Consensus Counts

by Ben Albahari 19. November 2010 13:00

In the case a hypothesis is supported by clear-cut tests, there is no need for consensus. Consensus is however useful when a plethora of tests are required to test the hypothesis, and domain expertise is required to understand and synthesize those tests to attain the likelihood of that hypothesis.

Hypotheses vary in how easy they are to test. The General Theory of Relativity, while a difficult theory to understand, has hypotheses with clear-cut tests that a layperson can easily grasp. Will a clock on a plane have lost a little time on its trip? Will the position of a planet be correctly predicted? If the theory predicts the outcome of these tests, a layperson has a straightforward reason to believe in that theory, even if they have little grasp of the complexity of the theory itself. In comparison, AGW is much harder to test. The central theory, of the greenhouse effect, is actually remarkably simple, as is the hypothesis it suggests, that CO2 warms the planet. But testing that hypothesis in the real world requires synthesizing data from a plethora of tests, and you can't really do this without being pretty familiar with climate science.

I believe the complexity of testing the AGW hypothesis is what ultimately troubles earnest climate skeptics. If they can't understand the tests, how can they feel comfortable signing off on the hypothesis? The strategy of only believing in hypotheses that have been demonstrated with simple tests is not actually a bad rule of thumb - in fact - I wish more people used it! It will correctly lead you to believe in General Relativity and the Theory of Evolution, while rejecting homeopathy and astrology. The problem is that it's an overly-aggressive "truth-filtering algorithm". It will cause people to reject theories that they would otherwise accept if they could understand the science needed to understand the tests. It's like using a skeptical axe rather than a skeptical scalpel.

In the case one doesn't personally understand the tests for a hypothesis, the rational thing to do is to outsource the assessment of that hypotheses to the experts. If an expert consensus exists, our task is easier - they're probably right, and the dissent is most likely from deluded or deliberate noise-makers with claims of conspiracies and ignored evidence. However, we shouldn't dismiss such claims - contrarians are occasionally right. A fast screening process you can use on a contrarian without spending your precious time delving into their arguments is to see if they've made radical claims elsewhere. For example, Roy Spencer, a climate skeptic, is also an evolution skeptic, claiming that Intelligent Design is "no more religious, and no less scientific, than evolutionism". The more rigorous approach however, is to examine specific claims challenging the consensus view for consistency.

Contrarians are usually "contrary for the sake of being contrary". Rather than disagreeing on a few specific points, they tend to disagree with everything they can about the consensus view. This inevitably leads to inconsistent claims. If the consensus view claims A and B, the contrarian will often claim NOT A and NOT B. But is it consistent to believe both NOT A AND NOT B? For example, climatologists believe A: That we know the cause the cause of warming, and B: That the cause is humans. Skeptics often believe both NOT A: That we don't know the cause of warming, and NOT B: That the cause is very likely natural. But NOT A AND NOT B are inconsistent!

In the absence of consistent contrarians, the rational thing to do is to trust the consensus. While a broad consensus has the negative effect of shunning dissent, it also has the positive effect of raising the number of people eyeballing the science, and raising the potential glory for the eagle eyed maverick. If a consistent alternative to the consensus view can be made, it will readily surface.


Expert opinions here.


Pingbacks and trackbacks (1)+