When Charles Darwin was studying barnacles, he realized that the division of particular barnacles into species, and into higher taxonomic classes, was to some extent arbitrary. When does the variation within a species cross into variation across species? There are some general rules to be applied, but just how the rules should be applied to particular cases can get thorny enough to become a taxonomer's judgment call, in the last case an arbitrary decision. And it is just this interpenetration of intra- and inter-specific variation that makes evolution possible. If there were an uncrossable chasm between species, evolution would be impossible.
Science is another general category whose ultimately undefinable frontiers enables evolution. What sorts of propositions and activities should properly count as scientific? Evolution of such a category is not likely to happen through wholesale redefinition, but instead through incremental shifts at the frontiers. One notion of the frontier of science is where new scientific propositions are created or where their truth value is in an active process of determination. This frontier corresponds roughly to the regular introduction of new models of automobiles. This is an evolution of the various things that we class in the category of Automobile, but it is not a change in our very notion of what makes a thing an Automobile or not. The rise of electric bicycles is a bit more interesting, as it puts in question the boundary between bicycles and motorcycles. The evolution of the category Science is pushed where there are propositions or activities that we're not quite sure should properly called scientific.
A collection of phenomena becomes recognizable as a pattern, can be seen as an object, when it repeats itself often enough across a long enough time, which then requires some regularity or stability in its environment, the associated phenomena that are causally intertwined with the object. As conditions shift, the object responds, evolves. Science is an evolving category in just this way. For example, the use of automated experimental observation and mathematical calculation and deduction have raised the question of whether scientific activity is properly limited to human activity and the acceptable degree of indirection.
Rather than gambling on any particular forecast of future conditions, a wise strategy is to look at a family of most likely futures and to position oneself to be able to respond effectively to any of them. A major fork in the road ahead for the world is whether energy production can continue growing or whether instead we are facing declines as precipitous as the advances of the past century. Given the fundamental role that energy consumption plays across the full range of our patterns of living, a significant decline would have a huge impact on all of our institutions, including science. Already with impeded progress in experimental particle physics and space exploration one can see some of the early effects of resource limitations.
What kinds of science will we need in an era of declining resources, and what kinds will we be able to afford?