Evaluating a research programme › When an Adjustment Counts as Progress

Michael BrenndoerferUpdated June 30, 20253 min read
PhilosophyAll chapters
Reading tools Notes & comments
Notes & comments
Part 9 of 12Science and Explanation
Chapter 113Evaluating a research programme › When an Adjustment Counts as Progress
Reading section 4 of 7 · When an Adjustment Counts as Progress · 3 minSection 4 of 7 · 3 min

Section 4 of 7 · 3 min read

Evaluating a research programme › When an Adjustment Counts as ProgressSection view · Sections in this chapter4 / 7 ↓

Lakatos distinguishes theoretical from empirical progress. A sequence is theoretically progressive when a new theory has excess empirical content over the preceding one. It puts forward additional claims that can be tested, including novel facts that the earlier account did not lead us to expect. Empirical progress requires that some of this excess content be corroborated. There can be a gap between the two. One can often identify what an improved theory predicts as soon as it is proposed, while it may take a long time to find out whether the prediction holds. Lakatos, “Falsification,” §2(c), discussion of progressive and degenerating problemshifts

In the orbital case, suppose the new calculation can reproduce what the earlier observations already show. That is a useful start. Suppose it also gives investigators a region of the sky to inspect at a stated time, and a way to tell the predicted body from several nearby objects. The result of the search can now bear on a claim they did not have to make to reproduce the original discrepancy. A verified body in the relevant location would still have to be identified and shown to have the appropriate properties. The illustration has done its job once it shows the difference between describing the data that motivated a change and using the change to reach an additional piece of evidence. Lakatos, “Falsification,” §2(c)

A sequence can go the other way. When a calculation fails, a group can introduce a new quantity chosen to recover the old result. The next discrepant case can call for another such quantity, with no claim about where else it can be measured or what else will change by a predictable amount. The revisions may bring the proposed laws into agreement with more records. If they do not add testable claims, that agreement does not show the kind of progress Lakatos wants. His account asks what each change has permitted the investigators to find that they could not have found by staying with the old version. Lakatos, “Falsification,” §2(c)

Judging novelty can require more than checking the dates on two documents. Compare two ways of developing the example. In one, the extra source of force is assigned whatever value will reproduce the orbit that was already measured. In the other, the value is constrained by observations of another system, and working out its consequences directs the investigators to a further feature of the original orbit. These choices determine which results the new account was designed to fit and which it puts at risk. Lakatos's distinction gives people a reason to record those dependencies, since simply adding another line of confirmed data cannot tell you which part of the new theory earned support. Lakatos, “Falsification,” §2(c), discussion of excess empirical content

Comments

Add to the conversation

A question, a correction, or an example from your own work.

No comments yet. Be the first to share your thoughts!

Newsletter

Stay up to date

Get new articles, book updates, and notes on AI and technology.

Unsubscribe at any time. Privacy policy