Monday, May 3, 2021

The half-life of facts - it can be an investment problem


A provocative book from 2012 is The Half-life of Facts: Why everything we know has an expiration date by Samuel Arbesman. I was intrigued by the title and found the conclusions surprising. I know that science is rapidly changing, and we are constantly adding to our knowledge, but I may not fully appreciate how change, evolution, and facts are integrated to create the dynamic knowledge environment we inhabit. 

Knowledge is not just a process of addition. It is also a process of subtraction. We are constantly finding facts that are wrong and do not stand the test of time. Our increased pace of discovery means that more of what we may have held as true is now false. There is a half-life for knowledge. Different fields have different half-lives, and research is constantly being overturned. From Thomas Kuhn and his monumental work, The Structure of Scientific Revolutions, we have learned that science can have paradigm shifts. These can only occur when our facts shifts. Changing facts need new theories which creates shifts in science. From the book, a figure and table on half-lives. 



The speed of knowledge diffusion is increasing, but with so many facts and knowledge being created, there is the potential for hidden information. Knowledge may be created but not disseminated. 

This does not mean that we shouldn't learn facts. It also does not mean there is not truth, or all knowledge is relative. This is not a perception or feelings issue. It is a progress issue. 

The impact on finance and investment is significant. The academic studies of the past will likely be overturned. The ideas that drove investment decisions a decade ago may not be true today. Our common knowledge may be wrong. Assumptions concerning market behavior are subject to change. This half-life of finance facts creates potential for investment failure. For those that are gathering and assessing new facts, there is opportunity for profits, at least for a time.

No comments: