4 ms·
Except that's unnecessarily inefficient and still incomplete, since it doesn't tell you anything about what to do with multiple games that predict the same symb
by naasking 9y ago
Except that's unnecessarily inefficient and still incomplete, since it doesn't tell you anything about what to do with multiple games that predict the same symbols, about what sort of games you should focus more effort on, what should you do with games that don't produce the right symbols, why you should or shouldn't modify game rules that fail such tests so they pass such tests, and more.
Philosophy informs these questions. For instance, a scientific process modelled on Solomonoff induction would give you more insight on what kind of games you should set up and why, which games are better than others for the same symbol outputs and why, to what extend modifying a game is a valid approach, etc.
- whatshisface 9y agoInsights on how to write down symbol games that match sensor readings would come from trying to do that, and gaining experience, would they not? It has an empirical kernel: whether or not you are successful at writing down good symbol games. That might technically be classified as philosophy, but since "the art of writing down good symbol games" is clearly founded in a task where you can easily evaluate your success, it strikes me as having more in common with other bodies of knowledge that grew around goal achievement than it does with philosophy. If you have multiple matching symbol games, maybe just either prove them mathematically equivalent (a la matrix mechanics and Schrodinger's equation), or if they aren't equivalent, write them both down and keep an eye on them until you develop an experiment that can distinguish between them. As far as I can see, that remains within the paradigm of "go forth and match the symbols."
- naasking 9y ago> Insights on how to write down symbol games that match sensor readings would come from trying to do that, and gaining experience, would they not? But your rules provide no guidance to do so, which would require philosophy. > As far as I can see, that remains within the paradigm of "go forth and match the symbols." Absolutely, I'm not suggesting changing the spirit of the basic empirical game you outline, merely that a different set or a few additional rules would permit you to generate better games more efficiently. Blind trial and error would eventually work too, but we ideally want to converge on accuracy and precision as fast as possible. What this means in practice is a philosophical question worth answering.
- whatshisface 9y ago>merely that a different set or a few additional rules would permit you to generate better games more efficiently. It seems suspicious that this is classified as philosophy - it has an empirical foundation! You can even design experiments to test these "theories of metascience:" just check to see how fast the scientists that work according to them produce successful theories. So it sounds like what you're advocating for is a human science (i.e. very difficult to design experiments, but fundamentally based on empiricism), as opposed to a branch of philosophy. Roughly speaking, the division I am making here goes: experiments plus arguments makes science, arguments without experiments makes philosophy. In this context, I can say, "scientists are only (professionally) interested in experiments and therefore uninterested in philosophy." However if something involves empiricism but is also labeled philosophy, that's just language and I gladly concede that scientists might be professionally interested in it.
- naasking 9y ago> It seems suspicious that this is classified as philosophy - it has an empirical foundation! You can even design experiments to test these "theories of metascience:" just check to see how fast the scientists that work according to them produce successful theories. I'm very skeptical that such a foundation would work faster or achieve more optimal results than an analytical approach. You can keep going meta, but so can I, and the analytical approach will always outpace the empirical one. > Roughly speaking, the division I am making here goes: experiments plus arguments makes science, arguments without experiments makes philosophy. I suppose it depends what you classify as "arguments". Science traditionally produces theories that provide predictive power, but also explanatory power. This latter aspect is often overlooked, but it's essential since it provides insights into further possible experiments. If your "arguments" includes an ontology for scientific theories, then I could agree with your breakdown, but notice that an ontology overlaps with philosophical inquiry. Also, there's a new and growing branch called "experimental philosophy" which includes experiments that test intuitions on philosophical questions, so it's still not quite clear cut!
- whatshisface 9y ago