Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Jiro comments on The Thyroid Madness : Core Argument, Evidence, Probabilities and Predictions - Less Wrong Discussion

10 Post author: johnlawrenceaspden 14 March 2016 01:41AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (132)

You are viewing a single comment's thread. Show more comments above.

Comment author: Jiro 16 March 2016 10:11:44PM 1 point [-]

Suppose diabetes was diagnosed by insulin levels instead of blood glucose. And there were two sets of patients, who had roughly the same symptoms, but one lot weren't treated because the insulin test showed that their problems weren't diabetes. Would you not say that the insulin test was broken?

No, not in general. It might be for diabetes, but that's fact-specific. Let's try substituting in something else:

Suppose fractured skulls were diagnosed by X-rays. And there were two sets of patients, who had roughly the same symptoms (head pains and bleeding), but one wasn't treated because the X-ray test showed that their problem wasn't a fractured skull. Would I say that the X-ray test is broken? Of course not.

Comment author: johnlawrenceaspden 16 March 2016 11:27:44PM 0 points [-]

Nicely done, thank you! My brain is broken, and this "informal reasoning" is harder than it looks.

In your case, the X-ray test is doing its job perfectly. And if the posited type 2 hypothyroidism needs different treatment from the type 1 version, which it probably would, then the TSH test will be a great way to tell them apart.

What I don't think you're allowed to do is say 'no problem, can't be anything to do with your car crash' when what you mean is 'your skull is not fractured'.

So the TSH test is a great test for TSH, and probably a good test for circulating thyroid hormones (although it doesn't give the whole picture). But I don't think that means that the TSH test is a good test for 'no thyroid hormone-related problem'.

Do we still disagree? Can I phrase my A&B&C=>(D OR E) thing better? Or do I need to abandon it?

Perhaps: Hypothyroidism (by which I mean any failure of thryoid hormones to act on cells).....

Comment author: Jiro 17 March 2016 03:49:57AM 0 points [-]

What I don't think you're allowed to do is say 'no problem, can't be anything to do with your car crash' when what you mean is 'your skull is not fractured'.

That example only works because fractures are involved in a subset of car crashes and car crashes are involved in a subset of fractures; either one can happen without the other. If that relation doesn't hold true, you would be allowed to say that. For instance, saying "no problem, can't be anything to do with your car crash' when what you mean is 'you weren't anywhere near a car at the time of the crash'.

Comment author: johnlawrenceaspden 18 March 2016 05:31:50PM 0 points [-]

Agree again, thanks