4 ms·

To add to this: I think that what appears to us to be stagnation in scientific interest was due to the fact that Ptolemaios was so brilliant. Contrary to popula

To add to this: I think that what appears to us to be stagnation in scientific interest was due to the fact that Ptolemaios was so brilliant. Contrary to popular belief, the empirical quality of his cosmology in terms of predictability was not surpassed by Copernicus, but only by Kepler about 100 years later. There were some minor discrepancies, that bothered experts in the late middle ages, which let to Copernicus. But even he could not convincingly solve them. (In his theory the Sun is not at the center, but the mean Sun, as is the center of Ptolemaios deferent not exactly the Earth.) With Ptolemaios, however, cosmology had stabilised to such an extent that the fundamental questions had found their answers and astronomers turned their attention to practical issues and refinements, such as calendars and the related problem of the very odd movements of the moon. (You need Newton and gravitation to solves this, more or less.)

chairmansteve 8mo ago
I often wonder if Quantum Mechanics is a "Ptolomaic" understanding of the sub atomic world.
- srean 8mo ago
  Yes I have wondered about that. I feel the same about Attention based networks, may be we are not using the most befitting coordinate system to understand them. https://news.ycombinator.com/item?id=46544981 https://news.ycombinator.com/item?id=46544981
srean 8mo ago
Yes this exactly. The Ptolemaic system is an adhoc form of Fourier analysis (circles upon circles). These are universal approximators (another universal class are the neural networks) of (preferably smooth) periodic functions. So with enough epicycles one can fit any complicated periodic motion. But it will not help extrapolate the motion of Pluto from the motion of Jupiter, it's a data fit description, not a causal description. This is also why I worry about the recent trend of using DNNs in astronomy. We may go Ptolemaic with them.

chairmansteve 8mo ago

I often wonder if Quantum Mechanics is a "Ptolomaic" understanding of the sub atomic world.

srean 8mo ago
Yes I have wondered about that. I feel the same about Attention based networks, may be we are not using the most befitting coordinate system to understand them. https://news.ycombinator.com/item?id=46544981 https://news.ycombinator.com/item?id=46544981

srean 8mo ago

Yes this exactly. The Ptolemaic system is an adhoc form of Fourier analysis (circles upon circles). These are universal approximators (another universal class are the neural networks) of (preferably smooth) periodic functions. So with enough epicycles one can fit any complicated periodic motion. But it will not help extrapolate the motion of Pluto from the motion of Jupiter, it's a data fit description, not a causal description. This is also why I worry about the recent trend of using DNNs in astronomy. We may go Ptolemaic with them.