3 ms·

We need a large number of samples when we lack sufficient understanding of the system in study. But if we assume our current knowledge of physics and biology ar

We need a large number of samples when we lack sufficient understanding of the system in study. But if we assume our current knowledge of physics and biology are good enough then we should be able to find the best(whatever definitions is used) without needing multiple samples. We use multiple samples to create a model but if a model exists then non-contradicting samples add nothing. Although we could have a large discussion about bayesian epistemology regarding my last statement.

throwaway_pdp09 6y ago
> But if we assume our current knowledge of physics and biology are good enough That's a bloody big if, but let us let it ride for now > we should be able to find the best(whatever definitions is used) Agreed it hinges on that, assuming the preceding if is accepted. What 'best' do you nominate? I have no idea what it might be.
- harshitaneja 6y ago
  It is a big if. My choice of words was very deliberate as I didn't wish to convey that our knowledge is sufficient. But the question here doesn't hinge on real knowledge(our ability to ever gain it is in itself a huge question) but rather on belief in possessing the said knowledge. Otherwise every real world calculation of probability is meaningless as physical systems can't have proofs but rather more and more tested models acting as the prior in our calculations. I don't nominate any best. I don't have an idea either. I merely used a word used by a parent. Best could be minima, maxima or any other criteria of the function they choose to refer here. My point is if you have a belief that your model is good enough, you don't need a large number of trials to calculate a probability on whatever characteristic you choose. The accuracy of this result is a question but the question would remain similar to the scenario where you calculate the probability from a larger number of trials.

throwaway_pdp09 6y ago

> But if we assume our current knowledge of physics and biology are good enough That's a bloody big if, but let us let it ride for now > we should be able to find the best(whatever definitions is used) Agreed it hinges on that, assuming the preceding if is accepted. What 'best' do you nominate? I have no idea what it might be.

harshitaneja 6y ago
It is a big if. My choice of words was very deliberate as I didn't wish to convey that our knowledge is sufficient. But the question here doesn't hinge on real knowledge(our ability to ever gain it is in itself a huge question) but rather on belief in possessing the said knowledge. Otherwise every real world calculation of probability is meaningless as physical systems can't have proofs but rather more and more tested models acting as the prior in our calculations. I don't nominate any best. I don't have an idea either. I merely used a word used by a parent. Best could be minima, maxima or any other criteria of the function they choose to refer here. My point is if you have a belief that your model is good enough, you don't need a large number of trials to calculate a probability on whatever characteristic you choose. The accuracy of this result is a question but the question would remain similar to the scenario where you calculate the probability from a larger number of trials.