SXSW Notes: Tagging 2.0

Loose notes from SXSW 2006 session: “Tagging 2.0”

This was the more nuts & bolts session on tagging, being called “THE key element of Web 2.0”

Adina Levin VP Prod, Socialtext
Prentiss Riddle Tech Evangelist,
Rashmi Sinha Uzanto
Thomas Vander Wal Principal & Sr Consultant, InfoCloud Solutions Inc
Don Turnbull Asst Professor, Univ of Texas at Austin

Vanderwall coined the term “Folksonomy”

Levin: Cognitive deconstruction, cognition on the web.


Is tagging THE key element of Web 2.0? Are we even at tagging 1.0 yet?

The word folksonomy proves its own point. We can use the word without explanation, which demonstrates its traction.

Tags can provide metadata to systems that don’t already have it.

Tags are good for RE-finding my own information. For creating personal metadata.

The new command line – a faster way to access content you already know the command/keyword for.

2.5 years of tagging on the web already, but slower to take off on the desktop.

What about using tags as verbs? “buy” “sell” “print”

What about *, **, *** for ratings? (a la iTunes/iPhoto)

Tags let you focus on the user view of the system, not the system itself.

People-centric view of data. Linking users by interest.

It’s hard to do a Google search on something when you don’t know what it’s called. Tags can help with that.

Tag Fraud: We already have tag spamming and gaming. Could there be a tag optimization business? Google paid $90 million fighting click fraud last year!

Tagging systems need ways to let you share some, but not all of the things that interest you.

Granularity – tagging individual items/elements, not just entire pages.

Tags as identity: If I look at your tag cloud, what do I learn about you?

Tagging interfaces: What’s the best way to get people to tag? How can interfaces teach us tagging vocabulary?

Tagging 3.0: Helping systems understand subtleties. Making retrieval systems smarter.

Wht if tagging systems become too powerful? No one tag anything about reading lips! [slide of HAL watching astronauts talk]

VANDERWAL: Tagsplosion 2.0

Washington Post and NPR now have links to tag their articles. So now the general public / normal people are in the tag cloud, and it needs to be explained. How can we do it without the explanation?

60-70 sites are using tagging as their main methodology.

A triad: The object that’s being tagged, the identity (could be anonymous or a screen name) of the person who is doing the tagging, and the “razor sharp” pieces of metadata. Identity and Object are connected by “interest.” Identity and Metadata are connected by “vocabulary.” Metadata and Object are connected by “definition.”

But this is expanded by the Dual Folksonomy Triad, where a triangle overlaps with another with “Community” at the other extreme from Identity. “interest” becomes “terminology” and “vocabulary” becomes


When tagging systems work, it’s because a lot of attention went into making them work (sounds obvious, but it’s a difficult problem).

“Normal” people presented with tagging fields tend to type in an English sentence. When faced with a tag cloud they say “looks hip, but too many words” (when the whole point is to expose all the words).

Tagging systems are plagued with interoperability problems, character set conflicts, etc.

Delimiter issues — delicious uses spaces, yahoo uses commaas, flicker tries to use both, and flattens the term after it’s been entered. The solution does not nec. translate into good interoperability.

Interoperation amplifies imprecision. “cat” vs. “cats” — ratios are inverted between flickr and delicious (people write about cats, but take pictures of their cat – but these terms should overlap in the tag cloud).

The delicious API will let you export your bookmarks and tags. The Flock browser will move your bookmarks back and forth between tagging systems.

Nobody wants real tags: Simple keyword metadata, no control, no hierarchy, no syntax or semantics, minimal cognitive effort by the user. What people really want is “Tagginess” [slide of Colbert]

Hierarchical tagging: (this makes Riddle’s skin crawl) — because folksonomy is not SUPPOSED to be hierarchical – who gets to decide what the hierarchy is? This is back to taxonomy.

Faceted tagging: Facets are just buckets of tags. So you dump “boston” into the “place” bucket.

LEVIN (Socialtext):

Refactoring tags for interoperability: Consolidate synonyms, fix and standardize spelling, add hierarchy (Venice, California, and Venice, Italy)

Tagging must combine human contributions with machine intelligence to reach the whole that’s greater than the sum of the parts.


The personal alternative: Categorization:
Object worth remembering –> Multiple concepts activated –> Choose one … (Tag it! solves this)

Four conditions:
Cognitive Diversity
Easy Aggregation

Tagging is “conceptual consensus”

How are you serving the individual motive?
Doe the individual understand and want to fulfill that goal?
What is the relationship between social and personal?
Is it too easy to mimic the tags of others? [I thought this was desirable!]
Is finding all about the most popular, most tagged?
Enable discovery, exploration, finding new things.
Don’t force users to do thing differently than than what comes naturally.
Solve problems by ensuring good findability.

One Reply to “SXSW Notes: Tagging 2.0”

  1. Thanks for distilling this session so well.

    I’ve been thinking a lot about tags as identity, and have been using a lot more. It’s great to be able to pass along a link and say, here’s what my attention stream has been like over the last year.

    I’m also very interested in using tags for collaborative sense making and multimedia production.

    Anyway, thanks for great write-up

Leave a Reply

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