I read papers depth-first recursively. I read the abstract and see how much I understand. If there's a lot of stuff I don't understand, I hop down to the references, find one of those papers, and try again. I do this until I get to a paper I more or less understand and bubble upwards.

I take pretty aggressive notes in Obsidian for each paper [1], which carries the benefit of being able to MediaWiki-tag definitions as I find them and build up a dictionary of terms I can reference.

I've never really seen the point of highlighting, it takes zero comprehension of the material to rub a marker over a page. I try my best to summarize each paragraph into a bullet. I figure that if I can summarize stuff accurately, I at least have some understanding of the material, and again this builds up a repository of notes I can read later (though I rarely do because I usually have a decent enough memory of the source material afterward).

Some day I will start sharing my archive of paper summaries for the world to not-actually-read, though I can't right now because they're kind of intermingled with personal notes that will take some time in order to decorrelate.

[1] I have actually been experimenting with Logseq lately, and I use Codex to synchronize back to Obsidian for the time being.

I highlight as a way to categorize my annotations. I highlight in Zotero as I go, and in the highlight's comment section briefly jot down why (e.g. something to follow up on, or whether this reminded me of something else it contrasts with). I dedicate a certain colour to "background references I should have", another to ~ "things I disagree with" etc., which I find useful when coming back over the paper to type up my notes.

In a sense the highlighting is just a way to localize my thoughts to a particular passage of the text, and the colours (or even highlighting at all) are secondary.

There's some considerable duplication of effort (notes in Zotero, then I type up notes in Obsidian, then also extract out some of those ideas into their own files). But, much like the recent posts about "outsourcing thinking" and GP noting that people sometimes do nothing with their highlights, I find that the work is useful for understanding and remembering.

Out of interest, why have you been considering Logseq?

I read the abstract and skim the intro before committing to a read. The authors have to convince me that they know the field, they think they’ve done something interesting, and I think what they’ve done is plausible. If it passes that bar, I assume the most adversarial possible mindset and look for holes in their methods. If their methods are junk, I may skim the conclusion just to see what kind of unfounded nonsense to watch out for in the future, but otherwise I’m done, and really most papers are done at this point. Papers in my field are mostly bogus, unfortunately. Every now and then, somebody uses plausible methods, and only then do I really bother to sit down and read the whole thing.

Hopping to the references right away doesn’t work for me. I prefer just going through the whole paper once and coming back. To me it’s analogous to reading a book in a language not familiar to me. You read it lengthy chuck of texts even though you come across words you don’t know. Then you can find the meaning of words and read the texts again for better understanding . This way I think is more helpful to grasp the whole topic and intention of texts faster.

I don’t highlight often, but when I do it is for the same reason I take notes. I never refer back to either one. But they focus my mind to stay on track so it cannot wander as much, in the moment, and prompts me to immediately reflect on or quickly reread what I think might be important or interesting.