2 ms·

It's reasonable to dismiss hypotheses that are untestable, provide no predictive power and have questionable evidence to support them. That's not so much about

It's reasonable to dismiss hypotheses that are untestable, provide no predictive power and have questionable evidence to support them. That's not so much about being close-minded, but just a practical methodology. If you accept non-testable hypotheses, then how can you tell whether it was the result of a specific deity or an invisible pink unicorn?

mistermann 4y ago
How is "provide no predictive power" achieved?
- ndsipa_pomu 4y ago
  > How is "provide no predictive power" achieved? It's very similar to being untestable. If a hypothesis provides a prediction, but can alter the interpretation after the event to fit the facts, then it's providing no predictive power. e.g. If someone predicts that <insert deity> rewards/punishes people when they die (which alters their near death experience), but then amends their interpretation of good/bad to fit the data, then it can't be used to predict anything.
  - mistermann 4y ago
    > If a hypothesis provides a prediction, but can alter the interpretation after the event to fit the facts, then it's providing no predictive power. e.g. Is this an accurate and comprehensive description of the situation we have before us though? > If someone predicts that <insert deity> rewards/punishes people when they die (which alters their near death experience), but then amends their interpretation of good/bad to fit the data, then it can't be used to predict anything. That you can identify one attribute of this nature is no guarantee that all attributes (which is not known) are of this kind.
1827163 4y ago
I don't by default accept the existence of such phenomena, instead I don't want to summarily, without investigation, dismiss the existence of it. There might still be a way we can test these hypotheses at some point in the future. It's just that we can't do it now. But not being able to test it now, is no reason to dismiss it, because then we will never go down the path that might eventually result in it being testable.

mistermann 4y ago

How is "provide no predictive power" achieved?

ndsipa_pomu 4y ago
> How is "provide no predictive power" achieved? It's very similar to being untestable. If a hypothesis provides a prediction, but can alter the interpretation after the event to fit the facts, then it's providing no predictive power. e.g. If someone predicts that <insert deity> rewards/punishes people when they die (which alters their near death experience), but then amends their interpretation of good/bad to fit the data, then it can't be used to predict anything.
- mistermann 4y ago
  > If a hypothesis provides a prediction, but can alter the interpretation after the event to fit the facts, then it's providing no predictive power. e.g. Is this an accurate and comprehensive description of the situation we have before us though? > If someone predicts that <insert deity> rewards/punishes people when they die (which alters their near death experience), but then amends their interpretation of good/bad to fit the data, then it can't be used to predict anything. That you can identify one attribute of this nature is no guarantee that all attributes (which is not known) are of this kind.

1827163 4y ago

I don't by default accept the existence of such phenomena, instead I don't want to summarily, without investigation, dismiss the existence of it. There might still be a way we can test these hypotheses at some point in the future. It's just that we can't do it now. But not being able to test it now, is no reason to dismiss it, because then we will never go down the path that might eventually result in it being testable.