When it comes to linking in future web systems, people should look to the past, according to Geoffrey Bilder's presentation at the recent ALPSP conference in Old Windsor, UK
Much is made of the new ideas in Web 2.0. But some of them may not be as new as you might think. At ALPSP’s international conference in September, Geoffrey Bilder of CrossRef illustrated this with the story of the ‘simple’ hyperlink.
‘The World Wide Web introduced the concept of the link and it was very powerful,’ he said. However, there was a problem: early browsers and web systems were very complicated. Over time, in an effort to simplify things for users, the WWW concept and its browsers have lost many of their early functionality, he explained.
And now, that very functionality that was lost in the efforts to simplify is starting to be required again. ‘So many Web 2.0 and semantic techniques are trying to put back in what was initially in early hypertext systems,’ observed Bilder.
One example of this is link persistence. Early systems had it but it was rejected because it required some sort of centralised system with huge maintenance overheads that are not really necessary for much of the information on the web. To cite journal articles electronically, of course, link persistence is very important and this is how CrossRef came about. ‘I say we are like a post office box that redirects mail,’ said Bilder.
The simplification process of the web also limited the possibilities of the hyperlinks themselves. For example, ‘early hypertext systems never assumed that links were unidirectional,’ said Bilder. ‘Today you can’t find out who cites your page without lots of work.’ This is another area that CrossRef has been working in, adding an overlay to alert people when they have been cited.
Today’s web links do not allow multiheaded and multitailed links either. The former is where something – such as a word or chemical formula – might appear several times in a document but all link to the same definition as a single link. The multitailed link is where one link leads to multiple places.
An example of this might be where a link would take you both to an article in the IET digital library and to the same article on IEEE Xplore. A similar concept is the version aware link, which Bilder believes is going to become increasingly vital, particularly with books. As new versions of books come out, links might be updated to point to the latest version. But if you have cited something in Version 4 of a book then you don’t want the link from the reference to take readers to Version 6. And this potential problem increases greatly when you factor in the rapid changes in wikis and blogs.
There is also sometimes the need for conditional links, which will only work on certain dates or if you have been authenticated, for example.
Another issue is that there is no way to qualify how items are being linked to. ‘Links are votes and are very important,’ noted Bilder. ‘However, some links will be negative and should be categorised differently.’
‘We have to start paying attention to links and anticipate problems,’ concluded Bilder. ‘If you look at old hypertext systems you will find interesting examples of how it was done in the past.’