Of course it’s complicated! I’m saying, there’s serious grounds for suspicion here. And the problem, if it exists at all, is likely to be gigantic. So we need to pay attention even though it doesn’t look very likely. A genuine Pascal’s Wager. We aren’t allowed to shrug our shoulders in response. Scope insensitivity is one of the sins.
All these funny diseases that look like mixtures of type 2 versions of well understood endocrine disorders. That I didn’t know about until after I’d made up the idea. And a very simple hypothesis that explains them all and should be easy to refute. I predict low body temperature in every different group. Patterns of differently low body temperatures correlating with how much the disease looks like classical hypothyroidism.
I have a hypothesis formed by whatever dodgy method I like, and which has turned out to have been commonly suspected by many different people, all starting from different observations, which I am now using to explain and predict lots of other facts that didn’t figure in the original making-it-up process.
Does the order in which I learned these facts matter? How should I adjust my conclusions to account, even given that I probably can’t remember the precise order? I am going through periods of puzzlement, enlightenment, and then spectacular rewards of confirmation followed by terror at the implications.
And the competing explanations all turn out to be philosophically suspect.
This science business turns out to be quite hard. And we claim (and I believe us) that we are unnaturally good at this sort of thing. Where have I erred, Brothers in Bayes?
What do you know that I don’t know? What conclusions (that are safe to draw in public) do you draw from my idea and do they turn out to be true? What are the odds and why? What is a yes worth. What is a no worth?
Are doctors actually trained to ignore these symptoms? Because they’re everywhere? How common are these diseases?
Are the patterns of occurrence the same in every racial group? Are they different in different countries? Are there places where some mysterious cause is making itself particularly obvious?
How much confounding has this caused in all epidemiological data ever? That might be the biggest prize.
Should I take my thoughts to medical statisticians? Or can I actually get a better answer here?
I’m saying, there’s serious grounds for suspicion here.
That’s still handwaving.
Let’s invoke Popper and ask for specific, testable, falsifiable statements. What exactly do you claim and want to test? What outcomes will prove you wrong? I don’t think the details of how you came to formulate your hypothesis matter.
Should I take my thoughts to medical statisticians?
We aren’t allowed to shrug our shoulders in response.
Actually we are. Changing the status quo is hard even if you are right.
Should I take my thoughts to medical statisticians? Or can I actually get a better answer here?
I don’t think mainling the original post to any medical statistician will get you anywhere. You would beforehand have to be clearer about your thesis and the evidence you have. It helps to cite the evidence.
which I am now using to explain and predict lots of other facts that didn’t figure in the original making-it-up process.
A prediction is something that has a credence value especially if you see yourself as Bayesian. At the moment you don’t state those.
Changing the status quo is hard even if you are right.
Shouldn’t be. If I can sharpen my argument to the point where I believe it myself, then I can take it to the ivory towers of the wise and they will listen. I know these people, and I trust them. They will do the right thing.
Of course it’s complicated! I’m saying, there’s serious grounds for suspicion here. And the problem, if it exists at all, is likely to be gigantic. So we need to pay attention even though it doesn’t look very likely. A genuine Pascal’s Wager. We aren’t allowed to shrug our shoulders in response. Scope insensitivity is one of the sins.
All these funny diseases that look like mixtures of type 2 versions of well understood endocrine disorders. That I didn’t know about until after I’d made up the idea. And a very simple hypothesis that explains them all and should be easy to refute. I predict low body temperature in every different group. Patterns of differently low body temperatures correlating with how much the disease looks like classical hypothyroidism.
I have a hypothesis formed by whatever dodgy method I like, and which has turned out to have been commonly suspected by many different people, all starting from different observations, which I am now using to explain and predict lots of other facts that didn’t figure in the original making-it-up process.
Does the order in which I learned these facts matter? How should I adjust my conclusions to account, even given that I probably can’t remember the precise order? I am going through periods of puzzlement, enlightenment, and then spectacular rewards of confirmation followed by terror at the implications.
And the competing explanations all turn out to be philosophically suspect.
This science business turns out to be quite hard. And we claim (and I believe us) that we are unnaturally good at this sort of thing. Where have I erred, Brothers in Bayes?
What do you know that I don’t know? What conclusions (that are safe to draw in public) do you draw from my idea and do they turn out to be true? What are the odds and why? What is a yes worth. What is a no worth?
Are doctors actually trained to ignore these symptoms? Because they’re everywhere? How common are these diseases?
Are the patterns of occurrence the same in every racial group? Are they different in different countries? Are there places where some mysterious cause is making itself particularly obvious?
How much confounding has this caused in all epidemiological data ever? That might be the biggest prize.
Should I take my thoughts to medical statisticians? Or can I actually get a better answer here?
That’s still handwaving.
Let’s invoke Popper and ask for specific, testable, falsifiable statements. What exactly do you claim and want to test? What outcomes will prove you wrong? I don’t think the details of how you came to formulate your hypothesis matter.
Will they listen to you?
http://lesswrong.com/lw/nbm/thyroid_hormones_chronic_fatigue_and_fibromyalgia/
I hope so!
Actually we are. Changing the status quo is hard even if you are right.
I don’t think mainling the original post to any medical statistician will get you anywhere. You would beforehand have to be clearer about your thesis and the evidence you have. It helps to cite the evidence.
A prediction is something that has a credence value especially if you see yourself as Bayesian. At the moment you don’t state those.
Shouldn’t be. If I can sharpen my argument to the point where I believe it myself, then I can take it to the ivory towers of the wise and they will listen. I know these people, and I trust them. They will do the right thing.
For the rest, see:
http://lesswrong.com/lw/nbm/thyroid_hormones_chronic_fatigue_and_fibromyalgia/
How much have you talked to people inside the system? From my conversations with stakeholders I have the impression that change is very hard.