It sounds like Gould is arguing that natural selection has not been as
important as often thought, but that simple genetic drift and variation
can account for what we see as evolution. Over time we would expect
a wider range of life forms because the random changes to the genome
introduce more possibilities.
At the extremes in any given parameter, we would expect a gradual widening
of the curve, so the effect is that the extreme gradually increases.
This would be true of every parameter: body size, brain size, brain
convolution, visual acuity, jumping height, temperature tolerance,
speed, etc. If the model is a random walk, then the rate of increase
should be inversely proportional to the square root of the elapsed time.
However it is not clear what the right way is to measure the parameter
in order to observe this.
It is true that some parameters, like the mass of land animals, have
not exactly followed this model. There are no land animals as big as
the largest dinosaurs. However it is possible that this is merely a
random fluctuation. Given that we are on the extremes, the population
size is relatively small. It is not too unexpected that there should be
occasional backwards steps even though the trend is towards increasing
size. The model would predict that eventually there would be dinosaur
sized animals again, and even bigger ones.
The problem with this model that I see is that it is hard to know how
to factor in the limitations imposed by the universe. It could be that
there is an upper limit to body size beyond which life forms based on
the kind of genetics used today just won't work. So the engine of
variation has to work harder and harder as it pushes up against this
limit. Variation is still occuring, and marginally larger life forms
are being produced, but the rate of progress in pounds per year seems
to have slowed.
The same complication arises in considering how the phenomenon works
in a competitive universe where natural selection is occuring. We
can still choose to focus on the random nature of the changes. But
the rate at which externally visible change happens will depend on how
easy it is for changes to work in that direction. If the change is one
which causes the animal to get eaten more easily, the change engine has
to work very hard to move that way.
So ultimately I don't see this point of view as really providing any
insight into the causes of evolution, since it just pushes the effects
of natural selection off into another part of the model, where they
serve to limit the rate at which changes of certain types can occur.
It doesn't mean that selection isn't important; it is just a matter of
what you are paying attention to.
Hal