Hi

First time here? Text your name and email address to 631.871.4864.

Your coffee's on me, just tell the barista you're with the design workshop.

Hiring? Find me at http://hirevito.com.

Looking for work? I have some leads.

Holla

WHEN I SAY "TAG CLOUDS" YOU SAY "SUCK"

Challenge

TAG CLOUDS

Response

Background

Modern tag clouds were ostensibly invented by Jim Flanagan in 2002, based on a visual design of his from 1997, for his Search Referral Zeitgeist. They were later popularized by Flickr and Del.icio.us.

For the record, Jim Flanagan "builds software systems and reads books and walks in the mountains," but is not a designer.

About "tag clouds"

"Tag cloud" as a term is a bit overloaded.

A "weighted list" is what Wikipedia says the generic term for this particular visual presentation style.

A "tag cloud" suggests tags, which suggest user-created classification taxonomies (folksonomy is a terrible portmanteau), which suggests flat organization structure, which might be considered lazy IA.

What we're here for

We're not here to discuss the pros and cons of tagging in general, nor of things like dynamic re-parenting of hierarchies around tag co-occurrence, or tags as faceted metadata, or accessibility problems with tags, or whatever. There's Refresh Austin tomorrow, or we can do that in another workshop.

We're here to examine if visually weighted lists make any sense as a presentation structure, from information and visual design perspectives.

Problems with weighted lists

Users don't understand them

"Despite choosing to use the tag cloud, some participants reported that they had found it difficult to use. Relatively few participants realised that the terms in the tag cloud were displayed in alphabetical order, and many participants reported that the use of different sized fonts in the tag cloud was disruptive." Using tag clouds to facilitate search: an evaluation

Users don't understand them

"The feedback and ratings of the study participants suggest that fun and aesthetic aspects largely affect the user.s interaction with tag clouds. Participants partly preferred layouts that did not yield the best performance." Comparison of tag cloud layouts: task-related performance and visual exploration

They're not designed properly

17: the ways we can make things stand out visually.

Information Visualization: Perception for Design, Colin Ware

They're not designed properly

3: the ways a weighted list usually does it.

Sometimes you'll also see Enclosure (boxes or background colors), Grouping (related tags near each other) or Numerosity (repeated tags, or tags followed by parenthetical numbers).

They're not designed properly

2: the good ways to visually quantify data.

Information Dashboard Design, Stephen Few

They're not designed properly

5: 2 good + 3 mediocre ways to visually quantify data:

Only intensity and size overlap with how we normally build tag clouds.

They're not used properly

Weighted lists as navigation?

They might be good for helping users start a casual browse, but studies suggest they're worse than search if you know what you're looking for, and often don't expose all your content. (The folksonomy tag cloud: when is it useful?, Tag clouds for summarizing web search results)

They're not used properly

Weighted lists to form impressions as to what this page/site/data is about?

One study suggests that the biggest impression weighted lists convey is that "collective tagging is used here." (Tag clouds: data analysis tool or social signaller?)

They're not used properly

Weighted lists as filler?

"I have this blank space and I want to look Web 2.0."

Not Web 2.0

Weighted lists were in 1995's Microserfs

microserfs4.png

Not Web 2.0

And as the cover of the 1992 German translation of A Thousand Plateaus, the second volume in the French treatise, Capitalism and Schizophrenia.

tausend_plateaus.jpg

Exercises

What data do weighted lists encode?

Imagine a tag cloud.

What do these different attributes mean? Individually, write down the possibilities. You have two minutes.

(Discuss)

What data do I want to convey?

Imagine a tag cloud used for navigation.

What information is important to convey to the user? What information are they expecting?

Individually, write down the possibilities. You have two minutes.

(Discuss)

What data do I want to convey?

Imagine a tag cloud used for impression forming.

What information is important to convey to the user? What information are they expecting?

Individually, write down the possibilities. You have two minutes.

(Discuss)

Individual descriptions

For each of the visual representations, describe at least one data point it can convey and how.

You have five minutes.

(Discuss)

Individual sketching

Sketch or describe as many weighted list constructions to fulfill the use cases of navigation and impression forming.

Don't forget: you are designing for navigation and impression forming first, so other visualization methods or information architecture organizations may be relevant.

You have five minutes.

Group iteration

Review your ideas with your partner(s). Let them critique yours and you critique theirs. Collaborate on one new tag cloud format for navigation, and one new tag cloud format for impression forming.

Don't forget: you are designing for navigation and impression forming first, so other visualization methods or information architecture organizations may be relevant.

You have ten minutes.

Presentations and Critiques

(Discuss)

Metrics

How can we test our tag cloud formats to see which ones perform the best?

What defines "the best?"

How can we measure that?

Other references

Handouts

thumb_IMG_1173.jpg thumb_IMG_1176.jpg

Originals: IMG_1173.JPG IMG_1176.JPG

DesignWorkshop/TagCloudsSuck (last edited 2010-06-07 05:12:26 by Vitorio Miliano)