Latest revision |
Your text |
Line 14: |
Line 14: |
| | | |
| Isn't there something about knowledge being true information arrived at by logically sound reasoning? This meets the first criteria but not the second. [[User:RegularSizedGuy|RegularSizedGuy]] ([[User talk:RegularSizedGuy|talk]]) 17:31, 25 February 2024 (UTC) | | Isn't there something about knowledge being true information arrived at by logically sound reasoning? This meets the first criteria but not the second. [[User:RegularSizedGuy|RegularSizedGuy]] ([[User talk:RegularSizedGuy|talk]]) 17:31, 25 February 2024 (UTC) |
− |
| |
− | I feel there is an additional explanation that White Hat did not intend. The Sun and Earth, the entire Solar system for that matter, orbits the center of the Milky Way Galaxy. Which may also orbit the center of the known Universe? I am not an astrophysicist or knowledgeable enough to attempt a proper explanation. [[User:Vampire|Vampire]] ([[User talk:Vampire|talk]]) 15:31, 26 February 2024 (UTC)
| |
− |
| |
− | "The Earth-Moon barycenter is located approximately ¾ of the way from Earth's center of mass to its surface, towards the Moon's center of mass. Our tidal bulges (oceanic and otherwise) occur along that line. One bulge is towards the Moon because of the gravitational attraction, and the other is in the opposite direction, by centrifugal force from the Earth's rotation around the barycenter." This is wrong for a couple of reasons. First, both bulges are the result of the same effect, and they would be there even if the Moon and Earth weren't producing centrifugal force by rotating about each other. In the usual explanation of the opposite bulges, you look at the acceleration of a particle on the near side of the Earth towards the Moon by more, the acceleration of the center of the Earth by a medium amount, and the acceleration of a particle on the far side of the Earth by less. Then, to look at things in the frame of reference of the center of the Earth, you subtract the center's acceleration, and find that the near side accelerates toward the Moon by a little, and the far side accelerates away from the Moon by a little. But even that explanation is wrong, or at least, very incomplete - the main driver of the tides is due to the vastly larger volume of water *away* from the line through the Earth's and Moon's centers; when you do that same vector subtraction of the local acceleration from the Moon minus the acceleration of the center of the Earth, you find that you get a tangential component of acceleration, and since water can flow, it does, until it reaches a surface of constant potential (it's not a lot of distance, but it's a lot of volume moving, so the tidal bulge is a significant volume of water). See https://web.archive.org/web/20220115060446/https://www.lockhaven.edu/~dsimanek/scenario/tides.htm [[Special:Contributions/172.70.110.30|172.70.110.30]] 15:46, 26 February 2024 (UTC)
| |
− |
| |
− | Ok, so regarding [https://www.explainxkcd.com/wiki/index.php?title=2898:_Orbital_Argument&diff=336067&oldid=336066 this argument], the problem is removing the clarification that this is not subject to consensus 'averaging'. Two totally different opinions which cannot be averaged, merged, subsampled or intermingled. It is maybe useful to mention taking multiple weather predictions and generating the most supported trend, much as natural language processing algorithms, but here the two statements cannot be combined in simple numerical or tokenwise ways (that is the point). Yes, state that mid-point estimations are useful (I'm happy with such a statement, and preserved/enhanced it), but do not remove the salient issue that ''in this case'' it is not a useful process. It's beyond even [[2893: Sphere Tastiness]] illogic. '''Which is the joke'''.... [[Special:Contributions/172.71.242.55|172.71.242.55]] 18:51, 28 February 2024 (UTC)
| |
− |
| |
− | Re: expert averaging, it's non-intuitive to consider that different symbolic statements in text can be averaged, but with a neural predictive model (NN, human brain) this actually is possible. You *can* average the latent vector representations of two text inputs (x,x') in the (vectorized) latent space of a neural model (z + z'). Latent averaging is often used in ML as an empirical heuristic to improve performance (sentence embeddings, mixture-of-experts models), and can be hypothesized to operate via the mechanism of improving the efficiency of a learned Bayesian circuit that performs abductive reasoning. Averaging is more obviously seen in the output token space, for example, you might symbolically average temperatures from two weather models (formalized as ensemble models, consensus models, etc.). So although it sounds weird at first glance, averaging experts in either latent or manifest space is often a good heuristic for a guess, and is rewarded as a result. Arguably, White Hat is using this algo and actually making a good guess here, although if he can't explain his thought process symbolically (he's just doing it because it feels "nice"), his accuracy may come off as an "accident." Caveats: there are definitely conditions under which averaging experts can go awry in both latent and manifest spaces (false balance, non-expert data, partial observability, etc.), but this arguably isn't the case in either the orbital or wave-particle initial observations.
| |
− | [[Special:Contributions/108.162.242.37|108.162.242.37]] 01:21, 29 February 2024 (UTC)
| |