Another Shot At Falsifiability

Suppose you notice that a woman walks by a market every single day at noon.  Do you have any rational justification that she’ll walk by that market at noon tomorrow?

Is this question any different than saying something like “We expect Drug A to eliminate 93% of strep throat cases, because we’ve observed that Drug A eliminates strep throat from 93% of all test subjects we’ve observed.”

In principle, these two statements are not significantly different.  They both rely on empirical observations and some level of rigidity about the data collection.

But in both matters, there is a logical problem about how justified a person is in making conclusions about what will happen in the future.  Just because a woman tends to be in front of a market at noon every day says nothing about whether she will be in the market in the future.  Same goes for the effectiveness of a drug.

This is historically known as the problem of induction.

Karl Popper put forward a variety of premises about what makes a position scientific.  One thing that escaped my attention when I first blogged about this is that, in Popper’s paradigm, a position can be completely wrong, yet be scientific.  For instance, Keppler’s 3 laws, though close approximations, are not correct.  According to Popper, Newton’s theory also falls into the category of false, but scientific.

Had I considered the implications of this, I probably wouldn’t have spent much time worrying about creationists’ invocation of Popper to refute evolution.

But I think the rational justification for favoring a hypothesis that is wrong AND scientific is that it allows you to comfortably say that the hypothesis is wrong.  In other words, there is a justification for the hypothesis’ analysis and rejection; whereas, a hypothesis that is unscientific is more difficult to coherently reject.

Popper grappled with challenges inherent in a field that is tentative and inductive.  An obvious example of this challenge is the problems that arise when, for example, you always observe the woman in front of the market at noon.  Science might say “the woman is always in front of the market at noon,” but that position is rendered 100% false the moment we observe the woman not being in front of the market at noon.  So, even though 99.999% of all noon-time observations showed the woman in front of the market, the absence of the woman in front of the market at noon means that she is NOT always in front of the market at noon.

A person is indeed being a pedantic dipshit when they spend too much time railing against the statement that the woman is always in front of the market at noon, but they’re logically and factually justified to do so.  The blindspot the arguer assumes is that the goalpost might easily be moved by stating that “the woman usually appears in front of the market at noon”.

Popper didn’t like this goalpost-moving.  The introduction of the word “usually” merely serves to add wiggle room to the premise, and ultimately reduces the strength of the argument, in Popper’s mind.  He disdained Marxism for this very reason.

Popper’s solution to this matter was to avoid making claims on induction alone.  Just because all past observations are consistent does not mean that all future observations will be consistent.

Instead, Popper put forward that a claim’s predictions ought to lead to successful bold predictions, and that it to being open to empirical refutation.  In other words, it should be easy to imagine and test how a particular outcome could disagree with the implications of a claim.  Enter falsifiability.

A scientific hypothesis should lead to the ability to test its implications.

One of the claims I’ve seen creationists make about evolution is that, since we can’t observe species changing today, that we can’t test evolution.  This is not only a gross mischaracterization of Popper, but it’s also patently incorrect.

It’s easy to imagine how to falsify evolution.  If you find human skeletons in the same soil layer as dinosaurs (or layers beneath dinosaurs), that would destroy the hell out of evolution, and the observations we’ve already collected.  Likewise, genetic evidence of irreconcilable differences between humans and Neanderthals (or Denisovans) would break all our assumptions about evolution, too; so would major genetic differences between chimps, monkeys, and humans.  Wild variations or inconsistencies in radiometric dating of observed fossils would also nullify evolution.

On the observation side, it’s easy to observe evolution in action, and predict the future based on it.  We see micro evolution in the lab.  We also are able to test particular tenets of evolution, such as the assertion that, in evolution, species gain and lose attributes based on how advantageous or disadvantageous they are.  For example, as humans’ skulls continue to shrink, we should see our need for wisdom teeth to go away.  Similarly, as our diets and other internal chemistry changes, we will probably see that humans will lose the need for their gallbladder, appendix, and maybe one of their kidneys.  As it turns out, we see about 0.09% of humans born without a gallbladder, upwards to 35% of people never develop wisdom teeth, and about 1 in 750 people are born with only one kidney.  These features may again become more useful in the future, so it would be interesting to know how this plays out, but the hints we’re seeing provide corroboration, despite the fact that we are stuck only being tentative about their likelihood.

We live in a tentative, probabilistic world, but that does not mean we can’t be successful.  Every decision we make comes after an assessment of potential outcomes.  We have an internal database of empirical evidence to support our decision.  As it turns out, our database may contain faulty or incomplete data, and we may see outcomes we did not predict.

Science relies on continuous collection of empirical data and refinement of models to improve our understanding of reality.

Over the past few centuries, as the roles of scientists and philosophers have become more concrete, it has become the job of the philosopher to tidy up the scientist’s work.  Where the scientist lives in a world of induction and empiricism, the philosopher must work to convert that into deductive postulations.  Karl Popper attempted to transform induction into a deductive process.
The fact that one can be wrong, but still remain scientific, is ok in itself – that’s one of the cornerstones of science, so long as one refines their hypothesis to fit new observations; however, rigid adherence to a collection of principles risks the very thing Popper tried to avoid – replacing scientific tentativeness with ideological dogma.

I think it’s better to simply ask “do the implications of this hypothesis agree with reality?”.  If they don’t, you’re wrong.  In those terms, we give away blind certainty, and in exchange, we get higher likelihood of being correct.

And as to the question of will the woman be in front of the market at noon tomorrow?  Probably.


Author: Tim...Stepping Out

Tim Stepping Out

1 thought on “Another Shot At Falsifiability”

  1. There is a tendency to confuse truth and reliability in discussions of science. The whole purpose of science is to make predictions, otherwise it would be a sterile intellectual exercise. Even our exploration of the origins of the universe is not fueled by mere curiosity. It is also to get a deeper understanding of the way things work and why things are the way they are so we can better anticipate what will happen later.

    No scientific measurement is “true” or false” and I am not just talking about measurement error and her sisters. If one were to attempt to measure the length of their dining room table, employing more and more precise tools (yardstick, steel measuring tape, laser interferometer, etc.) one would quickly find out the the two ends of the table are neither smooth nor are they parallel (and if the table splits in the middle, that length becomes variable also. So, what do we mean by the length of the table?

    The same kinds of difficulties exist in every scientific measurement. There are limits to how well we can measure things and no limits on our suppositions, e.g. is the 2 in E=mc^2 exactly 2 or approximately 2? Answer, we can’t tell. Really, we would never be able to prove it to be exactly 2 and not 2.000000000000004 or something. We resort, instead, to determining how reliable such descriptions are. We say things like, this equation works for numbers/values/parameters in the range of …. This is why Newton’s Laws and Kepler’s Laws are not “incorrect,” their “correctness” is limited, as are all laws of nature because we discover them, not invent them.

    So, the reliability aspect is one that shows the humility of science and would be totally lost on naysayers as they already emphasize that “science thinks it is oh so correct” as well as saying things, like “you don’t know … absolutely” in contrast to their beliefs which are absolutely knowable and correct in the fairylands of their minds.

    The problem is one group is using a reasoning mental faculty and another is using a trust mental faculty and never the twain shall meet. Neither side in this debate (which is really not a debate) will accept the other’s tools, world views, etc. so the dispute(?) cannot be resolved. That doesn’t make it any less fun to poke holes in each sides attempts to hop the fence and use the others tools, though. Them folks are fair game!


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s