Galaxy-brained hackers and engineers need to learn how to inoculate themselves against premature excitement over this kind of thing. Listening over the years pr

Galaxy-brained hackers and engineers need to learn how to inoculate themselves against premature excitement over this kind of thing. Listening over the years primarily to Eric Helms of Iron Culture, of all things, has given me a pretty good framework in which to look at research like this. We have an outcome we care about. That is, we want to live longer. We have a proposed intervention. Eat some proprietary diet. How do you study, scientifically, whether undergoing this intervention will produce the desired outcome? The gold standard way to know for sure is you run this. You take a broad sample of all humans on the planet, cutting across races, congenital conditions, metabolic phenotypes and what not, divide them randomly into two groups, feed one whatever else they might normally eat, and feed one this diet, and see who lives longer. Clearly, we can't achieve this standard. This is the basic reason quantum physics is so much more settled than nutrition science. We can easily produce and mess with hundreds of quadrillions of photons and electrons a day in labs across the globe and measure exactly what happens to them. It would take centuries and no such thing as ethics boards to do this with humans. So instead we typically end up with a few other options. One is epidemological studies. Collect lots of data over the decades and survey how people ate. See how long they lived. Try to find patterns in that data. Not a bad idea in principle, but survey data like this is often unreliable and this can't work for a new proposed intervention in which no past humans did this. Now we get down to mechanistic studies. Okay, we can't see directly if doing something will make you live long. Let's see if doing it will change some other quantity we can measure that other research has suggested maybe correlates with longer life. It's getting iffier, but if it's the best we can do, so be it. Then we get to the bottom of the barrel. That is mechanistic studies but not even conducted on humans. Let's do something to mice instead and see how biochemical markers change in the short term. It's a long way from that to demonstrating with any kind of reasonable assurance that doing this to a randomly-selected human will produce the outcome we actually care about. Funnily enough, Eric has chimed in on this so much largely because his research is in muscle hypertrophy, and for a long time, exercise science simply hadn't been doing work for long enough to answer questions we cared about, like feed one person 30g of whey within 30 minutes of lifting, let another eat normally, otherwise they train and eat the same, will one have measurably more muscle in ten years? As we've gotten that data, we've found that no, we can't find any difference. But a lot of people thought for a long time that it would make a difference. Why? Because we had mechanistic studies. We could measure muscle protein synthesis in the hours following lifting and see that the rate at which it occurred was higher if you ate a fast-absorbing protein within some limited time window after a workout. Sufficiently decent logic suggested that would lead to your body actually building more muscle. Well, it turns out no. It's interesting in general to see how sport outcomes have so drastically improved over the past few decades compared to health and to think about some of the reasons that might happen. At least one seems like athletes and teams have been trying various different training, lifestyle, and game-time strategies for decades and measuring what actually happens in terms of the outcome they truly care about. Will they run faster, jump higher, lift greater weights, and win more?. That can be tested in a reasonable amount of time and you have plenty of willing participants who will try anything and actually comply. All of the people doing it are otherwise pretty similar and trying to accomplish the same thing. Health, in contrast, is just much harder to study. We're never going to have the time, resources, ethical, practical, or political ability to tweak one variable at a time in otherwise identical populations, wait 90 years, see if one lives longer, then do it again until we find something that works. Most likely, we will never have real answers to a question like if I eat a particular diet for the next year or next decade, will I live longer? And remember when the comments get filled with "I did this and felt so great," whether it be for fasting, carnivore, keto, anything that severely restricts what you eat and is, in practice, a somewhat less principle ad hoc version of an elimination diet, that some proportion of the population is always out there living with some kind of food allergy or sensitivity they're not aware of. When they do any kind of restriction, some of them are going to eliminate whatever it was that was giving them a problem. Ergo, they suddenly find they feel much better. When you do this for real, under the supervision of a dietician, you reintroduce what you eliminated until you find what the problem actually was. On the Internet, you conclude this must be a magic diet that any person on the planet would benefit from over the full course of an entire lifetime. At least be aware of the logical leaps you are making and how warranted they truly are.