One might even conclude that the inability to predict is an indication of a lack of scientific support for the reasoning. Yes, both environmental pressure theories seem logical, but that doesn't make them true. When you have a logical theory to cover every base you can't go wrong. If they get bigger, the theory works. If they get smaller the other theory works.
What if they changed size for a different reason, but we never explored it because we already accept the catch all theories. The presumption is that they had to change for the better, but we never assume that extinctions were the result of evolution gone awry.
Yes they are insightful theories and might be correct, but I always question theories that can't be disproven.
The classic way to disprove the evolutionary time line is to find the classic violations: monkeys fossiles with dinosaur fossils for example.
Of course such findings have not been found, but the absence of evidence is not the evidence of absence.
The things we know in science are by its nature, unsure: they can be overturned tomorrow by a good experiment, and unexpected data.
The errors of the past we have confidence that they are errors because of past good experiments and data that was once unexpected, and used to develop a new theory that explains or permits the known data.
The Lodka Volterra equations explain year to year variation in species populations. Back some 40 years ago, it was presumed that eventually all the swirly motion would damp out and you had some kind of equilibrium point.
_ _ ________
/ \ / \_/
/ \_/
After the Chaos work of the last 35 years we have a different understanding: that there is no equilibrium point to find, that swirly motion of Lodka Volterra equations is the natural state of things. That makes predicting an equilibrium point impossible (or trivially easy since they are all equally wrong!)
The problem is that the new theories are hardly ever simpler than the old ones, so there is the concern that we may be adding the equivalent of ‘retrograde motions’ to the stars to explain the variance of crude or messy data to our beautiful theories.