Top to bottom is usually the best way to go, especially read first abstract & introduction, but don't be afraid to skip sections if they seem unclear or you jus

Top to bottom is usually the best way to go, especially read first abstract & introduction, but don't be afraid to skip sections if they seem unclear or you just don't understand what they are trying to tell you. These days, you could put the paper into NotebookLM, and generate a podcast, that will generate something friendly, but might be partially wrong or skip over essential stuff.