategy, what sort of experiences or concepts are most useful to have? This is a difficult thing to determine since our goal is to update the highest-level concepts, so we can’t choose on the basis of a yet-higher concept.
1.4.2.1 Given this, one strategy might be to study the areas which we are currently maximally uncertain about. Another might be to try to learn as many novel ideas as possible.
1.4.2.2 But we should not choose the above areas or ideas purely at random; rather, the areas should be chosen based on how near they are to our current conceptions of the highest-level notions of reality, goodness, etc.
1.4.3 Given the prominence of science and mathematics in determining our world-view, studying these areas seems like a natural choice. This will be the strategy we pursue here. Thus it seems worth saying a bit about how I think the epistemic process works in these domains.
1.5 Math & Science: flywheels of modernity
1.5.1 Modernity has seen an explosion of our scientific and mathematical knowledge. The core mechanism behind this can be described as follows: virtuous feedback loops between conceptual progress and empirical/calculational flywheels.
1.5.2 What is a empirical/calculational flywheel? By this I mean a set of well-defined procedures and technologies which enable one to pose precise questions and problems about some aspect of physical or mathematical reality.
1.5.2.1 As an example, in astronomy sextants enable precise records of the position of objects in the sky. This allowed Tycho Brahe to record the motion of the planets and led to Kepler’s laws.
1.5.2.2 In biology, empirical flywheels include Mendelian breeding experiments. Fields can also provide empirical tools to be used in another, as in the X-Ray diffraction experiments which revealed the structure of DNA.
1.5.2.3 In mathematics, the distinction between “empirical” and “conceptual” research is not as clear. Roughly speaking, explicit calculations and proofs using a given set of definitions could be considered a “calculational flywheel”.
1.5.3 The full process of science consists of an alternation between discovery/work within one of these flywheels, and developing new high-level concepts on the basis of the results obtained from doing this. These new concepts in turn suggest new ideas for experimental paradigms, technologies, and calculational techniques.
1.5.3.1 These feedback loops are themselves enmeshed in larger loops of engineering technology, political change, etc. which define modernity in full.
1.5.4 Notably, the concepts discovered in the course of investigating a given flywheel can be surprising. The process of investigation takes on a life of its own and shows you things you didn’t know would be there.
1.5.4.1 An obvious example is quantum mechanics, a theory whose conceptual interpretation is still not clear to this day, despite being ubiquitously and successfully applied. QM was initially as a solution to technical problems in the theory of radiation.
1.5.4.2 It might seem that this would be less the case in mathematics. But in fact calculations are constantly giving rise to surprising results that naturally suggest radical new concepts. Take complex numbers. It’s natural to try to find roots of cubic and higher order equations: this leads to the introduction of the imaginary unit i. Extending differentiability to this new number system gives rise to the amazingly rich holomorphic functions.
1.5.4.2.1 Mathematicians’ sense of the richness of an area largely consists of the extent to which investigating it tends to lead to surprising new concepts “naturally” coming up in the course of an investigation.
1.5.5 These examples suggest to me the mental image: when investigating, do not merely doggedly pursue the answer(but do pursue it, to be clear). When stuck, let the phenomena take you by the hand and guide you to the correct concepts.
1.5.6 The above “virtuous circle” between conceptual analysis and experimentation bears some resemblance to my description of ideal philosophy. Indeed, I think the activities are similar.
1.5.7 My project in this text is to attempt to do philosophy—that is develop new concepts describing very high-level structures of the world, and our existence within it—by examining the mathematical structures discovered by physicists.
1.5.7.1 Relative to this lofty goal, I have only really succeeded at producing a sketch of some concepts which might be a useful description of the world.
1.5.7.2 As a secondary goal, I wish to introduce some new “flywheels” that might be especially productive in generating new concepts in this area.
Figure: Philosopher guided toward correct concepts by empirical flywheels
1.6* The Usefulness of Philosophy
1.6.0 You might be thinking: “That’s all well and good, but who cares about having good high-level concepts? Isn’t it basically useless? Philosophers aren’t exactly super effective at getting stuff done in the world.”
1.6.0.1 (If you’re not thinking that feel free to skip to the next section where I actually start talking about physics)
1.6.1 I think that having good high-level concepts can be very impactful. In slogan form: “philosophy helps you choose a good direction to move in; doing a good job at moving in a direction depends on attention to detail and work ethic. But a good direction can be very impactful”.
