There has been some debate of late over the role of falsifiability in science. Falsifiability is the philosophical notion advocated by Popper as an acid test to distinguish between ideas that are scientific and those that are not. In short, for a theory to be scientific, it has to be subject to falsification. It must make some prediction which, were it to fail, would irrevocably break it.
A good historical example is provided by the phases of Venus. In the geocentric cosmology of Ptolemy, Venus is always between the Earth and the Sun. Consequently, one should only observe a crescent Venus. In contrast, in the heliocentric cosmology of Copernicus, Venus can get to the other side of the sun, so we should see the full range of phases.
Galileo observed the full range of phases when he pointed his telescope at Venus. So: game over, right?
Well, yes and no. In the strict sense of falsifiability as advocated by Popper, yes, geocentrism was out. That didn’t preclude hybrid pseudo-solutions, like the Tychonic model. Worse, it didn’t convince everyone instantaneously – even among serious minded people not impeded by religious absolutism, this was just one piece of evidence to be weighed along with many others. One might have perfectly good reason to weigh other things more heavily. Only with the benefit of hindsight can we look back and say Nailed it!
Nevertheless, this is often taught to young scientists as an example of how it is suppose to work. And it should. Ellis & Silk make an eloquent defense of the ethic of falsifiability. I largely agree with them, even if they offer a few examples which I don’t think qualify. They were motivated to mount this defense in response to the case made against falsification by Sean Carroll.
Without commenting on the merits of either argument – both sides make good points – it occurs to me that these is also a human element. One of personality and proclivity, perhaps. It has been my experience that those most eager to throw Popper (and Occam’s razor) under the bus are the same people who fancy ornate and practically unfalsifiable theories.
The debate about standards is thus also a debate about the relative merit of ideas. Should more speculative ideas have equal standing with more traditional explanations? If you’re a conservative scientist, you say Absolutely not! If you like to engage in theoretical speculation, you say Hell yes!
Clearly there is value to both attitudes. The more conservative attitude teaches to refrain from turning our theories into Rube Goldberg machines that look really neat but teach us nothing. (Many galaxy formation simulations are like this.) On the other hand, some speculation is absolutely necessary to progress. Indeed, sometimes the most outrageous seeming speculations lead to the most profound advances.
In short, our attitudes matter. There is no such thing as the perfectly objective scientist as portrayed by the boring character in a white lab coat. We are human, after all, and a range of attitudes has value.
In this context, it seems that there should be a value system among scientists that parallels the standard of falsifiability for theories. We shouldn’t just hold theories to this high standard. We should also hold ourselves to a comparably high standard.
I suggest that a scientist must be persuadable. Just as a theory should subject itself to testing and potential falsification, we, as scientists, should set a standard by which we would change our minds. We all have pet ideas and tend to defend them against contrary evidence. Sometimes that is the right thing to do, as the evidence is not always airtight, or can be interpreted in multiple ways.
But – at what point does the evidence become compelling enough that we are obliged to abandon our favorite ideas? It isn’t good enough that a theory is falsifiable. We have to admit when it has been falsified. In short, we should set a standard by which we could be persuaded that an idea we had previously believed was wrong.
What the standard should be depends on the topic – some matters are more settled than others, and require correspondingly more compelling evidence to overturn. The standard also depends on the individual: each of us has to judge how to weigh the various lines of evidence. But there needs to be some standard.
In my experience, there are many scientists who are not persuadable. They are not simply hostile to speculative ideas. They are hostile to empirical data that contradicts their pet ideas. Sadly, in many cases, they do not seem to be able to distinguish between data – what is a plain fact – and contrary ideas. One sees this in the “debate” on global warming all the time: solution aversion (we don’t want to stop burning oil!) leads to cognitive dissonance and the rejection of facts: we don’t want to believe that, so the data must be faulty.
Sadly, this sort of behavior is all too common among practicing scientists. I advocate that this be considered unscientific behavior. Just as a theory should be falsifiable, we must be persuadable.
It’d be nice if we could be civil about it too. Baby steps.