> Concerning falsifiability, let me first say that the nub of the matter
> is not that evolution hasn't yet been falsified, but rather that it is
> incapable of ever being falsified.
To put it glibly: Falsificationism has been falsified as a criterion
for scientific legitimacy.
This group has discussed falsificationism before, and every time we discuss it, we seem to forget how important it is to define the term "evolution." It is ESSENTIAL to define that term when debating falsificationism.
You can define a "theory of evolution" which: E1) Is impossible to falsify.
You can define a "theory of evolution" which: E2) Can be falsified in principle, but which in practice would require such an UNUSUAL observation that no one (not even opponents of evolution) would expect such an observation to be made.
You can define a "theory of evolution" which: E3) Can be falsified, but has not yet been falsified.
You can define a "theory of evolution" which: E4) Has already been falsified.
Let me expand on these a little bit.
E1) You might phrase it this way: "Evolution happened by natural mechanisms." It's hard to see how that could be falsified.
E2) You might phrase it this way: "Take a collection of known natural mechanisms [NM1, NM2, NM3 ... NMn]. Evolution happened by some combination of those natural mechanisms." If you phrased the theory this way, each natural mechanism in that list might be fairly well understood empirically (though not necessarily). However, if you are allowed to construct the global theory so that it can combine the natural mechanisms in almost any combination, the global theory winds up with very few empirical constraints, and is almost impossible to make predictions because of the complexity of the interactions. It would require a very unusual observation to falsify a theory phrased this way.
E3) You might phrase it this way: "Take a collection of known natural mechanisms [NM1 ... NMn]. Evolution happened by a combination of these mechanisms, with NM1 predominating under conditions C1 and historical periods H1, NM2 predominating under conditions C2 and historical periods H2, NM3 and NM4 predominating under conditions C3, C4, and C5 ... etc." The key here is to make specific predictions about when certain natural mechanisms are important and when they are not, and WHY they are more important than other mechanisms under those conditions. If you phrase your "theory of evolution" in this way --- as many scientists have done in the past and are doing in the present --- the theory can be falsified.
E4) Essentially, you can do E3, but in a way which has already been falsified. Note, however, that if a particular E3-version of the theory is falsified, it doesn't necessarily falsify an E2-version of the theory.
This doesn't just apply to biological macroevolution. This kind of theory construction also happens for the evolution of the very early universe, for the formation and evolution of galaxies, for the formation and evolution of stars, and for the formation and evolution of the planet earth (or other planets in the solar system).
For stellar evolution (but not stellar formation), we have very good quantitative models which fit extensive data, so some version of E3 is well-established.
For stellar FORMATION, the complexities of magnetohydrodynamics mean that we can't be as quantitative. Almost no one doubts an E2-version of stellar formation, but scientists who work in this field are constantly trying out improved versions of E3-theories (after their OLD favorite E3's turn into E4's).
For galaxy formation, it's a lot more difficult. We have a little empirical data on galaxy formation, but not nearly as much data as we have for stellar formation. What's worse is that several of the natural mechanisms believed to be important for galaxy formation (such as the gravitational attraction and conglomeration of "dark matter") have weak empirical constraints. When some of the important natural mechanisms have weak constraints, any global theory which uses them is bound to have weak empirical constraints.
Planetary formation, I think, is in better empirical shape than galaxy formation, but not so good as stellar evolution.
What about biological macroevolution?
I've seen plenty of pro-evolution books (and articles and posts) which spend most of their time explaining E2, toss out some specific examples of E3 which work on sub-sets of the data in order to bolster their case, and then make some grandiose conclusions about E1 --- all the time using the same phrase ("theory of evolution") and never explaining to the readers how they use that phrase differently in different places.
And I've seen plenty of anti-evolution books (and articles and posts) which cite a few examples of E4-versions of evolution, make some overly- simplistic and empirically flawed attempts to falsify an E2-version of evolution, and then complain bitterly than all the evolutionists are REALLY doing is preaching E1.
Folks, if you want to know which E3-type "theories of evolution" scientists are working with day-to-day in the lab, and how they deal with data which potentially falsifies their favorite E3-type (or even E2-type) theories, then the popular literature on evolution is the WRONG WRONG WRONG place to look. You've got to look in the professional journals. Popular books on evolution (both pro- and anti-) can be quite good (much better than the professional journals) at explaining the various proposed natural mechanisms. But the popular literature does a poor job of showing how scientists today wrestle with those natural mechanisms and the data in order to construct new E3's.
What most biologists work with everyday are hybrid E2/E3 version of macroevolution. Various specific E3's are applied in certain conditions and/or in certain historical periods where there are theoretical reasons for expecting it and some data to back up the choice, then falling back on E2 for the rest of biological history. As new data comes in, some of these E2/E3 hybrids are effectively falsified and discarded, while new ones are created which are (A) consistent with existing data, (B) falsifiable, and (C) not yet falsified.
This same sort of dynamic of creating and falsifying E2/E3 hybrids is going on in the study of galaxy formation. And this same dynamic USED to go on in the study of stellar evolution, until they were finally able to get enough data and computing power to really nail down a solid E3.
The observed examples of Punctuated Equilibrium in the fossil record successfully falsify some hybrid E2/E3 versions of the theory of evolution, but not other E2/E3 hybrids. Here's another way to say more- or-less the same thing: The E2-version of macroevolution is sufficiently weakly constrained (empirically) that it has not been falsified by the observed examples of PuncEq in the fossil record.
It's not too difficult to write down an E2-type "theory of biological macroevolution." And it's not too difficult to think of some remarkable observation which, if it were actually made, could falsify it. For example, we could imagine finding a species of field mouse somewhere in the Andes whose fossil record showed continuity with other species in the region, but whose genetic sequences and genome organization were so totally different from supposedly related species that there is simply no way it could have evolved via known natural mechanisms.
Now of course, IF that happened, some evolutionists would retreat to an E1-version of the theory. (And they would look for new natural mechanisms to explain that field mouse.) However, the fact that some people would retreat from E2's to E1's does NOT invalidate the fact that there currently exists an E2-version of the theory of macroevolution which is falsifiable, but has not yet been falsified.
On the other hand, the mere existence of an unfalsified E2-version of the theory of macroevolution doesn't say too much about the likelihood of it being true. All the "heavy lifting" gets done in nailing down the data and the empirical constraints on natural mechanisms, to move E2- versions of the theory into E3-versions. E2-versions get falsified (IF they get falsified) in the same way: by nailing down the data and the empirical constrains on natural mechanisms so tightly that it becomes clear that there just aren't any E3-versions which will work, forcing abandonment of the theory or a retreat to E1-versions.