Jon Lebkowsky
Whuffie in Links

The Emergent Democracy tribe's been discussing a possible enhancement of href links. Since Google's page ranking uses the number of times a page is linked as part of its algorithm, it might make sense to include other information about your evaluation of a page when you link to it. The idea is to contribute your assessment of the whuffie (reputation) of the link, so that you wouldn't assign more credibility to a bogus page if you linked to it for some reason. Broad implementation of a method like this could improve Google's assessment of value, and it might have other uses as well.

There's a debate about the best way to implement something like this. My opinion is that you would add an attribute called "whuffie," after Cory Doctorow's term for reputation in the imagined future of Down and Out in the Magic Kingdom. Whuffie could have a value of -10 to 10, so you might have <a href="http://www.weblogsky.com" whuffie="10">.

This is a great opportunity to identify the blogroll from the entry. Various sections of your blogroll indicate a static vote of confidence about the blogs they point to, whereas entry links are more about the articles they point to. Blogstreet should look at only the blogroll, Blogdex, only the entry links and Technorati a combination of both. I guess the URL tells you, in a sense, if you are pointing at the entry or the blog, but you could make it explicit in the tag.

This meme reached mailing list escape velocity in only one day! Almost missed it. ;-)

19 Comments

Eeek! Before we go adding non supported "whuffie" attributes to HTML (oh please don't!!!), let's look at the possibilities avaliable to us!

At SxSW I chimed in at the "Future of Blogging" panel about how "FOAF" can be used exactly in this way. Please give me a few days to create an MT based template and demo!

In a nutshell: linking to "friends" and articles via FOAF (or another similar standard) could not only be an indicator of a positive indicator "mark" or "flag", but because it uses RDF (as in the same stuff RSS 1.0 is made of), it can be aggregated and harvested ...)

Hey Boris. I knew someone would react quickly. 19 minutes... not bad. ;-)

Don't worry, weren't not implementing yet. (Maybe this sort of thing is a great way to get people to make tools more quickly...)

Agree, that FOAF makes a lot of sense. This discussion started in the context of a voting tool for blogs.

Also, noted that many people have talked about this sort of thing for a long time, but probably more people hardcoding href tags than any time in history. (I thought blogs would make me free of html, but in fact I can now write html by hand better than any time in my life!)

Ah! Hehehe!

As things move forward, only those who wish to hand code will do so. (hello XHTML 2.0!).

I strongly urge those of you who are not "in the trenches" of web technology keep an open dialog with folks like Karl (and myself, and others... and others) because there is so much truly amazing stuff going on down here in deeper reaches of web-tech and the ideas are bubbling and the tools are coming. I promise!

As for a voting tool, well.. just visit my weblog! I have an early alpha version of a voting system concept running there already! I have many issues, both conceptual and technical, to iron out but it's a start... Ask me about it sometime! ;)

Sent this to the edemo list:

After that comment Joi cross-posted about FOAF ("Friend of a Friend"), I
went back and looked at the spec... not sure it fits our intentions. I
haven't used FOAF so far, and I'm not sure how it's implemented, but I
certainly don't think an expanded RDF format gives us what we want, which is
a way to assign values to links.

As for "adding non supported 'whuffie' attributes," I don't think this was
our intention (supported attributes begin as proposals).

A modified or expanded FOAF format, or one derived from it, could contain your entire blogroll for syndication. Because it would use RDF, you could tak on any kind of metadata you desired, be it trust indicators, votes, comments, etc.

FOAF is at work and is not finished. It's a spec where all people are invited to participate.

The more people will participate to this work, the more we will have a chance to make it useful.

For the link aspect, it makes me think, I wanted to make an entry on profile and rel in HTML, I think it's time to explain.

You can use standard mechanisms in HTML 4.01 to add properties to links.

Do you want, I explain a bit here or can you wait one day or two?

Karl, it would be great if you could explain... The mailing list is going at full speed and your thoughts would be very timely.

I think that FOAF could have some value on the blogroll side of things, but I don't think it's appropriate for the embedded links.

If I link to an article on CNN's site, or a posting on your site, I think the link itself should be able to carry an "I support this" or "I do not support this" attribute. It's not a comment on the author, necessarily (although a series of "I do not support this links" would certainly send a message). It's a comment on why you're linking to the content.

My suggestion on the mailing list (which I'll blog tomorrow when/if I have a little more time) was that this be a modified version of the a tag, either through a variant on protocol, or through a new name. The "new" link tag would have two required attributes--the href itself, and a value/quality/vote type tag, with a constrained -1/0/+1 value. (I should note that not all of that is my idea...it builds on the various components others have suggested.)

I think Andy Edmonds has been thinking about this.

it's quite simple in fact and straightforward. It just need a bit of consensus on the way to do it and maybe look extensively on the Web if someone has not categorized the information.

I have written something to explain it, it needs work but it explains the concepts: trust Meta Information

Sure, FOAF sounds like it works for site2site links, but, as Liz points out, there is a need over and above this for page2page links.

The whuffie concept seems like a potential solution for non-endorsing links, although I'd prefer a simple "affect" attribute. I also think it's better as a -2 to +2 scale.

There's a huge amount of accumulated wisdom on this topic from hypertext research.

Would someone post some pointers into the Emergent Democracy list that give a quick intro to their thinking on this topic.

PS -- Seb, Thanks for connecting the dots!

Thanks for that hypertext research link. It occurred to me that those of us who were discussing this on the Edemo list should look around at what others are doing. As Karl has noted, we need to follow a standard (W3C) process for integration with the xhtml spec. Also Pete Kaminski et al are discussing the creation of a standards group for "social software."

I'm not clear on how a 'whuffie' system would apply to links. Whuffie would be tied to individuals, not specific posts.

