Probably advocating this approach because writing and reading is often linear in nature. So when you are focusing on trains of thoughts, you are essentially building up linear arguments that can be used at a later time.
If you look at how the brain works, we have a network of linked neurons, and the ones that tend to fire together, wire together because that allows for easier repeat usage. But that doesn’t stop other neuronal routes from firing off.
The approach that I would take is a balance between a network and trains of thought, as Andy Matuschak advocated for. The key word in all of that is “Meaningful Links”, not “All Links”. So I don’t do one train of thought, but multiple, while aiming for quality.
The better question is “What is a meaningful link?” (I have my own theory on that).