Skip to content

Simple Link Types (SLT)

I am throwing this out there for your comments. Can you blog replies or comment here please?

When we follow a link we can think of it as tearing a slit through the document and going to, or retrieving, another document or resources. Hence the fanciful name SLT.

It is very clear that link types can be useful in some instances and it is equally clear that they are not useful in all instances and the last few decades have demonstrated that there is a huge issue with building any kind of consensus of types.

So why not make the link types super simple and ‘dumb’ and let the smarts happen on the connections? Why not let anyone define a type simply by appending the type to the end of a link and optionally including a reference to a glossary/definition which the reader software can choose to parse or ignore?

These are questions and I am sharing this with you because you are very smart and may very well have thought about this. Shall we try to do something about this?

If ‘q’ is used to say query, as show here:

Why not simply add ‘lt’ for link type:

And an optional link to the glossary like this:

I would implement something like this in Author and Reader, where maybe in Reader some of the link types could result in info boxes as is done in Wikipedia, since Reader will know the link is to a small bit of further information. & etc.

What do you think? Too complicated or too simple or not useful enough?

I was reminded of link types today because Mark Anderson wrote:

Link types were more of a thing in 1980s hypertext as they reflected/allowed argumentation and simple computational inquiry across the network. Randy Trigg (who worked on Notecards with Halasz, at PARC) in his 1983 PhD thesis defined a complex list of >80 link types. It was around this point (source – convo with Mark Bernstein) the hypertext decided it was all getting too much and link types got kicked into the long grass.

See: Interestingly, in the last para it says “Storyspace has no link types, nor do HyperCard and its descendants, nor does the Web.“  Interestingly, Tinderbox—which was designed c2000 and first published ( c.Feb 2002 *does* have link types, though hardly anyone uses them except as visual labels on links.


Published inThoughts


  1. This seems to fall in to the chick-and-egg problem of what’s practically do-able in the immediate term vs an idealised solution. Did the web ever come up with a standardised method for sharable link-bases as using in pre-Web hypertext systems?

    As well as the technical implementation (mark-up, embed, transport, parsing, whatever) the is the human interface problem. If we are enabled to indicate argumentation/trust for citations, how is this data captured. I’m not sure folksonomic free-tagging works as well here as it does for something like photos. In this context, intent should be deliberate and unambiguous or the worth is lost, yet that fist poorly with ‘just tagging’ things.

    • Yes, I think you are right, specific use-cases need to be catered for.

  2. Maybe this should be implemented under the category of citation rather than link? Hmm… And thought of as tags rather than types.

  3. Christopher Gutteridge

    This muddles address and semantics which I’m not a fan of. The purpose is either to tell the browser what the link is and/or to tell the target webserver what the semantics of the link were, which is none of it’s business.

    HTML already has a light way to describe the semantics of a link;

    This is most often used for “nofollow” though, to indicate that the link is not an endorsement. Google et al ignore nofollow links for ranking so it’s used in wikis and blog comments to reduce the value in spammers adding links.

    The key thing is that the semantics of a hyperlink are part of a relationship between two works, whereas a URL is just the location. Many hyperlinks can exist to the same location with differing meanings.

    I think you could extend the semantics of the rel=”” relationship to include a URI indicating your own meanings.

    All of this suffers from HTML’s issue of links embedded in documents.

    • Very useful, thanks Chris. I also realised that they are not really link types, but tags about what the author feels the linked item is, such as ‘canonical’ or ‘wrong’ etc. – in addition to relationships for creating and following graphs.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.