As I understand it, whuffie woudl enable you to see if the site/person you are looking at is held in high esteem by others.

Even better is the ability to know if the people who have given whuffie are people who think like you (give whuffie to the smae kinda people).

This allows the left-hand and right-hand whuffie.

For example both Andrew Sullivan and Noam Chomsky would have high whuffies, but from drastically different kinds of people.

Is this what others are thinking or is the current thinking all link based?

I think I am with you (on this one)Michael. Unless there is something I am missing about the whole "link" thing.

This post is a defence of the term 'vote' and a simple yes/no/abstain formulation.

The pushback against 'vote' seems to be first that it implies a degree of formal organisation that is not present in one person authoring a link, and second that a hard-edged choice is insufficiently nuanced

Perhaps this is a transatlantic linguistic distinction, but to my English way of thinking, a vote is an individual act - marking a ballot. 'One man, one vote'.

In existing institutions, a vote on its own is not much use - a formal process needs to surround it - an election, poll, motion or referendum.

The point of this proposal is to enable such processes to be built from the bottom up - in an emergent way. Already, our links are counted as votes by existing statistical engines, but only their presence can be counted, not their absence. This proposal is meant to provide a small building block of individual decision, enabling a web author to represent clearly agreement, disagreement or neutrality with the page linked to.

Expressing finer graduations of agreement is unnecessary. The historic purpose of calling a vote has been to make a choice. If the motion in question is unsatisfactory, you can always propose a clearer formulation, link neutrally, and start a new poll.

By combining the vote attribute with an id attribute, Ross's polling site could track votes and discussions of a topic, and maybe even identify those contributions that swung most votes, as I discussed earlier.

These motions and referenda could be formally initiated, or could be detected by spiders monitoring 'vote' links, and flagged and formalised when the number of votes exceeds a threshold (the magic 150?)

On the 'whuffie' front, this is not meant to provide that; it is for ideas not people. However, the vote plus id model could be used to track the progress of an idea, and could be used to identify the most influential posts in a discussion - those that triggered the most downstream votes.

Thank you for your patience, after pretty much not seeing the use of "votelinks", your points finnally clicked with me. I now see how Google pagerank favors links that are networked regardless of actual favor from visitors. This could changed.

However, I do not like the idea of more codebloat. What say you?

You don't need formally add a new attribute to href links. You can just add an attribute. The attribute should be ignored by browsers. Those "in the know" will be able to do something useful with it. Those who don't care won't be affected.

Another neat idea, for some html like:
Purple Cow
Beer!

you can make it easier for people to see what you mean if you add a little bit to your style sheets, like so:
a[vote] { font-size:1.5em; }
a[vote='+'] { font-weight: bold; }
a[vote='-'] { text-decoration : line-through; }

(It works in Mozills 1.3, and shouldn't affect any one else)


One other thought, inspired by Michael (above), it seems to me that you can vote for/against both a site or ideas, depending on how specific you link is. I would interpret a link like: weblog as an endorsement for the site (or whoever's responsible for it), while a link like:
Patriot Act would be intrepreted as disagreement about a specific thing/idea. It's not fool proof, but maybe it's good enough?

Your link votes don't work on my browser. EI 6 (mac) But I will scope them on a PC next timed I am in the office.

Leave a comment

9 TrackBacks

Listed below are links to blogs that reference this entry: adding more information to links.

TrackBack URL for this entry: http://joi.ito.com/MT-4.35-en/mt-tb.cgi/585

Anti-links, vote-links, value-links from Stochastic Aleatory Ontological Expostulations
March 15, 2003 1:15 AM

...Joi Ito has picked up the subject; read the comments on his site, very interesting. Some people there are advocating FOAF or RDF solutions, which I think are way too complex for simple page-to-page links... Read More

Blog popularity indexes like Daypop and Blogdex showed that The Raging Cow story was one of the top-linked items last Read More

FOAF as Whuffie encoding from anil dash's daily links
March 16, 2003 4:50 PM

http://joi.ito.com/archives/2003/03/14/adding_more_information_to_links.html Read More

There has been some discussion (Jon Lebkowsky, Joi Ito, Kevin Marks and Greg Elin) on the possible addition of an opinion attribute to the link tag. This would allow people to show their approval or disapproval for a link at... Read More

Phil and I are going to disrupt the Power Laws of Weblogging. Yeah, just the two of us. Through, uh, mutual TrackBacking and, uhm, catchy jingles. Seriously though, I don't think Technorati's Interesting Newcomers system is going to help expose... Read More

Phil and I are going to disrupt the Power Laws of Weblogging. Yeah, just the two of us. Through, uh, mutual TrackBacking and, uhm, catchy jingles. Seriously though, I don't think Technorati's Interesting Newcomers system is going to help expose... Read More

Phil and I are going to disrupt the Power Laws of Weblogging. Yeah, just the two of us. Through, uh, mutual TrackBacking and, uhm, catchy jingles [update: now with actual catchy jingle!]. Seriously though, I don't think Technorati's Interesting Newcome... Read More

One Note from AKMA’s Random Thoughts
November 13, 2004 9:33 PM

I acknowledge that by posting this note, I repeat and perpetuate my fixation on one as-yet-unimplemented feature of the major search engines. Nonetheless, this morning I encountered another setting in which I would love to be able to run a seeded searc... Read More

Might it soon be possible to Aggregate my Blogroll/OPML/FOAF file, and aggregate their Blogroll’s, Links and entries. Add weight with... a whuffie="10″ attribute, and use the rel="tag” to discover how a link or blog entry has been ... Read More

About this Archive

This page is an archive of recent entries in the Business and the Economy category.

Books is the previous category.

Computer and Network Risks is the next category.

Find recent content on the main index.

Monthly Archives