The idea that HyperTexts should attach types to links.
- I generally argue for TypingNotCategorizing as an InformationArchitecture strategy
2020 : revisiting this theme in a DigitalGardeners discussion with BillSeitz.
He wants rel="tag" attributes for in-wiki links (http://webseitz.fluxent.com/wiki/2005-07-26-MarksDecentralizedTagsonomy)
I'm not sure that merely linking from PageOne to PageTwo implies that PageOne should be tagged as part of the PageTwo category.
OK. It looks like the value is the word "tag".
But I'm not sure I understand. What is "tag" supposed mean in this context?
That if PageOne links to PageTwo, then that's equivalent of tagging PageOne with a #PageTwo tag or classification?
That seems to conflate two different ideas.
If I have a page that's a local index called "PagesThatNeedWork" and it links to a page called "ChocolateCake", that doesn't mean that "PagesThatNeedWork" is part of the set of pages that discuss ChocolateCake.
True, for a listing-page it's kinda stupid.
But if you think of every WikiWordAsTag, and PageA links to PageB, it kinda makes more sense.
How I see its use... If PageA is a longish unique page name, and PageB is a more general term being referenced, then more spaces could have a PageB, and you could follow a PageB tag, and browse lists of fresh pages linking to PageB.
Well that works if, say, the BillSeitz page has a CategoryPerson link, then yes, clearly BillSeitz is in CategoryPerson.
On the other hand, if, to take a real example ... today I just cleaned up a paragraph from the ThoughtStorms BillSeitz page where I was inviting you to try SdiDesk. (So that's only 15 years out of date :-)
There's no particular reason to infer from that link that you are one of the pages "about" the topic of SdiDesk.
The bigger picture is that the ideal is that wikis would have different TYPES for links between pages, that mean different things. Which is why, initially, I assumed your rel="tag" was meant to be more like rel="broader-category" or rel="counter-argument" or rel="admin" etc.
Obviously having types is useful. Putting the types in, explicitly, by hand, is too much work and prone to error.
So the big question is how much "type inference" is possible / valuable?
I don't have explicit tagging in my new wiki engine yet. But I plan to very soon. And obviously ThoughtStorms is full of CategoryCategory type classification. But I think I really want to distinguish between explicit tags (or Categories) vs. "merely links to"
But really, if we could work out some rules for easy inference, that would be very useful
- RandallTrigg's thesis :
There's a blog : http://is2.xspaces.org/TypedLinks
There are typed links on ThoughtStorms. For example (WarpLink) (and I suppose (IoB) )
TechnoRati just introduced them for the internet : VoteLinks
PeterKaminsky considers FolkSonomies and wonders when we'll see "tagged links" : http://peterkaminski.com/archives/000425.html
Turns out that HTML actually has them : https://developer.mozilla.org/en-US/docs/Web/HTML/Link_types
DavidWeinberger suggests social links are typed : http://www.corante.com/many/archives/2004/08/24/typed_links.php#comments
Which clearly connects to the ClayShirky point that structure is important, along with size, for social groups. (ScaleAndStructureOfSocialGroups / TypesOfComplexSystem)
2010, see DareObasanjo's comments on the SocialGraph
See also :