In previous post Snow, a thing of the past I explored the positions of both sides of the debate concerning the connection between snow and global warming. One thing I didn’t touch yet was the skeptics’ remark that “One can not have it both”. There were statements before that snow was a thing of the past because of global warming, but now when snow falls in abundance, there is the statement this is also because of global warming. If this is the case, it can not be proven right nor wrong.
By their nature, skeptics are, well… skeptical. They want testable statements that can prove or disprove something. This is rooted in science and is called falsifiability. In the most simplest way it looks like this: a hypothesis (testable statement) must predict at least one observation by which it can be refuted. If the evidence is in line with the prediction, the statement may be right. If the evidence conflicts with the prediction, the statement is wrong.
Maybe a bit theoretical, so an example to make it clear. Let’s suppose the statement: “All ravens are black”. It can be made testable by stating that if all ravens are black, this means no raven with another color will exist.
- If we find a white raven, then the statement “All ravens are black” is clearly wrong: it is falsified (proven to be false)
- If we see only black ravens, then the statement might be right. “Might” because there is always the possibility that we just didn’t find the ones with another color. They could exist, even if we didn’t find them (yet).
By the way, white ravens do exist (although they are rare).
This is a very simple example and it should be nice if this was also applicable to global warming. The reality is not always that simple. Predicting something is hard, especially about the future. Especially about a complex thing like the climate. So statements about global warming will be stated in a less clear manner. Something like: “it will be more likely”, “it might” or “it is unlikely”. These are all stated as a probability.
One problem: statements based on probability are not falsifiable by default. This is not difficult to understand: suppose one has the statement It is very likely that the frequency of heavy precipitation events will increase. Suppose we define very likely as more than 90% chance. What happens when we test the precipitation in certain places over a certain time?
- If we find an increase in heavy precipitation, the statement might be true.
- If we find a decrease in heavy precipitation, the statement isn’t necessarily false. There is a possibility of at least 90% of an increase, but this also means a possibility of up to 10% equal or a decrease in heavy precipitation. It might even be possible that the precipitation was measured in a different place where the increase was not observed. Or maybe the time frame wasn’t long enough to be able to see the increase.
In conclusion, even if the outcome is not consistent with the statement, it doesn’t falsify it.
Of course, first the meaning of increase in frequency and heavy precipitation and probably also the time frame have to be defined. If not, the statement is almost meaningless and one could prove about everything.
This seems disturbing. Does this means that many of the global warming statements can’t even be falsified? I was puzzled about this for a long time. Now I think it is not necessarily true. In the real world we realize that falsifiability is important, but not the only thing being taken into consideration.
Suppose a probability statement that eventually fails. As we saw this doesn’t mean the statement is false, but it doesn’t exactly add to the credibility of the statement and those who make it. The higher the probability with which the statement was stated, the more credibility loss. It is not surprising at all to see fading confidence if observations are not in line with what one would expect from the theory.
A non falsifiable statement looks like a horrible thing when trying to prove or disprove something. But look at it from the other side: a probability statement has, well…probability. Snow and heavy precipitation are consistent with the theory of global warming, but they are also consistent with other things. There isn’t necessarily just one theory. By the fact the theory has to be proved by probability statements, there will be at least one competing theory, maybe even more. All with their own probabilities, explanation of previous observations and predictive value.
This post started with the remark “One can not have it both”. Well, can it? Of course not. At least not in the sense of piling up statements so a theory will be confirmed with any observation possible. That would be the same as stating that “All ravens are black, except when they are not”. This doesn’t learn us nothing new, the predictive value of this is exactly zero.
It is easy to come with an explanation after the facts. The real test is how a theory predicts data we haven’t seen yet and also describes previous and current observations best. This doesn’t necessarily means falsification in a strict true|false way as we would expect with the “One can not have it both”-remark. But rather a comparison between competing theories.