What should I have for dinner? (A case study in decision making)

Every­one knows that eat­ing fatty foods is bad for you, that high choles­terol causes heart dis­ease and that we should all do some more ex­er­cise so that we can lose weight. How do I know that ev­ery­one knows this? Well, for one thing, this gov­ern­ment web­site tells me so:

We all know too much fat is bad for us. But we don’t always know where it’s lurk­ing. It seems to be in so many of the things we like, so it’s some­times difficult to know how to cut down.

...kids need to do at least 60 min­utes of phys­i­cal ac­tivity that gets their heart beat­ing faster than usual. And they need to do it ev­ery day to burn off calories and pre­vent them stor­ing up ex­cess fat in the body which can lead to can­cer, type 2 di­a­betes and heart dis­ease.

See, it’s right there in black and white. We all know too much fat is bad for us. Ex­cept… there are a lot of peo­ple who don’t agree. Gary Taubes is one of them, His book, Good Calories Bad Calories (The Diet Delu­sion in the UK and Aus­tralia), sets out the case against what he calls the Die­tary Fat Hy­poth­e­sis for obe­sity and heart dis­ease, and pro­poses in­stead the Car­bo­hy­drate Hy­poth­e­sis: that both obe­sity and heart dis­ease are caused by ex­ces­sive con­sump­tion of re­fined car­bo­hy­drates, rather than dietary fat.

Taubes is very con­vinc­ing. He ex­plains how peo­ple have con­sis­tently recom­mended low-carb diets for weight-loss for the past 150 years. He ex­plains how sci­en­tists roundly ig­nored stud­ies that con­tra­dicted the link be­tween high choles­terol and coro­nary dis­ease. There are de­tails of the mechanism by which eat­ing re­fined car­bo­hy­drate af­fects in­sulin pro­duc­tion, lead­ing to obe­sity. He gives a plau­si­ble nar­ra­tive for how the Die­tary Fat Hy­poth­e­sis came to be ac­cepted sci­en­tific wis­dom de­spite not ac­tu­ally be­ing true (or sup­ported by the ma­jor­ity of the ev­i­dence). He ex­plains how stud­ies of low-fat diets sim­ply ig­nored over­all mor­tal­ity rates, re­port­ing only deaths from heart dis­ease, and how one study wasn’t pub­lished be­cause ‘we weren’t happy with the way it turned out’. All in all, the book is very con­vinc­ing.

I ex­pect a rel­a­tively large per­centage of peo­ple on LW are already aware of this. Search­ing the LW archives for ‘Taubes’ gives sev­eral, mostly pos­i­tive, refer­ences to his work (Eliezer seems to be con­vinced “Die­tary sci­en­tists ig­nor­ing their own ex­per­i­men­tal ev­i­dence have kil­led mil­lions and con­demned hun­dreds of mil­lions more to obe­sity with high-fruc­tose corn syrup.”). How­ever, I do ex­pect it to be news to some peo­ple, and I think it raises an im­por­tant ques­tion. Given that ev­ery­one needs to eat some­thing, we all need to de­cide whether we be­lieve Taubes or whether we be­lieve Change 4 Life.

Good Calories, Bad Calories is 601 pages of rel­a­tively small type, and con­tains 111 pages of refer­ences. Most of you prob­a­bly don’t want to read a book that long, and you definitely don’t want to check all of it’s refer­ences. Even if you did, Taubes openly ad­mits that his book is at­tempt­ing to ar­gue for the Car­bo­hy­drate Hy­poth­e­sis—he is try­ing to con­vince you, why should you be sur­prised if you find your­self con­vinced? (He claims not to be cherry-pick­ing but then, he would, wouldn’t he?) So how can you de­cide whether to trust the gov­ern­ment or whether to trust some jour­nal­ist with no train­ing in biol­ogy? Even if you do de­cide to as­sess the ev­i­dence for your­self, how ex­actly should you go about it?

This is the key ques­tion of ra­tio­nal­ity. How can we be­lieve what is true? And I think this makes a great case study—it’s an area in which we all have to have a be­lief (or at least, act as though we have a be­lief) and one in which there is (or at least ap­pears to be) gen­uine con­tro­versy as to what is true and what is not.

If you’ve already thought about this, do you be­lieve Taubes’ the­sis, and how did you come to this con­clu­sion? If this is the first time you’ve ever heard of Taubes, how far have you shifted your prob­a­bil­ity for the Die­tary Fat Hy­poth­e­sis based on read­ing this post? What more re­search do you in­tend to do to de­cide whether or not to con­tinue be­liev­ing it? How much weight do you place on the fact that I be­lieve Taubes? On the fact that Eliezer be­lieves Taubes (Eliezer, if your po­si­tion is more nu­anced than this, feel free to cor­rect me)? How much did you up­date your be­liefs based on what other com­men­tors have said (as­sum­ing there have been any)?