We dealt with the computer in computer science by showcasing some hardware access speeds in Chapter 2, Why Performance Is a Feature. Now, it's time for the science bit.
It's important to take a scientific approach if you wish to achieve consistently reliable results. Have a methodology or test plan and follow it the same way every time, only changing the thing that you want to measure. Automation can help a lot with this.
It's also important to always measure the use case on your systems with your data. What worked well for someone else may not work out great for you.
We will talk more about science and statistics later in the book. Taking a simple average can be misleading, but it's fine to use it as a gentle introduction. Read Chapter 10, The Downsides of Performance-Enhancing Tools, for more on concepts such as medians and percentiles.