Read to Forget
I read to forget. Even when studying or working on papers for a PhD, I approach texts with the same mindset: I'm not a storage device that needs to save all bits of information. I am more of a system of Bayesian beliefs, constantly evolving and updating in small, incremental steps.
I remember co-workers highlighting large chunks of text, sometimes 40%. That doesn't make sense to me. We can only read a text once, given the number of compelling works and the limited time available to us. So, I read to forget. When I start reading, I'm prepared to lose 98% of what's in front of me. From most texts, I only want two things: First, I want it to subtly alter my thinking, an incremental update that moves me towards a refined world model. Second, I want to pull out a few key pieces of information that I might use later in my writing. For instance, if I come across a well-written methodology section in a paper, I’ll save that. Reading should stimulate my thinking and produce new ideas. I've found myself reading a paper, pausing midway, and immediately experimenting with some variation of the algorithm described, leading to new ideas or even a new paper.
If a non-fiction text doesn't spark new thoughts or actions, it may not even be worthwhile reading. Anything beyond that clutters my note-taking system. You can't possibly keep track of everything, nor can you work with hoarded pieces of information.