The opening is unfair to Kuhn. He characterizes most of science as 'normal science', which is constituted by 'puzzle solving', and is therefore a form of cumulative progress. And it is clear that the candiates in a revolution are only those that already address a majority of the solved puzzles adequately. The result cannot be anything other than cumulative progress, with the occasional step sideways taking us a little bit back. He does not posit a driving force, but he does not rule one out.
Kuhn was refining Popper, not denouncing him, basically by adding detail to the overly-abstract notion of falsificationism. He emphasizes over and over again that they 90% agree, but that the bald notion of a single judgement criterion does not fit with historical reality. Normal science is falisificationist in principle, even though it fails to be so in practice. And revolutionary science is done by normal scientists. So by sheer force of social psychology, it inherits all the biases of normal science that can be accomodated without leaving too few candidates.
Feyerabend was right to characterize Kuhn as historicist, though perhaps in a weaker sense than dialectical stances promote. Any adopted method must be somewhat teleological. Dogma needs a selling point, because freedom is traded for something. That something, in the case of sciences, is the quasi-mythological promise of eventual convergence. (I put in 'mythological' because I think it takes on some religious character, not because I doubt its truth -- I am a convert.)
You can see punctuated equilibrium and the selfish-gene theories as revolutions within Darwinism. The original theory involved a notion of gradual accumulation of adaptations and a sort of 'convergent design as sculpture' that became less tenable when we realize how discrete genetic information was.
The modern version of plate-tectonics based upon subduction is also a (slow and low-impact) revolution that eradicated the two contending camps of geological theories that had been at war for centuries and replaced them with a genuine peace. So one might say that plate tectonics was the foundational revolution in geology, before which it was not a fully-formed science, but still had two rival schools and no single paradigm. The assertion of this paradigm was underway during Kuhn's life, but it took decades. It finally came together during the 1970's, after the publication of "Scientific Revolutions" and in the face of famous holdouts with sterling reputations.
The creation of the discipline of psychology started an ongoing revolutionary process, which we are still in the middle of. Freudianism almost succeeded at being a paradigm briefly, and so did Behaviorism. So we have firsthand experience currently of how revolutionary processes take place and how normal science manages to assert itself during the process and coalesce shared results across paradigms, despite the lack of genuine traction.
To my experience, the history of psychology to date backs up Kuhn. But it also suggests that the process has a sort of fractal, self-similar nature to it, where potential paradigms create local sciences with miniature revolutions as they seek to expand themselves onto contested territory to gain traction in the larger ongoing revolution.
I also would contend Kuhnianism as a sort of 'percollational* synecdoche' (where the process writ large is made up of approximations to itself on various levels, so that each level provides a modelling metaphor for other levels), is an excellent model for politics, group dynamics and individual human thought -- a moderate improvement on sheer dialectics.
* the word 'percollation' is Benoit Mandelbrot's word for processes like the growth of a sponge, which form self-similar solids within a large range of scales. (We probably need a better term, less open to coffee puns.)
