3 ms·

>If I shake a dried gourd over a cold sufferer and he gets better, I can claim to have been the cause of the recovery. And guess what? No one can prove that cla

>If I shake a dried gourd over a cold sufferer and he gets better, I can claim to have been the cause of the recovery. And guess what? No one can prove that claim false. Umm, sure they can? They repeat the experiment (or perform a better one) and don't get the same results. Lack of controls makes for weaker evidence, but it doesn't magically make things unfalsifiable. And so I ask you again: What makes you say these studies are unfalsifiable?

lutusp 13y ago
> Umm, sure they can? They repeat the experiment (or perform a better one) and don't get the same results. Lack of controls makes for weaker evidence, but it doesn't magically make things unfalsifiable. Yes, in fact, that's exactly what it does. My dried gourd example should have demonstrated why this is so -- I can always claim a correlation with the gourd and the outcome, and without proper experimental controls, that's irrefutable. My dried gourd example is meant to show the need for experimental controls -- another group that recovers in the same amount of time without benefit of gourd-shaking undermines the claim that the gourd had anything to do with the outcome. But "undermines" doesn't equal "falsified" -- the latter condition requires that we test an explanation, not just an observation. The method of performing more experiments, gathering more data, decreases the p-value of the outcome, but it cannot turn a hypothesis into a theory until we're testing an explanation rather than measuring an effect. As long as we measure effects without trying to explain them, we're doing an expensive kind of accounting, but not science. Science requires that we try to offer and then test an explanation for what's being observed, and toss the explanations that don't pan out. The tl;dr: science requires theories. > And so I ask you again: What makes you say these studies are unfalsifiable? The problem we're having here involves a definition of terms -- what falsifiable means. More here: http://arachnoid.com/building_science http://arachnoid.com/building_science
- aninhumer 13y ago
  I thought this argument sounded familiar, and looking at my comment history confirms it. I argued with you about the necessity of an explanation for something to be scientific previously. Clearly you feel strongly about this issue, so I'll make a better attempt to understand your argument this time. Can you clarify for me exactly what you mean by an "explanation"? The linked article starts to talk about a difference between description and explanation, but then just arbitrarily says that Newton's theory of gravity constitutes an explanation. I don't see how Newton's theory is anything more than a very detailed description. Why is the hypothesis "There's a magic force called gravity that behaves according to these laws." any more valid than the hypothesis "Stuff falls because magic." Both provide testable predictions, and can be falsified. Indeed one could simply point to the moon and say "that doesn't fall, theory disproven" just as one could point to the unusual orbits, and disprove Newton. I don't see how that makes the hypothesis itself unscientific.

lutusp 13y ago

> Umm, sure they can? They repeat the experiment (or perform a better one) and don't get the same results. Lack of controls makes for weaker evidence, but it doesn't magically make things unfalsifiable. Yes, in fact, that's exactly what it does. My dried gourd example should have demonstrated why this is so -- I can always claim a correlation with the gourd and the outcome, and without proper experimental controls, that's irrefutable. My dried gourd example is meant to show the need for experimental controls -- another group that recovers in the same amount of time without benefit of gourd-shaking undermines the claim that the gourd had anything to do with the outcome. But "undermines" doesn't equal "falsified" -- the latter condition requires that we test an explanation, not just an observation. The method of performing more experiments, gathering more data, decreases the p-value of the outcome, but it cannot turn a hypothesis into a theory until we're testing an explanation rather than measuring an effect. As long as we measure effects without trying to explain them, we're doing an expensive kind of accounting, but not science. Science requires that we try to offer and then test an explanation for what's being observed, and toss the explanations that don't pan out. The tl;dr: science requires theories. > And so I ask you again: What makes you say these studies are unfalsifiable? The problem we're having here involves a definition of terms -- what falsifiable means. More here: http://arachnoid.com/building_science http://arachnoid.com/building_science

aninhumer 13y ago
I thought this argument sounded familiar, and looking at my comment history confirms it. I argued with you about the necessity of an explanation for something to be scientific previously. Clearly you feel strongly about this issue, so I'll make a better attempt to understand your argument this time. Can you clarify for me exactly what you mean by an "explanation"? The linked article starts to talk about a difference between description and explanation, but then just arbitrarily says that Newton's theory of gravity constitutes an explanation. I don't see how Newton's theory is anything more than a very detailed description. Why is the hypothesis "There's a magic force called gravity that behaves according to these laws." any more valid than the hypothesis "Stuff falls because magic." Both provide testable predictions, and can be falsified. Indeed one could simply point to the moon and say "that doesn't fall, theory disproven" just as one could point to the unusual orbits, and disprove Newton. I don't see how that makes the hypothesis itself unscientific.