Thanks for sharing these insights. I’m mostly bookmarking things when I come across them and then process my bookmarks and capture them to PDF while processing (adding tags, moving to the right groups) in DEVONthink. I’ve got a small script that I use to view the website (if it’s still available) instead of the PDF in DT. When enabled I can use that to simply copy/paste specific parts in an annotation file.
My main point is that I also like browsing or viewing the URLs / websites of captured PDFs inside DTTG (like when viewing a bookmark). I don’t want to create an accomapnying bookmark for a captured PDF for just that purpose, but haven’t found a way to do that yet (other than navigating to the captured PDF, clicking the link under Info and viewing them in Safari)
In light of the general discussion, I too like @rmschne, prefer clipping PDFs - it’s the most future-proof format IMHO and indeed images and layout are saved which helps me often understand an article / website better when I visit it later (sometimes years later). But there are definitely some problems: sometimes issues with glyphs can garble up the text layer, or copy-pasting text means you end up with a lot of hard breaks in lines. I’ve thought about capturing Markdown _and_PDF but don’t like more versions of the same content cluttering up my database.