No internet connection
  1. Home
  2. Inexact Sciences

Ontology is what fills in the details

By hazard
    2021-02-15 23:22:10.733Z

    (I Was getting bored and losing motivation writing this, so I switched to a more brutalist writing style to get it out.
    This was prompted by a convo with @snav the other week)

    Ontologies are sort like programming languages for concepts, but with a lot more cultural baggage.

    Ontologies aren't true or false in the same way that programming languages aren't true or false. They just aren't propositions.

    Any ontology that passes a meager threshold of complexity can be used to simulate any other ontology, just like how any programming language that's Turing Complete can simulate any other language.

    This doesn't mean that your choice of ontology doesn't matter. You can tell me choice of programming language doesn't matter after you write an operating system in brainfuck. Different ontologies can express different ideas more readily. If this inclines you to go on a quest for "the most powerful/expressive ontology", watch yourself and read gwern's take on this.

    The affordances of a programming language are governed by the particulars of how you think, and the reality of the compiler. The affordances of an ontology are governed by the particulars of how you think, and all the cultural baggage/association/implicit-cruft.

    One reason I don't find a Cartesian dualist ontology useful is it obscures to me the way that the mind and body interact. It's harder to see why working out might help with your depression. It's harder to see how internal conflict can sap you of physical strength. You could make it do that, but it's not the direction that the learned associations push you.

    Sometimes I hear materialism or physicalism brought out in opposition to things like souls, spirits, psychics, etc. And yet the literal face value version of either ontology doesn't forbid such things. You could have a soul made of matter. James Randi did some really great stuff promoting. Was he ever building evidence in favor of the materialist ontology? No.

    If you were to catalogue every last assumption and association that's active in someone's mind when they reason with and talk about the material ontology... what you'd have would be something much more than an ontology.

    A trick that I really enjoy is the way that both consequentialism and deontology can simulate the other. You do it in a way that completely violates the "spirit" of either. You really have to clarify your concepts and articulate your assumptions before you make two things which can't simulate the other.

    That task is fun and righteous work. Over the summer I had my eye on the difference between reductionist and hollist "the whole is great than the sum of its parts" ontologies. Might have a post on that later.

    • 5 replies
    1. crispy
        2021-02-16 07:11:45.697Z

        Love this general direction. A lot to say, but my mind is fried so here are two thoughts:

        A trick that I really enjoy is the way that both consequentialism and deontology can simulate the other. You do it in a way that completely violates the "spirit" of either. You really have to clarify your concepts and articulate your assumptions before you make two things which can't simulate the other.

        This is amazing and is, exactly, the concept of a reduction math, especially complexity theory (as I'm guessing you know). But what if we took this further? What I personally want out of conceptual logistics (see Concepts Are Tools, Not Artifacts) is the ability to break the same thing down multiple ways and see which components are useful. What if the defining element of this game is reductions of the one you've described? I've had the same thought re: consquentalism and deontology, and this need for spirit obviously connects with @suspendedreason's surrogation. But there's something more still, what if we can allow "the spirit" of something to be a first class property, and getting at various "levels" of the spirit describes on what level you're copying the idea "as it was in it's context" and what things you're letting loose? This is the beginning of my thinking about metonymic robustness, the idea that a given conceptual tool has various holes it allows to be filled or not, but on viewing that tool for the first time it's not clear which holes allow you to fill them which ways. I know this all a bit too abstract, but I figured I better get this down and come back to it later or let it perish.

        I also love this parallel/depiction of "freedom of expression within a given representation" this seems key, and it seems like most actual conceptual work is just putting people in a place where they always understand what's going on. Every really good proof of a theorem looks like that, where the final few steps seem hilariously obvious. But can we use this engine in reverse and discover what the "contextualizers" are?

        1. In reply tohazard:
          suspendedreason
            2021-02-16 08:33:03.384Z

            @crispy Can you clue me in to some reads/launch points for this reduction math/complexity theory stuff? I feel like I have a vague sense of what you mean but no relation to a discourse. And unpack "first class property" for me too? Same reason.

            1. crispy
                2021-02-17 04:32:59.043Z

                This is on OK introduction: https://youtu.be/n0zd5hcOSQI but the truth is most of it is just very mathy, so the question is how much you want to go into it. I have the course notes for a complexity theory class from a while ago somewhere, and some other pointers I can scrounge. If you want to get into it, let me know and I'll gather them up.

                Re: first class property, I'm referring here to the fact that in many languages there are types of data that can't be passed around e.g. many languages don't allow a variable to contain a class (in the object-oriented sense) so classes aren't first-class. In this case, I'm saying that instead of saying "the spirit" is just intangible and hard to get to, can we treat it as something real, but something that so far we've only managed to define negatively? Can we bifurcate different parts of the spirit (e.g. intentionalism vs. cultural interpretation elements?) in order to define it. Let's treat spirit as a real object, instead of a ghostly "who knows"

                1. hazard
                    2021-02-17 14:26:58.732Z

                    Ah, dope, I didn't get what you meant by treating "spirit as first class" but now I'm onboard.

                    I think this is also a super important move for us to make as a small group, because I think most of the divergence between spirit and letter is due to adversarial action. It's not that James Harden can't understand the mysterious and ineffable thing which is "the spirit of basketball", it's that he found a hack in the letter, saw potential for huge gain for him and crew, and took it.

                    I do still think there is some baseline difficulty in syncing up around a spirit, but it's the thing that can be overcome with shared reading and lot's of conversations. Which is exactly what we do. So yes, I'm in huge support of treating spirit as first class.

                2. In reply tohazard:
                  hazard
                    2021-02-16 13:07:13.912Z

                    This is the general wikipedia link. Say you want to solve a particular problem with primes, but all you've got is an algo that solves a particular graph theory problem. You'd hope there'd be an "efficient reduction" of the prime problem to the graph one, i.e a fast algo that turns the prime problem into a graph problem.

                    You can always do such a reduction, but often the reduction will be just as hard as solving the original problem. Normally people hunt for reductions that are polynomial time.