An interesting phenomenon I observed while I was updating list of books I've read on Goodreads, is that humans discount the rating of books they read later in their life. The reason is not so complicated. When a person develops a habit of reading, the marginal learning of each book is high, but as he gets on reading and has accumulated some concepts over time, the marginal learning starts plummeting. The problem is that this trend is not consistent. For instance, if a person ventures into a new domain, the marginal learning will once again be high, and if he returns to a field he has already explored, it will be low. Due to that, we cannot use a formula to discount the earlier ratings of books.
The phenomenon I described earlier applies when ratings are made at the time of reading. However, if you ask a person at any point in time, to rate all books he has read, he now might rate earlier read books lower than what he would had he rated it on time of reading. This is because when a reader rates a book in retrospect, the book's learnings seem smaller relative to the accumulated learnings at the time of rating.
This leads to some interesting questions:-
- What is a more appropriate way of getting rating for books
- Does this phenomenon also applies to other things like grading. e.g. a teacher starts grading papers, and the earlier papers get graded higher than the last ones, given same quality of paper attempted?