The GraphTheory analysis... Links have to be content because: : there is no way of deriving where the edges are in the hypertext graph from the nodes alone. : thus, they carry information and hence are ''content.'' If you can embed information in links, you have be just as careful where you put them as with text. You can be too spartan or too babbly. ''But we don't embed information '''in''' links. Links '''are''' information. ''No, they are meta information.'' They should justify themselves on a page just as the words do, no more, no less.'' You embed information in a link the same way you embed information in a page. A link is an unordered 2-tuple {from,to}. As such, the structure of the link is fixed, the structure carries information ("a link is undirected between two points") independent from the actual information of which nodes it connects. Similarly, a page by itself is a node with a title and a (possibly empty) set of referants. You embed content in the page by writing something on it. This is getting ''very'' esoteric. I apologize. I agree with your conclusion about justification, though. ''At least in the case of the WWW, links are not a 2-tuple; they are {from, to, link text}, and, if you're feeling generous today, {context} of the link as well. Evidence of the theoretical importance lies in the often ironic use of links, where the '''to''' contradicts the link text, or subtly undermines the context. They are definitely content; done correctly they can completely shift the meaning of an otherwise-unlinked block of text.'' ---- As far as I know, the task of language itself is to refer to concepts and links between them. If a news article writes about a recent political event and its effects, the article refers to the link between the events and its effects. Therefore the link is the ''object'' of the article. And as by magic the two objects of event and effect are joined by a third object. This, to me at least, is the magic of writing. If you refer to two concepts and you can find a name for the link between them, all you need to do is promote the link to an object, just by making it your subject. -- StijnSanders ---- I have found it very rewarding to systematically create new links which reference a page where the entry previously referred to is in a way which did not make a link. Example: SWIG does not make a link, but SimplifiedWrapperAndInterfaceG''''''enerator does link to the page describing SWIG. The ''reward''? I have learned a lot from the pages I would not otherwise have visited. -- JohnFletcher ''Yes, very rewarding. An excellent way to ReadTheWholeWiki. Well, almost.'' ---- See also HyperTextHistory ---- CategoryDiscussion