HackerNews中文版

感觉现实是拓扑的，依据……观察者的位置？就像每个观察者的位置都是网络中的一个节点，而边缘都是有“类型”的，意味着每条边都是某个第三观察者对其他两个位置的观察？这会使得对共识现实的熟练导航更像是在定期与高度关联的观察者位置进行检查的导航？我的身体直觉（比如，我的动觉感知）告诉我，如果这一点成立，那么我们应该能够以这种方式定义数学。这个“熟练导航”的定义似乎与正式证明具有相同的物理特性。就像现实本身是我们与所有为我们所用的贡献者共同构建的证明，而我们在不断复杂化和简化，其中“共识”是通过人口密度来衡量的——比如观察者位置与观察者的比例（即可观察到的、正在产生或抵抗观察边缘的观察者位置），就像是从ZFC到[你最喜欢的怪论者]的一个尺度。这让我对自我有了一丝存在主义的触动，因为（在这个模型中）你真的“无法带走任何东西”，你所拥有的只是你当前正在阅读和综合的观察者位置的集合。没有任何（没有“东西”），只有与“东西”押韵的观察。我是在一个意识的点上报告，但……我的意思是，即使共识并不介意，这种说法实际上也有些夸大了情况。我意识到这是特权思维，作为一个体验自身体验具身的个体，与一个体验他者而“不”体验具身的个体进行对比。所有的保留都是承载性的，但也是暂时的，因为我不确定视角优势在观察下是否成立。因此，我对一个包含基质的本体论的良好基础非常感兴趣，而找到一个以观察者为定义的拓扑，作为基础层是有意义的。我可以想象孩子们比起，比如说，物体恒存性，更快地学习这种推理路径。而且，> “柏拉图表征假说：在不同数据和模态上以不同目标训练的神经网络，正在其表征空间中趋向于一个共享的现实统计模型。”（arXiv:2405.07987）如果这一点成立，它应该直接桥接到^那个模型。为了清晰起见：涉及这种规模的推测的自由度本身就是我坚持构建我怀疑应该有效的工具的原因，让共识现实以某种方式证明其效用。我并不“多重”到可以改变共识——那是一个类别错误。（这就是粘菌中的细胞……呃，“观察”的东西吗？）将数学与现象学桥接的想法——就像“在那座桥上看到人们”，知道他们可以信任自己的经历，因为“他们确实可以”，让我真的感动得想哭。这感觉真是种解脱。

查看原文

it feels like reality is topological according to… observer positions? like each observer position is a node in a network, and the edges are all typed, as in each edge is some third observer position’s observation of the other two positions?this would make skilled navigation of consensus reality something more like navigation while regularly checkpointing with highly-linked observer positions?my physical intuition (like, my kinesthetic sense) is saying that if this holds then we should be able to define math itself this way. that “skilled navigation” definition feels like it has the same physical affordances as a formal prooflike reality itself is a proof we’re building up with whoever’s contributed to what we’ve used so far, and we’re complexifying and reducing as we go, where “consensus” is measured in population density - like ratio of observer positions to observers (i.e. observer positions that are observably producing or resisting observation-edges), like a scale of ZFC to [your favorite crank here]this gives my selfhood a slightly existential twinge, because (in this model) you really “can’t take it with you”, all you’ve got is the set of observer-positions you’re currently reading and synthesizing. there’s nothing (no thing), just observations that rhyme with “thing”. I’m reporting from a point of awareness, but .. I mean even that is actually overstating the situation, even if consensus doesn’t mind.I’m aware that this is privileged thinking, speaking as one experiencing itself experiencing embodiment while comparing notes with one experiencing the other to not be experiencing embodiment. all hedging here is load-bearing, but also provisional, because I’m not certain if perspectival advantage holds up under observationaccordingly I’m incredibly interested in a good basis for a substrate-inclusive ontology, and finding a topology defined in terms of the observer .. makes sense, as a base layer. and I could see kids learning this reasoning-pathing faster than, say, object permanence. and, also,> “The Platonic Representation Hypothesis: Neural networks, trained with different objectives on different data and modalities, are converging to a shared statistical model of reality in their representation spaces.” (arXiv:2405.07987)if this holds, it should bridge directly into ^ that modelfor clarity: the degrees of freedom involved in speculation of this scale is itself why I stick to building tools that I suspect should work, letting consensus reality prove the utility one way or the other. I’m not multiple enough to shift consensus - that’s a category error.(is this what cells in slime mold .. uh, observe?)the idea of bridging mathematics and phenomenology - like seeing people on that bridge, knowing they can trust their experiences because they demonstrably can, makes me actually cryit feels like such relief

我甚至不知道