It's a shame how poorly web browsers handle bookmarks. In Firefox, for example, there is the ability to tag bookmarks and even a feature that suggests previously created tags to cut down on duplicates. However, they lack two features which I see as vital. First, they don't allow for convenient viewing of bookmarks on the basis of tags. The only way to see all of the tags is to open the bookmark manager. From the Bookmarks menu, the only tags shown are those most recently modified. Second of their problems, is that there isn't any recursion in the presentation of tagged bookmarks, even in the bookmark manager. Only a single tag can be used to specify the bookmark view at once, but what if the user wanted to limit the the display to bookmarks with two or more tags?
Often these don't pose a problem for users, because they only visit a few different sites which they either have opened in tabs all the time or available in a "quicklist" auto-sorted by the browser according to recent usage. However, when a user needs more access to his past reading, it can become difficult to pick out the data he is looking for from a large list of tags with titles that may not reflect the particular data he's searching for. When doing actual research, having to rely on searching by page titles (which sometimes don't have the pertinent relation to the contents) can be disastrous. While there may only be a few cases where a user runs afoul of the poor organizational structure, the times when he does will irk him and probably cost a lot of time both in the auxiliary work of finding data and in the time taken to get back in "the zone" to work efficiently---and this can even ruin a whole day's work if the distraction is sufficiently upsetting.
Still, there remains the fact that many people won't bother with bookmarks at all. They rely entirely on their browser history and web searches, probably never clearing their history to prevent the loss of their fragile knowledge network. While I would always encourage more active management of information, I tend to see that as the most natural approach. Organizing bookmarks takes time, and even if it's only a little, it's still valuable. Ideally, everything we read would be automatically organized into a well-structured web of information, that we could run through at will with the ease of accessing our own memories---the web would be an extension of our memories. However, science has not yet fulfilled this desire, and we are forced to make do with the tools at hand. I can, however, hope that there is data to be gleaned from user habits while surfing the web, and that those data might help to structure the information we view for later re-view.