1.6.2 Let’s consider some examples.
1.6.2.1 The nature of the mind. Materialism posits that the mind is instantiated by processes in the brain; combined with the Church-Turing thesis, this implies that making an artificial mind that runs on a computer should be possible. Thinking through the consequences of this naturally implies that AI development is a very impactful area to study. These considerations motivated people to work in AI decades before LLM-like systems made it obvious that such a thing is possible.
1.6.2.2 The nature of political power. European societies in early modernity believed that political authority was held by royalty, derived from natural law or divine sanction. Social contract theory instead viewed political power as the result of an explicit or implicit negotiation between people; this contributed to the revolutions in Europe and America in 17th-19th centuries.
1.6.2.3 The nature of money. From where does money derive its value? One answer: it is an arbitrary yet consistent fixed-point in valuing-space, emergently useful for coordination. This—in contrast to other answers such as that money’s value derives from the state—reveals that something like bitcoin should be possible.
1.6.3 Good highest-level concepts are not so different from low-level concepts in this regard; both should enable better understanding of, and action in, the world.
1.6.4 Can the world be fully grasped through concepts? Can we fully analyze the redness of red, the feeling of a late-summer breeze, yearning and nostalgia? Or even what it’s like to go swimming, or fix an old bike, or use a word processor? Perhaps not.
1.6.5 And yet, taking a step back, I often find it remarkable how well abstractions I read and pondered out of intellectual curiosity, have turned out to describe in miniature the great arc of history, the pattern of peoples’ lives, real opportunities and tragedies.
1.6.6 Thus I adopt philosophical analysis not as a game, but out of belief that the world rewards careful attention to its structure and the conviction to act upon it.
1.6.6.1 In particular: as mentioned in the introduction, I believe that AGI might come soon. If this is the case, a theory of abstraction and the mind could be key to influencing its trajectory. I believe that the ideas outlined here are very promising as paths to such a theory.
You could also go back a step: both “the state” and “social valuation for coordination purposes” reveal that fiat currency (even unbacked by any material goods) could work, in contrast to the view that the currency itself must be precious.
this is a very mindy explanation of philosophy, putting the “high level concepts” front and center. I’d like to frame philosophical undertaking in a way that highlights practical applications and consequences.
most of my personal pradigm shifts have added extra dimensions to lived experience by highlighting newly salient features. So it’s been less of looking up known-unknowns, but rather stumbling on unknown-unknowns
What do you think of my discussion of usefulness in 1.6? I agree my discussion is mindy but I feel like all philosophy is mindy(even the purportedly anti-mindy variety)
I think you’re conflating “thinking” or “developing mental models” with “doing philosophy”
In my mind, philosophical approaches tend to be bad. Extrapolation across incommensurable reference classes, doubling down on reifications, specialising in domains that are both abstract and ill-defined (not due to lack of trying!)
More on this: https://meaningness.substack.com/p/undoing-philosophy
Also, once you disentagle “philosophy” and “developing mental models”, it’s quite clear to me that mental models and empirics go hand in hand, in an interpenetrating fashion. Claiming progress in the name of philosophy seems more like stealing credit. And no, I’m not saying that no philosopher has ever come up with a fruitful frame; I think this happens in the rare cases where philosophy meets enough natural invariants through empirics; though I do claim that the people into philosophy likely would have contributed more if not infected by philosophical antipatterns.
I mean, have you ever been to a philosophical cafe? Or talked to someone very into either hegel or analytic philosophy? Brrrr
“Flywheel” seems like the most interesting concept here.
Does it mean something like “a well-defined way to measure something”, “representation/subset of the problem with a smaller state-space”? For example, integers is a flywheel because it’s a well-defined way to measure some (but not all) properties of objects. Thermometer is a flywheel because it’s a well-defined way to measure some (but not all) properties of the body.
Any deep discovery combines multiple very different flywheels into one. (Maybe. I’m just spitballing.) Like when Descartes discovered that two number lines define a plane (in such a way that many shapes have simple equations). Or like when the concept of a group was introduced, combining symmetries and something number-like. Or like when Newton realized that the same thing can explain both orbiting motion and falling/throwing.
Is this aligned with your thinking?
I think of a flywheel as being something that is easy to iterate on and get unambiguous results. So yes, basically “well defined measurements” + “smaller state-space”.
