Why Don't More Scientists Use Something Like the IPCC's Probability Scale?

The Intergovernmental Panel on Climate Change (IPCC) is the main scientific body that studies global warming. Although it takes some heat from the skeptical, it seems that in a broad sense the IPCC has done two things right.

The first thing it has done right is to avoid making only one projection of future global warming until the year 2100. Given all the uncertainties, it would be very dogmatic to do such a thing. Instead of publishing only one projection, the IPCC publishes a variety of different scenarios, which have names such as RCP8.5, RCP6.0, RCP4.5, and RCP2.6. Some of them are shown in the graph below (from this document).

The climate change implications of these scenarios are shown in the next IPCC graph, which uses a color-coding scheme matching the one used in the previous graph. So we see that the RCP 8.5 scenario (requiring a total carbon dioxide emission of about 6000 gigatons or more) translates into a very damaging 3.5 Centigrade increase in global warming. But the RCP2.6 scenario and the RCP 4.5 scenario (involving only about 3 or 4 gigatons of carbon dioxide emission) result in a much smaller increase of between 1.5 and 2..0 Centigrade. 

 

Now, you could argue that the RCP 8.5 scenario will never occur, on the grounds that we will never be extract so much carbon dioxide (6000 gigatons or more). The International Energy Agency estimates that the total world reserves of coal are only about 892 billion tons, and much of that is hard-to-extract coal that may never be extracted because it is not economically feasible to extract.

But given the IPCC's approach of providing widely different scenarios, such an objection does not damage its credibility. For one of its other scenarios such as RCP 2.6 or RCP 4.5 may still be valid, despite such an objection. Given its differing scenarios allowing such a variety of outcomes, we really can't fault the IPCC for being too dogmatic in this regard.

There's another thing that is good about the IPCC approach: the fact that it uses a probability scale. The IPCC has a scale that looks like this:

 

These probability assessments are used in the IPCC's Fifth Assessment report. So rather than the report speaking as if it was 100% certain about everything, we are told that some things are likely, other things are very likely, and other things are as likely as not.

The use of such a likelihood scale is to be commended, because there is a great deal of uncertainty in our understanding of nature. If scientists are not sure about something, it is is good that they state assertions in a shaded away, expressing only moderate confidence or low confidence.

It would be good if other sciences followed such an approach. But do we find such a likelihood scale used in fields such as biology, physics, neuroscience or cosmology? No, we do not.

The general tendency in contemporary science is to follow a very different approach. It is as if there were two big lists: the list of approved doctrines, and the list of disapproved doctrines. All of the items on the list of approved doctrines are pretty much taught as if they were gospel truths. This is not too different from the way that the Catholic Church teaches religious doctrines.

Once some teaching somehow manages to get on this list of approved doctrines, the doctrine has it “made in the shade.” We are then told that there is a scientific consensus, so it's kind of “case closed.”

But what if a different approach was taken, an approach using a probability scale, like the IPCC is using? The results would be salutary. There would no more be a situation in which doctrines could endlessly “rest on their laurels.” Each scientific teaching would have to be graded on a probability scale. If such a probability scale was used honestly, it would soon become apparent that some of the theories that have chewed up the most research funding have a not-very-high ranking on the probability scale.

It would be best to make use of a probability scale rather different from the one used by the IPCC. The scale below would be better suited for grading a wide variety of scientific assertions.

Almost certain	Greater than 99% likelihood
Very likely	Greater than 90% likelihood
Likely	Greater than 50% likelihood
Quite possible	Greater than 5% likelihood
Possible	Greater than 1% likelihood
Remotely possible	Less than 1% likelihood
Apparently impossible	Apparently no likelihood

I can imagine a two-phase exercise to make use of such a scale. In the first phase, you begin with a chart like the one below. You draw colored lines between the items on the left and the items on the right, using a different color of ink for each level of confidence.

So the end result might look something like this:

In the second phase of the exercise, you would be required to justify each case in which you had specified that one of the items on the left had a probability of “quite possible” or higher. This would require actually presenting observations or experiments that show there is a basis for concluding a likelihood of at least 5%. Appeals to current popularity or a consensus of agreement would not be allowed.

Let us imagine how clarifying such an exercise would be. Using such a scale in the area of physics and cosmology would throw light on how much of modern theoretical physics is on shaky ground. For example, there would be no basis for giving either the theories of supersymmetry or string theory a probability grade of higher than “remotely possible” or “possible.” The fact that such theories are widely popular would be no basis for granting them an assessment of “quite possible” or “likely.”

Using such a likelihood scale sounds like a great idea, so why is such a scale not used in fields such as cosmology, physics, and biology? I suspect the reason is that using such a scale would involve introducing a level of humility that today's dogmatic theoretician would prefer to avoid.

Consider the dogmatic scientific theorist. He may advance some theory unlikely to be true. But he would rather that people not judge his theory based on some scale in which it is judged whether the theory is “quite possible,” “likely” or “very likely.” Because then it might become clear that there is no basis for concluding that the theory is probably true. Such a theorist would prefer the current approach, in which it is as if there is a “list of approved doctrines,” and you are supposed to accept all the items on the list. Having got his theory on such a list (often because of sociological considerations), such a theorist would not want there to be a system in which scientists use varying shades of confidence in their assertions. It might then become apparent to all that the theorist's pet theory is not on very firm ground. 

Similarly, an apologist for an organized religion would never want you to make use of some exercise like the one above, in which you drew lines specifying whether 15 assertions of his creed were “possible,” “quite possible,” “likely,” “very likely,” or “almost certain.” He would instead want you to accept the whole creed with complete confidence, regarding every item in it as a certainty.

Like such a theological apologist, the modern pitchman for the "official party line" of modern science would rather that you not use probability scale ratings that might make his "standard story" look like something that is largely speculative, largely a kind of social construct in which many a predominant theory is more of a speech custom than something that has been established as a likelihood.