Malcolm Ocean's roamblog

June 28th, 2024

edited 2024-06-28 · open in Roam
    • from Sahil >
      • > Second law: You cannot expect actual order (stable structure) to arise spontaneously. If you see it, there must have been a controlled process lurking behind the scenes which involves entropy production in the larger naturally unfolding context.
      • > An example would be: A priori, if I tossed a 100 coins and they all came up heads, or if all the gas molecules were repeatedly going to one corner of the room, you'd be surprised, and consider it a thermodynamic miracle. However, if you were more careful, you'd notice how this is an open system, and point out some systematic force that biases it towards a "low entropy" region, producing a bunch of heat in the process, in accordance with the Second Law over the whole system.
      • > Cosecond law: You cannot expect deep caring (stable commitment) to arise via control/force. If you see it, there must have been organicness/freedom lurking behind the scenes which involves connection, even in contingently-arranged controlled situations.
      • > An example: You and I might have met in order to work on a project; our relationship might be purely instrumental and it would be surprising if it extended beyond into true love. But not really: it's possible that we have liberated interactions within that container that allow back and forth openness and integration of our more personal sides, which is in line with the cosecond law.
    • MS > I directly perceive something about the 2nd law of thermodynamics
      • I don't believe it
      • (struck that I have some sense of this wrt 2nd law, but I feel like insofar as I've traced the proof it's pretty loose)
    • Sahil > ergodicity, minus-first-law
    • Sahil > causal bias in our understandings
      • science, progress
      • "gears" understanding
      • reductionism
    • bayes theorem
      • likelihoods describe things that are compact, causal, modularly defined
        • very good for atomistic, causally-defined things
      • posteriors are complex, depend on lots of factors
    • duals
      • LCM <=> GCF
      • cartesian ? sets <=> disjoint union sets
      • MO > same fact-ness? like integrals & differentials
      • S > yes, theorems for free
    • S > can't predict what chess grandmaster will do, but can predict he'll win
      • MS > but that's a statistical thing
      • (I think that that's still viewing it in the causal frame. in the telic frame it looks more like "no, it's because he wants to win and can")
    • 2 claims:
      • is it possible to dualize?
      • if so, what does it look like?
      • (but there's a choice of dual, depending on how you structure the first category)
    • S > thesis by Yoshihiro Maruyama
    • we're very good with likelihoods, not so much colikelihoods (which involve stable spirit, commitment, etc... rather than stable substrate)