Combining two flywheels into one...hmm, maybe. I would be a bit more general and say that deep discoveries compress a lot of your worldview, which could include uniting two small well-defined domains(or showing a particular small domain well predicts a bigger domain)
I guess I wanted to ask… in what direction do you want to take the concept?
You introduce the concept of a flywheel. It’s connected to “feedback loops”, “technology”, “questions”, “problems”… but all of those auxiliary concepts are pretty ugly. I mean from a mathematical/ontological perspective. Because they are pretty complicated and contingent.
So one way you could develop the concept is to “purify” it, simplify it as much as possible, make it as ontologically fundamental as possible. Define what a “flywheel” is independently from feedback loops / technology / questions, or in a way which generalizes all those auxiliary concepts.
I mean one direction you could take the concept of flywheels in is to try to define “(deep) knowledge” in terms of flywheels. Or something like that.
But the two ideas above are not the only valid ways to develop the idea of flywheels. Maybe you want to do something entirely different with them. Or maybe they are just not an important enough concept to develop like this.
I actually think of “questions” and “problems” as being fairly neat concepts. “Question” just being some thing you can query about the system, like “what is the position of Mars tonight?”, “what is the regulator of this elliptic curve?”, whereas “problems” would be searches over easily verifiable structures, like “does there exist a proof of this conjecture?”, “does there exist a way of synthesizing such and such a chemical element?”
Regarding ontology...I think of flywheels as being small, toy domains which give unusually clear, unambiguous answers, which can aid in the construction of theories applicable to messier domains. “High resolution” parts of reality, if you will. Take the movement of the planets: the laws of motion can be seen more easily when separated from the vagaries of Earthly existence such as air drag. Or particle accelerators are literally higher resolution than most things(they probe higher energy levels ~ smaller scales). This is connected to my overall schema of hot VS cold things which I’ll get into later.
Makes sense. IIRC many abstraction researchers (like Sam Eisenstat or John Wentworth) do the same.
One objection though: your definition doesn’t define what is an interesting question/problem.
So domains can be bigger/smaller (in terms of state space), harder/easier (in terms of computational complexity) or messier/clearer (in terms of measurement error and other problems with result interpretation). The last distinction seems less fundamental.[1] Do you really need to focus on it?
Separate question: would you say that e.g. Gettier cases are a toy subdomain of philosophy, allowing to test different definitions of knowledge?
(I’m trying to probe your conceptualization of flywheels in different ways to maybe help you write down some thoughts or inspire new ideas. If it doesn’t help feel free to say so or just ignore this message.)
“Messiness” seems to pop up only when you try to interpret domain A through domain B, but you’re confused about how to do it (e.g. you don’t know statistics) or lack capabilities to do it (e.g. you don’t have precise enough instruments).
Hmmm, so I think of messiness as being like...essentially, since the world is complex, the best models of most phenomena have many parameters.
But this means that, for long-running disagreements between world-models, most pieces of evidence won’t be able to resolve them. Because each world-model will have a lot of free parameters to interpret a new piece of evidence in any given way. This can be especially bad for controversial topics since people will engage in motivated reasoning, which there’s plenty of room for due to the abundance of free parameters. It’s also computationally expensive.
Thus to distinguish between world models, it’s often actually easier if the data comes from a small, artificially restricted domain so that there’s less wiggle room.
As an example, consider the “pots VS peoples” theories of the indo-europeans. That was definitively resolved by DNA data. Though people had plenty of archaeological data, it was not as disambiguating.
I think of interesting problems as being those whose answers (you think will) most help you disambiguate your global world model.
Gettier cases....yes there’s some similarity in examining an edge case. Lots of philosophical thought experiments are like that(but to be honest I never really understood the intuition that the weird JTB cases were not “knowledge” or why this distinction was important)
Disambiguating world models is a good motivation to focus on messiness/clarity.
Unexpected and interesting definition idea. Will look forward to future posts.
Flywheels are historical devices where weaights was put on strings around a spinning cylinder, adding momentum. Metaphorically, it points at a feedback loop that “gets going”, ie producing excess output, in a way accumulating some kind of capital.
When talking in terms of science, you could say that the invention of a new frame of reference unlocks a fair bit of arbitrage/early adopter dynamics. So the frame of references that keeps popping up serve as our swirling rocks, adding inertia to the axis of progress