I realized that I have not posted anything from CHI2006 and thought that putting some of the notes that I took during the conference may be useful to a number of people. Plese feel free to add to my notes as I may have missed out on something or have misrepresented what was said. Also please excuse the notetaking skills as I found it rather difficult to keep track of the conversation during panels) In general, I thought that the tagging panel was interesting and raised the fact that much more thought and work needs to go into Social Tagging. To echo what Eric said about the panels, I thought that they may have dwelt too much on the fundamentals of tagging resulting in too little time for the meaty issues of the subtle differences that resul from making minor tweaks Social Tagging systems. But all in all it was good to have heard from the people whose articles I have spent the last couple of months reading. I have followed up the session with writing email questions to both George Furnas and Rashmi Sinha. There were a number of issues that I needed to investigate further about the cognitive aspects of tagging. What I learn from them will be posted here as well ... so stay tuned.
Panelists:
- George Furnas (U. of Michigan)
- Caterina Fake (Yahoo) absent
- Luis von Ahn (CMU) absent
- Joshua Schacter (del.icio.us)
- Kevin Fox (Google)
- Scott Golder (HP Labs)
- Marc Davis, Cameron Marlow, Mor Naaman (Yahoo)
The panel sessiong began with a discussion of the question "What is a tag?"
- no agreement with regular folk
- is it too technical a term?
(Shinha)
- tags are loose associations of words and objects
- could be culture based
- some kind of association comes to mind
(Fox)
- a rich way of describing something
(Furnas)
- history of tags = library of alexanndria could marked collections with metadata
- 1980s text based interfaces = looking at what is essentially taging
- tagging as creating new handles to get at something
- how much are those new handles helping?
- but people come with an unintuitive broad variety of words
(Schacter)
- distinction between tags and metadata = additional layer of attention
- more attention
- different proposition for recall
(Davis)
- why didn't tagging take off in the early days of the internet
- tags have always been around
- but community sharing aspect is the new thing
- different from tagging with the intent of recall
- by definition flickr is not tagging = for other people and not for yourself
- there's a spectrum of behavior in the system
- who the intended receipient of the tags in flickr?
What's the difference between tags and categories?
why do tags get adopted by so many people?
- because we've put more information on the internet
- more social usage of the web
- tagging has been adopted because the cognitive cost is low
- digital categorization is inherently difficult
- simpler way of storing digital information
Discounting decision making during tagging?
- large amount of decision making
- fundamentally categorizing
What are tags used for? And do they work?
- works well for personal recall
- ok for distribution
- works better with more people
- both for personal and social purposes
* that's why you have the long tail
* ref to wisdom of crowds
* when do things in front of others = not presenting true self
* but with delicious = initial decision is made alone
motivation to tag something that other people will find = digg spam
human component to creating the tags
personal aspect is very important = self-referential aspect (newcar, jobsearch)
these tags weren't for other people
task oriented nature of tagging
there is not communication within delicious, it mostly happens outside of the system = personal and small group
Schacter originally built delicious because he had thousands of links in a fun file
can have a small minority who are bulding roads for other users on the system
tagging is a fundamental part of the UI now
allows pivot browsing'
- the way tagging relates to search needs to examined more
- tagging something with the tags in the text is very different from tagging something that doesn't have the tags in the text
furnas: what is the ultimate value of tagging?
- access subsequently
- graphic = probability of recalling
- even when you have up to 20 tags, there will still be failures of recall
- want to work to diversity of tags for recall instead of consensus tagging
- Coases's penguin paper
-- analysis of common space peer production
-- theory: when human endeavours organize themselves into markets or firms
-- large number of people need to be organized
-- various aspects of information goods allows a diversity of motivations
is tagging going to make it outside of the community of computer users?
what are some of the incentives for people to use flickr or delicious?
- done around sharing = to get people looking at your photos (flickr)
- delicious is different = social function of tagging:
-- distribution
-- does the social network affect the linguistic effort
-- if you are connected to other people your terms are more common
looking for a specific item or a genre of items
localized dialects = for leveraing small groups with a specialized vocabularies
- no present tool to map between local and more generalized vocabulary
- tools and systems that help mediate between vocabularies
furnas = can imagine tags as collaborative filtering devices
- need to regress to the norm at the beginning before you can get specific information about particular issues
perceptual differene between calling it tags, keywords or labels
sinha
tagging is self-expression and social expression
when myspace has tagging = it's going to be commoditized
tags are going to be very differently used
tagging people is very differnt from tagging objects = difference between self presentation and real self
- going to be a lot more opportunity for tension
does tagging work for certain applications only?
- sinha thinks that its an interface and design issue
- it was too busy
what are the design choices to make to facilitate participation
too egalitarian?
it wasn't personal
tagging that is done through the normal task that someone has done
does tagging scale to more than one group?
schacter thinks that this is more about community making rather than ...
things like delicious needs to put in ways for groups with in the system to occur
too many users in the system
nature of humans = doesn't scale
need to put in some mechanism for social proximity
- motivation for introducing pools or sets
- people in close social or ideological proximity are in that group
- there is a fundamental tension between the individual and the social uses of tags
- for eg. photos on flickr tagged as tokyo = difficult to comprehend outside of the individual context
flat tagging space and that of architecture
get a very rich set of social relationships
question of scale
what determines the tagging terrain such that it is no longer flat
does anyone have data on the length of tags?
can media be used to tag other objects?
No comments:
Post a Comment