Does Diet Affect Cancer Incidence or Severity? Decision-making under Uncertainty

Maybe. Maybe not. That's the gist of this article in the NYTimes:

The diet messages are everywhere: the National Cancer Institute has an "Eat 5 to 9 a Day for Better Health" program, the numbers referring to servings of fruits and vegetables, and the Prostate Cancer Foundation has a detailed anticancer diet.

Yet despite the often adamant advice, scientists say they really do not know whether dietary changes will make a difference. And there lies a quandary for today's medicine. It is turning out to be much more difficult than anyone expected to discover if diet affects cancer risk. Hypotheses abound, but convincing evidence remains elusive.

... So should people who are worried about cancer be told to follow these guidelines anyway, because they may work and will probably not hurt? Or should the people be told that the evidence just is not there, so they should not deceive themselves?

Speaking for myself, I want to know. I want to know that I'm not increasing the risk of cancer by refusing to eat cauliflower and parsnips and Brussels sprouts. Contrary to the hypothetical assertion in the article, eating those things really does hurt: they taste bad, and they make me feel ill. In econ-speak, they have a large and negative marginal utility, starting at Q = 0.

[h/t to Jack for the pointer; his question: "So, is broccoli a placebo?"]