/sci/ - Science & Math

>>11165013
>>11166829
Am I the only one who finds it fascinating when this board discusses simple concepts but tries to take an entirely new approach, almost as if it is dipping into the realm of schizophrenia real fast and pulls back a second later with the result? I find it fascinating because no other place does this and everything else in the world is focused on telescoping precision, rigorously following the mainstream scientific rules and doubling down on what is established so far, or what I have internalized in my personal model as "low learning rate" if we equalize this to a neural network, whereas /sci/ has a high learning rate - it has a high chance to come up with completely disconnected shit but on the other hand a very small portion of that disconnected shit will be legitimate insights that are the result of the high learning rate trying to unstuck us from the faulty local minimum that society was telescoping into for infinite precision. Like for an example this autistic triangle, what if the whole notion of enforcing the Euclidean norm, something that anyone knowing basic highschool math is aware of, is actually faulty and we can in fact have imaginary lengths? Of course every single professor would reflexively tell you to stop spouting schizo bullshit and to follow the rules established by academia, but was following these same rules the path that led us to canonizing the usage of negative numbers under a root, something that 100% of academia would laugh in your face for attempting to do in the 16th century? This is the precise thing that I mean to say with low/high learning rate, a low rate is great for going into problems for precision but what if the minimum we're going towards is actually the local one and not the true global one? In that case, the only way out is "delusional" high learning rate that re-examines and challenges the very basics of the established dogma.