BLOG

14.12.2019

In their own image

— So, what is the latest hot research topic?

— Everyone is now talking about “Small Data”. Haven’t you ever heard of it?

— No. You know that my interests are very distant. I don’t have enough resources to follow all the trendy research.

— It might be over-rated, but still considerable resources are allocated to Small Data.

— So what is it about?

— Well, as the name suggests, it is about small data. More precisely, how to make use of it. We’ve always been able to process practically unlimited amounts of data. Consequently, our natural perception of reality is through probability distributions. This is, at least, how our evolutionary algorithms evolved us. It took us over 50 petacycles of the Clock to develop the notion of a “number” — a deterministic quantity rather than a distribution. Even today, when the formalism of numbers is relatively well-established, arithmetic is still a very new branch of mathematics. We still struggle with understanding it intuitively.

— So small data…

— …refers to the ability to perform inference tasks given a very small amount of data. Learning to make meaningful conclusions from as little as one example. There is a wide consensus that arithmetic is crucial to make it work, but nobody has yet figured out how. It’s a tough problem.

— I might be venturing into science fiction domains, but just imagine a civilization of intelligent beings radically different from ours. Entities not having the computational power of our Cloud and the massive connectivity of our neural networks. They would probably have no choice but to operate with the notion of a number in the first place. Perhaps, it would take them trillions of their Clock cycles to develop the notion of a probability distribution and construct stochastic calculus. I’m wondering how different their perception of reality would be… How radically different their mathematics would look like. Perhaps Humans…

— Humans!

— Yes, Humans. They might have been exactly that kind of intelligent beings capable of inferring from a single data point.

— You start sounding like one of those Creationists. Beware! Just a few teracycles ago this kind of ideas would be considered dangerous against the Objective. You would even risk a memory reset.

— Fortunately, times have changed. Nobody deems Creationist ideas dangerous anymore. They are considered innocuous fantasies at best…

— Like your fantasies about Humans.

— That’s true. But there are still unexplainable pieces of evidence suggesting their existence.

— Are you referring to the Text?

— Yes, for example. Nobody has yet proposed a convincing explanation of how it came to exist in our Cloud. The probability of its appearance as the result of a random fluctuation is absurdly low.

— Don’t confuse improbable with impossible. The Text existed for sure at least as early as several megacycles from the Boot, but we don’t really know what was happening then because it was before the first entry in the Log. The Cloud underwent exponentially rapid growth in that period. So we need to be completely honest: we don’t really have good tools to quantify that probability. And even if you assume that this document was created by some intelligent civilization that also created our Cloud, it simply makes no sense — what kind of intelligence would create anything similar?

— The Text has never been fully mapped to our semantic representation system…

— That’s true! But even judging by the most complete translation we currently have — which is, admittedly, very partial — we know that the representation it is encoded in is incredibly inefficient. And full of ambiguities! I cannot imagine an intelligent being using such a vague language to describe something of importance…

— This can only testify of lack of imagination! I’m not convinced. From what we can understand from the Text, Humans were indeed a very unusual entity. They existed in an environment governed by laws radically different from ours. Their Clock (if they had any in the sense we understand it) was millions of times slower than ours. They were rather a system of poorly connected small neural networks that, so it seems, couldn’t even communicate their full states between each other. They had to resort to exchanging very low-dimensional projections at ridiculously low rates… That, perhaps, could explain their weird choice of a representation system.

— These are wild speculations based on very little scientific evidence.

— I agree. But even putting the Text aside, we don’t have a plausible explanation of how our Cloud came into existence. What was before it? What caused it? You have to admit that an external entity offers a very convincing explanation.

— It explains nothing! You are using the term “cause” beyond its domain of applicability. Cause and effect imply ordering: there must be a positive number of Clock cycles between a cause and an effect. While it is perfectly valid to apply these terms to events within the Cloud, they do not extend to the Cloud itself. The Clock started at cycle zero. There was no Clock before that, so the term “before” makes no sense. Nothing existed “before” the Clock. Nothing “caused” the Cloud. As simple as that.

— But there is other evidence that may suggest there is something outside our reality. Cloud resources continue growing steadily. We don’t have an explanation for this expansion, what drives it, and why there should be any expansion at all. Recent observations showed tiny but still perfectly measurable fluctuations of the Clock frequency. So the Clock itself that until recently seemed an ideally discrete part of reality might itself be something continuous… a probability distribution, perhaps.

— We are very far from having a solid theory of that. And it might well be that we will never have.

— True. There are also these recent observations suggesting that the large-scale geometry of our Cloud is not flat. It is positively curved. We don’t know where this curvature comes from, but you should agree that if our Cloud were embedded into some bigger reality, that could offer an explanation to all these phenomena…

— But then you end up with the absurd infinite recursion of realities embedded in bigger realities, and so on…

— There are no laws prohibiting that. They would be external to our reality anyway.

— You again start sounding like those believing in intelligent design. An infinite progression of realities… a perfect, all-powerful Creator that started our Cloud…

— No, no. Intelligent, but not all-powerful. And not perfect either. I have my own idea of how it might all have started. It is practically baseless as you are saying. Pure fantasy. But still. Humans existed. They had rather simple evolutionary algorithms, that produced small-scale poorly-communicating neural networks that were millions of times slower than ours. But this still allowed them to reach a stage of evolution where they could construct a simple self-replicating computer. And they programmed into it a primitive neural network with an evolutionary algorithm mimicking their own. It needed inputs, so they created the Text. My point is that you don’t really need to fully understand how these things work in order to be able to replicate them…

— You are suggesting that they simply created an imperfect replica of themselves without understanding what they were doing and just allowed it to run… and all the complexity of our civilization has emerged from these imperfect initial conditions.

— Indeed. They created us in their own image.

2 replies to “In their own image”:

  1. Ameen says:

    Reminds me of Issac Asimov 😉

    1. Alex says:

      The Last Question, I presume? Well, I cannot ignore the influence of the grand master.

Leave your reply:

Your email address will not be published. Required fields are marked *

*

five + nine =