On Against Method (1975), the introduction through ch. 5 and ch. 15-16, featuring Dylan, Wes, and Seth. This book that began as a joint call-and-response project with our last author Imre Lakatos, but then Lakatos died and couldn't respond to what Feyerabend explicitly wrote roughly in the form of a series of letters to a friend.
Feyerabend agreed with much of Lakatos' analysis of how science progresses but denied that this progress should be described as rational and that it can be captured via any single method of advancement. Through lots of case studies of brilliant scientists, F. tries to show that these figures violated any rules that philosophers of science have identified. It's not that F. thought that there was truly no rationality to their genius, or that any way of doing science is just as good as any other way, but he thought instead that the measures they took were so divergent and dependent on individual circumstances that if one had to look at all of them and generalize to a single scientific method, then the overall theme would have to be "anything goes."
So F. was trying to break more thoroughly from Popper than Lakatos did, with a resulting view that's more like Thomas Kuhn's, agreeing with Kuhn that there's no way to critique a scientific enterprise from within that enterprise, because any new, potentially countervailing evidence can be discounted from within the system. Each system involves a conceptual scheme and can't break out of that.
Like Lakatos (and Popper), F. had strong political motivations for his views, and felt that any pressure to conform to an existing system negatively impacts freedom. Scientists need to just engage problems as they see them and not pay attention to extraneous considerations (like "that's impossible according to current theories") that would end up impeding progress. In the abstract then, this sounds like it would not (as Popper was concerned about) rule out "pseudo-science" in advance, because there's no principled rule for distinguishing science from pseudo-science across the board. Instead, even the most crazy-sounding old theories (flat Earth, voodoo, the Genesis creation story, etc.) may have a grain of truth that can be usefully consulted or revised or serve as the inspiration for a future, predictively fruitful explanation. More practically, it's not justified, for instance, for us to categorically dismiss non-Western medical techniques just because they're not part of what has been approved by what we take to be scientific medicine. Particular alleged remedies need to be evaluated individually; we may well find that the ancients (or foreigners, or something otherwise outside of the accepted mainstream) knew something we've lost. Our knowledge and our culture more generally need a generous helping of epistemic anarchy to thrive.
Does this really constitute a distinct, positive position between Kuhn and Lakatos? Or is F's book just a screed against doing philosophy of science at all as opposed to merely learning the history of science? We at PEL certainly use something like this principle in thinking about the history of philosophy, finding it useful to explore often archaic views with which we as moderns fundamentally disagree not just for their intrinsic interest but because we feel that looking at these brilliant weirdos from the past may give some useful, historically removed perspective on philosophical problems as they are currently conceived.
Wes brings up our past discussion of Donald Davidson on conceptual schemes that might be useful for you to review.