Consensus is not Proof

[chris_gerrib]

So I read in today's New York Times (I think found it via instapundit) that there is no proof for the contention that a low-fat diet is particularly good for you.

So how come the surgeon general said in 1988 that a high-fat diet was a public health menace comparable to cigarettes? Well, that was the consensus scientific opinion of the time. Gary Taubes, whose book, “Good Calories, Bad Calories was reviewed in the article, claims this was a "concensus cascade."

Concensus cascades happen where a group of people, or even an individual, express a strong (if mistaken) opinion on a matter of some doubt. If other people then assume that the first person was right and say they agree, latecomers to the debate then feel greatly pressured to join the consensus. By 1988, the consensus was so strong that we had a "reputation cascade" - people who disagreed with the consensus were assumed to be paid agents for, in this case, the meat and dairy industry.

Now, if this sounds anything like what might be happening with the global warming debate, the similarities are purely coincidental. If you believe that, I've got a bridge for sale in Brooklyn. Nor is this a new phenomenon. Go read up on the consensus about luminiferous ether.

I personally think the Earth's average temperature is rising, and mankind is at least partially at fault. I thought at one time that the three Duke lacrosse players were guilty. Fortunately, this belief had to be tested in a court of law. In the end, it proved not to be strong enough to even step foot in a court.

Consensus is not proof. We as a society need to always challenge conventional wisdom, consensus and things "everybody knows." Usually, the consensus is right. But testing that consensus is healthy and necessary, and calling people who do that testing dupes is wrong.

Comments

[jetfx.livejournal.com]

If you study logic, the consensus trap is what's called the fallacy of numbers. Basically, if large numbers of people believe it, it must be true.

[bdunbar.livejournal.com]

there is no proof for the contention that a low-fat diet is particularly good for you.

I thought the consensus (snark) was that 'what diet is best for whom' is a factor of genetics and habit?

[chris-gerrib.livejournal.com]

You have correctly stated MY consensus. Unfortunately, that is probably not scientific consensus.

[jeff-duntemann.livejournal.com]

Superb post. One reason we're getting consensus cascades is that mass media concentrates the attention of immense numbers of people on relatively few sources of opinion and research. This is an issue separate from any supposed liberal (or in some minds conservative) bias in the media; the point is that mass media feeds on itself in nonobvious ways (reporters trade lists of sources among themselves, amplifying the influence of any given source on any list, etc.) and often yields a sort of "channel capture" effect in a particular debate. It's much worse in a field like health, where we actually know much, much less than we claim to.

For example, we're only just now gathering some reasonably reliable data indicating that sleep loss contributes to obesity, something I've suspected for many years. During the worst of my periods of sleep disruption, which mapped to the implosion of my company in Arizona, I gained weight--and then lost it almost immediately once I began sleeping well again. I'll summarize what I've gathered on Contra in a few days, but it goes against conventional wisdom and I don't expect people to take it seriously any time soon.