From 46a672b287a1e74f6213462e80cf37714e04eb1d Mon Sep 17 00:00:00 2001 From: Fadi Shawki Date: Tue, 4 Jun 2024 16:22:36 +0200 Subject: [PATCH] Moving some paragraphs around --- .../routes/papers/2024.AUniversalLanguage.tsx | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/orbitmines.com/src/routes/papers/2024.AUniversalLanguage.tsx b/orbitmines.com/src/routes/papers/2024.AUniversalLanguage.tsx index b87fd74..8c6a815 100644 --- a/orbitmines.com/src/routes/papers/2024.AUniversalLanguage.tsx +++ b/orbitmines.com/src/routes/papers/2024.AUniversalLanguage.tsx @@ -246,6 +246,17 @@ const AUniversalLanguage = () => { But this introduces a rather hard problem, namely that: . +
+ Move elsewhere? + + A lot of this comes from the realization. That most differences come from the context in which they're applied. But this presents a problem of how one often uses languages: Specific names for specific perspectives. And that makes useful generalization quite hard. [REPHRASE] + +
+ + Essentially the problem becomes. When do you decide that a particular kind of perspective, or switch in perspective should have a different name associated with it. Essentially what we're asking, is: Why is it so important to name this differently? Would it be harder to find if one didn't do that? [REPHRASE] + + +
Part of any task then, becomes this: . This should somewhere be quite intuitive: You can use tools without knowing how to make those tools. Essentially wanting to understand unknowns, might as well be called reverse engineering: How is it done? What aspects of it can be { Something like: Constantly all the rays as functions are executing (in orbits), then if something causes something else's behavior to change, you get the inconsistencies. Particles are seemingly temporally stable orbits/modular structures? "Wrong dynamics" from a particular perspective, often probably still keep traversing - they still work. It's just that from the perspective you wanted, they don't. - Reprogrammability as foundational? + Reprogrammability & inconsistencies as foundational? + Invariances at start hence a modular structure might be a necessity physically
@@ -362,7 +374,7 @@ const AUniversalLanguage = () => {
- +
Version control, causal histories, theorem proving, ..., a (programming) language are all rather similar. Always we consider some sort of