Nat Friedman September 18, 1998 My bookmarks are a mess. I once thought that I could keep them organized; I thought that I could responsibly categorize URLs as I found them, building a hierarchy of web addresses as I surfed. I envisioned a glorious tree structure of well-classified URLs hanging off of my browser: a gleaming taxonomy of the hotspots in webspace. "Congenital Achromatism?" I would say, "Oh, that's under Medical/Diseases/Genetic/Vision." Instead, what happens is that when I come across an interesting URL that I want to remember, I add it to my bookmarks, but I generally don't have the patience to put it into the right category. The result is an amorphous blob of unclassified and ill-named web addresses at the toplevel of my bookmarks list, next to an incomplete but well-intentioned skeleton hierarchy of categories that I set up years ago when I was hopeful enough to think that I would keep things well organized. And that's OK. But there are other problems; sometimes, people send me URLs in emails, or on IRC, or maybe I come across them in documentation. I don't always want to check these out immediately. Sometimes, I want to file them away for later review. When I do go hunting for a web page that I know I'd seen before, I find myself wading through the huge list of unsorted crud that makes up my toplevel bookmarks list. Most of the bookmarks are ill-named and many are 404; I have no idea where most of them came from and whether or not they're relevant to my current search. What I propose is the creation of a universal URL repository. Instead of enforcing the idea of a hierarchy of URLs, it would treat all entries equally, with the option of superimposing a hierarchical structure on top of the collection of items in the repository. The idea is to store URLs like in a database: URL: http://www.mozilla.org/ Name: The Mozilla Web Page Longer Description: ... Found Where: IRC, LinuxNET, #linux Referred by: Jamie Zawinski (IRC: "jwz") Context: "Nat: The URL is http://www.mozilla.org/" References: (1) Mail message June 17, 1998 (2) Mail message August 3, 1998 (3) IRC, LinuxNET, #linux, August 5 ... Category: Hacking/Big Projects Most of the time, I want to search my bookmarks for a key phrase or something like that. "Let's see, what was that page Jamie was talking about?" "Let's see, what was that page about simulated annealing?" Most of the time, a hierarchy is unnecessary, even bogus. Better still, the URL repository could be responsible for indexing the pages stored in it, giving you your own private search engine. And of course it could automatically prune or mark 404s in the indexing phase. I would love to see something like this implemented as a CORBA object which could be accessed by all the applications that I use. I'd love it if my mailer would automatically recognize URLs that appear in the messages that I'm reading and add them to the repository, or look them up and tell me the last place I saw them. I would love to be able to query this repository with search strings, or by person, or by date. This is the kind of deep integration which GNOME needs.