László Kozma


Home | About me | Projects | Links | Weblog | Ideas


Link to content [more ideas]

Perhaps the biggest problem with the world-wide web is the deterioration of hyperlinks. More and more services nowadays (such as URL shorteners) add additional layers of redirection, thus making the hyperlink system even more fragile. Suppose I write down some ideas and link to an essay that further expands on the topic. In a few years when someone reads my notes, chances are that the essay I linked to no longer exists at that location. The server might have gone offline, the domain might have expired or maybe the site was just rearranged and the essay is still there at a different address. Even if the page is still there at the same address it might be that it was substantially rewritten, perhaps the author changed their opinion and now it no longer supports my ideas, so I wouldn't link to it anymore. However, the original page might still be there somewhere in the original version, someone might have mirrored it, it might still reside in Google's cache or on the Internet Archive website. I could link to those services, but I'd rather the viewers visit the author's page directly.

Are there any 'best practices' to ensure the consistency of links ? Besides the above mentioned services, one could install some link checker software that periodically verifies that the linked pages still exist. It would be easy to include the option of checking also whether those pages have changed. However, link checkers generate unnecessary traffic. Of course, one could just copy the whole content over, making the whole linking unnecessary.

So is there some halfway-solution between linking to the address and copying full content ? Perhaps one could link to part of the content: find those phrases or words which are otherwise rare but characteristic to the text we want to link to. This is easy to do with simple automatic methods. Then one could use Google's "I'm feeling lucky" search, which automatically redirects to the first search result. If the query is accurate enough, chances are that it would redirect to the given article, even if the original address went offline and it is now hosted elsewhere. However, this is not a very practical solution: links could be hijacked by spam sites that would use 'just the right phrases', Google might change their service, this would not work with images, etc. ... Any better ideas ?

Note: The Xanadu Project had similar goals but it never reached mainstream popularity.

An example Google redirect link to my entry on reverse autocomplete.
(http://www.google.com/search?hl=en&q=bidirectional+autocomplete+reverse+kozma&btnI=I%27m+Feeling+Lucky)



blog comments powered by Disqus