Because I want to keep them spatially and temporally local; they are relevant to what I am doing and/or my current interests.
Because I don't have time to read them now and I figure keeping them open will lead to a higher likely-hood of them being read. Consider: When does misuse of "cache" effectively become RAM, become the same as disk storage/access, etc.
Because I don't want to say no, because I don't have a framework for evaluating media holistically against each other. Because there's no friction preventing, or significant consequences to, opening of new tabs. Because I as a human being cannot keep a model in my head for sometimes hundreds of pieces of content, their importance, their topics, or the date they were discovered -- and thus cannot effectively make decisions on how to address the overwhelming tab monster.
Yeah, probably. What are you playing at ... ?
Bookmarks are a black hole. They simply offers a slightly greater promise (and a smaller, albeit still unmaintainable search space) of finding a resource again than searching the web directly. Most context (other than page title and an optional category) is lost. For instance, there is semantic value encoded in having multiple tabs open -- their mutual existence can add value to the equation, as a reference. For instance, imagine translating a Chinese document and having a "dictionary" tab, a "stroke-order" tab, a "pronunciation" tab (not all resources are best at all 3 tasks). A bookmark does not sufficiently encode these relationships or their co-related value.
Because browsers are victorious by competing on standards and performance, foremost, and providing the best experience to the most users. Some features (like a directed graph of your entire browser history) may require fundamental changes to the browser's internals and negatively affect performance. It's unfortunate, because browsers impact a lot of people, but (disregarding the audience multiplier and assuming building-for-self), designing browser features has a low return compared to the time investment and the programmer has to address tangential concerns. It's like disassembling a car to install a new engine, and running diagnostics on all the other components to ensure there's no oil leak, versus installing NOS (yeah, it can leak, there are occasional fires, and sometimes catastrophes, but ZOOM ZOOM!). Besides, NOS allows you to more quickly achieve and compare speed (analogy: approaches) and get a feel before having to choose an engine design. So, this is why I'm not starting with the browser (although I do have a side project for prototyping these ideas within an unperformant webpage, rather than through the browser, proper). That said, the correct minimal solution shouldn't depend on an architecture (will only be enhanced by it), so we should be able to prototype it elsewhere (in a less-friction environment, lower development-cost environment).
Minimally, it would look like a simple, centralized google spreadsheet with a list of links, their titles, a rating, and the date completed. I would thus sort them according to my interest. If manually adding a link + desc to the list is too much friction, the resource likely isn't work it.
Chances are I don't want to read everything on my list. By making the document public + shared and asking others I respect what media on my list I should skip, or by asking for recommendations for alternative, superior resources on the same topic, I can achieve a superior outcome.
Mozilla Firefox has tab groups which I utilize (despite their implementation being graphical and frustrating).
I use tab groups as workspaces to manage collections of tabs that I actually do need open.
I have 3 sets of tab groups --
Image 1: a screen capture of the above tabgroup collections
I do not have more than 10 tabs open in a workspace.
Image 2: a screen capture of tabgroups overlapping and in chaotic order
Finally, I have one instance of chrome open where I have the following applications always pinned: