Laudan's piece is annoyingly jargon-laden and wordy. Nevertheless, he makes some very useful and important points about Kuhn.
I. Units of Scientific Change
Kuhn describes scientific revolutions as involving a shift from one paradigm to another. But what's a "paradigm"? Laudan suggests that Kuhn sees paradigms as involving the following three levels or components. (Somewhat annoyingly, Laudan's terminology and even abbreviations fluctuate during the article; I've tried to include all his labels below.)
1. Ontology/Facts/Theories: "a conceptual framework for classifying and explaining natural objects" (140). That is: what kinds of things there are.
2. Methodology: "methods, techniques, and tools of inquiry for studying the objects" (140).
3. Axiology /Values: "sets of cognitive goals or ideals" (140). Laudan takes this to include things like views about which properties are desirable in a theory, how they should be weighted, etc.
The "holist picture" is the idea that all of these things come together in a package deal, and cannot be separated. Thus, on (Laudan's interpretation of) Kuhn's picture, a paradigm shift looks like this (a beautified version of Laudan's Fig. 1, p. 143):
One problem with this picture: paradigms include values and methodological principles, which are the kinds of things one might use in choosing which theory to adopt. It may thus appear that there's nothing outside the paradigms to make the choice of which paradigm to adopt a rational one. As Kuhn himself writes (Kuhn 109-10, quoted in Laudan, 143): "In the partially circular arguments that regularly result, each paradigm will be shown to satisfy more or less the criteria that it dictates for itself and to fall short of a few of those dictated by its opponent."
II. Kuhn's model vs. Laudan's model
Laudan highlights two features of Kuhn's model of theory change:
1. Holistic: ontology, methodology, and values come in a big package that can't be divided up; in theory change, all the components change at once.
2. Hierarchical: values determine which methods are appropriate; methods determine what one takes the ontology and facts to be.
Laudan proposes instead a model which is:
1. Anti-holistic: some components of a paradigm can change while others remain constant.
2. Reticulational: all of the components affect all of the others, i.e. the model is non-hierarchical. ("reticulated" =def "having the form of a grid or network")
III. Examples of Piecemeal Change
Are there historical examples in which one of these pieces changed while others stayed constant? Laudan says yes, and offers these examples.
1. Methodological change, 1800 - 1860: a shift from enumerative and eliminative induction to "the logic of hypothetico-deduction" (151). Laudan claims that this methodological shift wasn't accompanied by any particular revolution in theories or values. (But the methodological change leads to an ontological one: an acceptance of unobservable entities. At least I think that's what Laudan is saying at the bottom of p. 150, but it seems odd, since corpuscularianism was already widely accepted in the 17th century, two hundred years earlier, and seems to involve a commitment to entities too small to see.)
2. Axiological change, 1900s: shift from certainty as a goal of science to mere plausibility or probability. (Laudan 152-153; though I wonder how well-defined this shift is, e.g. Locke has a lot to say about probable reasoning in the 17th century, though it's true he seems to regard it as a pretty poor substitute for actual knowledge, which on his view requires certainty.)
IV. Laudan on Kuhn's Critique of Methodology
Laudan discusses and criticizes the following four reasons offered by Kuhn for thinking that theory choice isn't rationally determined (you could think of these as four arguments for "ampliative underdetermination," to use L's terminology from his underdetermination article). I'll briefly indicate the arguments.
1. ambiguity of shared standards (e.g. even if scientists agree that "simplicity" is desirable, they may interpret it differently)
Laudan's reply: not all standards are vague in this way. Examples: (i) theories should be internally consistent; (ii) theories should be consistent with what is known in other fields; (iii) theories should be deductively closed; (iv) theories should be subjected to controlled experiments [but this doesn't seem like a criterion that will very often decide between competing theories]; (v) "theories should lead successfully to the prediction of results unknown to their discoverer" -- (i) - (iv) all from p. 134.
2. collective inconsistency of rules (different criteria for theory choice may disagree with each other in particular cases)
Laudan reply 1: there are lots of sets of rules that are internally consistent, e.g. Mill's methods. [But: Mill's methods are for determining which correlated factors are causes; they aren't about deciding which of two competing, broad theories is correct. Moreover, surely Kuhn has in mind the set of *all* the methodological principles a community of scientists employs, which will typically include simplicity, elegance, etc. Third, what Kuhn has in mind is surely not that standards are literally inconsistent with each other, but that they can lead to conflicting results, e.g. theory A may be simpler while theory B has a wider scope.
L mentions this interpretation of Kuhn ("could have strengthened his case considerably . . .") on p. 136; his response is just that while it's conceivable that this can happen, we need more study to know how often it actually happens.
3. shifting standards (disagreement among scientists over what the standards are)
L acknowledges that this may happen, but says that mere disagreement does not know that there's no rational method for resolving those disagreements.
4. problem-weighting (different theories/paradigms may solve different problems: how to decide which are most important?)
Example: Daltonian chemistry gave up solutions to a wide variety of problems that had been offered by phlogiston theory and elective affinity. Someone who thought those problems were especially important might want to hang on to pre-Daltonian chemistry.
Laudan: Scientists do not typically give the most *epistemic* weight to problems that they regard as important in some other way. Instead, the problems with the most epistemic weight are those that test the theory most "severely."