We also have centuries of history which show that practices people thought were good actually did nothing, or made things worse. People got better despite the t

We also have centuries of history which show that practices people thought were good actually did nothing, or made things worse. People got better despite the treatment, which caused them to believe that the treatment was the success. It took over 1,000 years for the humorism of Ancient Greece to be discarded as wrong, and Hippocrates' theories on bloodletting, once a mainstay of medicine, to be relegated to only a few modern uses. Mercury has been a part of traditional medicines for thousands of years across Eurasia. "There is hardly a disease in the nosology for which mercury in some form has not been used." "No doubt many patients died after years of suffering from slow mercurial poisoning." (Quotes from http://www.soilandhealth.org/02/0201hyglibcat/020134syphilis/020134syphilis-ch9.htm http://www.soilandhealth.org/02/0201hyglibcat/020134syphilis... ) . While mercury still has some medical uses, the fact is that centuries of reports claimed wide benefits and successes for mercury is evidence that claiming benefits for oneself, and the agreement of others, is not strong evidence that it actually is good. It may be good, but it isn't sufficient. It may actually be bad. There's also drugs which we know are effective, but which are usually riskier and less effective than other drugs. Eg, sulfa drugs vs. beta-lactams. While treating with sulfa drugs absolutely has benefits, promoting it over other, better possibilities leads to overall worse healthcare. That's why sulfas aren't a first line drug any more. In terms of a Paleo diet, it may be good, or not, or it may be that other diets are better. How do you know which it is? How do you know which traditional uses of mercury are toxic? Which traditional uses of bloodletting are worse than not doing bloodletting at all? We only have three solutions: trials (which gives the clearest evidence when done right), epidemiology (much harder to interpret), and mechanistic (there's no reason to believe that a leg amputation will help a cold, and the best case outcome quite severe, making it unethical to test that experimentally. This is mostly used to reject unlikely possibilities which have poor evidence backing it.)