From what i understand this argument is using the intuition that best explanations are replaced at one point and we eventually realize that they do not accurately describe the world. However, the fact that our “best” explanations can be replaced does not mean that we should support anti-realism. Could we say that we are getting closer and closer to the truth? Phlogiston was replaced by oxygen because anomalies arose and oxygen was a better explanation to explain those anomalies. In the caloric theory, while caloric does not exist, there was continuity across the theories, energy flows instead of a fluid substance. There is something we were getting right in the old theories that gave it its success and now we have better theories that have some continuity from the old. This leads onto structural-realism
Link to original
Structural realism: theories give approximately correct descriptions of the structure of the world (not the entities by themselves). Equations and models capture relations
• example: 19th century electromagnetism was not correct about the existence of ether, but the equations give approximately correct descriptions of how electricity and magnetism relate
- There are certain structural relations in the caloric theory that remained true in the current thermodynamic theories that there is flow from hot bodies to cold bodies
Of course, there are objects out there. But we learn from the failures in the history of science, something that continues is the relations …
Structural realism is also known as the best of both worlds idea (by philosopher John Worrall)
- the idea that both no-miracles-argument and pessimistic-meta-induction are both correct. As pessimistic-meta-induction tells us that we cannot trust what our theories tell us about objects and their properties
- what theories do get right (approximately) is the structure of reality i.e. the relation among objects
A problem that arises is filling in the gaps (what is structure? where does math go?) = the argument that there are continuities in particular cases rests on deriving equations of old theory from new in particular limits. Does this work beyond physics and some examples in chemistry?