The answer to your question is it depends entirely on the reader and the reading material. Let me explain with some background, and the results of my research.
The first thing you need to be aware of with respect to reading speed is that each person's aggregate reading rate is is influenced by a variety of factors. It can vary on the time of day, the ability of the reader to focus (or not focus), the complexity of the material, how the reader feels about the material, and certainly many other factors. Ultimately, the amount of time it takes the Kindle to hone in on your reading speed will vary from person to person and from book to book - hence there really isn't a singular answer to your question.
That said - I have spent virtually all of my free time over the last month reverse engineering the way in which the Kindle determines reading speed. At some point, I will write up a full explanation - but to focus in on factors that specifically relate to your question - for simplicity's sake, let's assume you're starting with a brand new kindle. Here are some basics about how the Kindle algorithm works.
- Every time you advance a "page" in the kindle, it captures the length of time you were on the page, and the number of words that were displayed. From this it calculates a words per minute (WPM) reading rate that we'll call a "sample".
- If the very first sample is less than 900 WPM, it keeps track of that sample - otherwise it discards the sample.
- Each time that you advance a page, the Kindle checks the new page level calculated WPM against the 900 WPM limit, discarding those samples that exceed the 900 and retaining those that don't.
- If this is the first book you've read on the kindle, or you have manually reset the reading timer, it will continue this until it has 15 samples.
- If you had already trained the kindle with your reading speed before this book, it will use a global WPM rate as a proxy for this specific book until it has sampled enough pages such that the total number of words read exceeds 600. At that point it switches from using the global rate to using a book specific average rate calculated from the samples that were kept.
- At this point, the sampling continues in this manner for the remainder of the book, throwing out samples that exceed that 900 WPM limit, and using some statistical techniques to only include the samples in the book specific rate calculation that meet the statistical criteria it uses (hint: it treats the samples as components of multiple normal distributions and moves the samples into and out of those distributions based on new observations - it gets fairly complex).
- Each time the book specific WPM calculation is updated, it uses that data along with some other details to estimate the remaining time in Chapter and Book.
So, ultimately, as answer to the question you posed: There is no "normally" that makes sense and there is no "correctness" of the reading speed that can be assessed. The reading speed is "historically" accurate from the very first page, but how fast it hones in on what an individual feels is representative of their reading speed
depends entirely on how consistently (from a WPM perspective) the reader actually consumes the material. It ain't magic - it's just statistics.
Hope this helps in some small way.
