I have written about uniformity before, such as here and here. This post looks at the need for a principle of uniformity.
David Hume’s principle of the uniformity of nature (PUN) asserts that unobserved cases closely resemble previously observed cases. This principle concerns the character of natural populations based on a sample as well as the sequence of natural events in the future based on past observations. Because of this principle, science can count on uniformity when making inferences and determining laws of nature. So they say.
But PUN is unnecessary overkill. It supports all sorts of false generalizations. If swans in Europe are observed to be white, PUN supports the assertion that all swans everywhere are white. This turns out to be either a bad definition of swans or a false statement, since there are black swans in Australia. If it rains on your birthday every year for 10 years, does that mean it will always rain on your birthday? PUN says Yes.
The problem is that PUN endorses far too much. It endorses good and bad inductions. And it does nothing to distinguish good from bad inductions or to improve inductions.
Instead of PUN science needs good definitions and conditions. Whatever fits unambiguous definitions and meets specific conditions is uniformly the same – by definition. Good definitions are ones that delineate the essence and only the essence of something. Good conditions are ones that specify no more or less than what is necessary for something to exist or to happen.
Because the sun has been observed to rise every the morning for thousands of years does not guarantee that it will rise tomorrow – unless one defines the sun in such a way as to include the property of its rising relative to the earth. That may or may not be a good definition. It will take other observations and definitions to determine how good it is.
PUN should be discarded. Science works best with the right definitions and conditions, without a PUN.