We all grew up knowing about tags. We had tags in our clothes, we had them on our holiday presents, we played a game called tag, and some even used spray cans to tag their turf. All of these uses of tag have different meanings, but unless we understand the context and/or the person using the word tag we do not know what they mean. This can be a problem with tagging on the web, but like everything else there are two sides to the story and there are some great benefits from tagging, if it is done well.
###The World is Flat
Tagging is a non-hierarchal (for hierarchies think of the old Yahoo! directories) means of organizing information or objects on the web. Information is found through associative correlations or inference. In a hierarchy one would go to people, then film, and then actress to find Cate Blanchett. Using tagging one could use actress and moviestar to get a similar result. Prior to computing hierarchies were an efficient means of organizing information. With computing we have more options, which can work just as well.
This past year or two the phenomenon of tagging has exploded on the web. Many of the initial roads to success point right to del.icio.us, a social bookmarking tool, which permits its registered users to add tags to help them refind the bookmarks. These tags as a self-organization tool work extremely well for the person tagging. But being “social”, the sharing of the tags with others also helped others find these same bookmarks to web-based information. The use common naming of items on the web lets people surf within a community of tags used with similar definitions. This tagging for one’s self in a social manner was new on the web and provided positive results.
Other sites have been using tags to their benefit such as Flickr, CiteULike, Furl, etc. These sites let the consumers of the media provide the tagging for the purpose of refinding the information (Flickr is slightly different in that the content creator is normally the person adding tags initially). Folksonomy was coined around this idea of the consumers of the content freely tagging the information or object, usually external to the object being tagged, for their own use as well as sharing that tagging openly.
Folksonomy is for One’s Self
How is folksonomy different from tagging? Folksonomy uses tagging, but it is the “who is tagging for whom” that is important. In a folksonomy the primary focus of the tag is for one’s own use, so to ease refinding the information or object. The is tag chosen because that is the name that the person tagging calls the object. This point of reference (or self-reference) is what changed the game. People applying tags for their own use (or reuse) select words they use and believe they understand. This perspective means the tags are correct for at least one person and more often than not for the community or discipline to which they belong.
The sharing component of a folksonomy is where they become very helpful. The individuals become data points that we can use as guides. The individuals can be used as guides to find other related information that uses the word in the same manner through out their usage (or there is a higher probability of this happening). If I am looking for information on information architecture I may want to follow the links that Lou Rosenfeld has tagged and shared rather than somebody with a different definition. I use Lou as my point of reference to find information to lead me to others that use the words in the same manner.
###Tagging is for Self and Others
Tagging in general uses a flat naming structure, similar to a flat category system. Often the tags are chosen freely (not out of a predetermined list), which is also known as free-tagging. Outside of a folksonomy, these tags are often chosen based on what the person tagging believes others would be seeking or would call the information. This scenario is where things get less accurate and in worst-case scenarios it becomes what Cory Doctorow labeled “metacrap”.
Objects and information can have more than one tag. In fact, they can have many tags, as long as the tags apply. If the tag is correct (in the viewpoint of the tagger) then it should be helpful. The use of tags for finding information is not always done through one correct word, it is through various avenues of understanding that gives tagging its value. There are many correct points of reference for each object and people from various cultures, disciplines, and points of view each person brings.
###Is Tagging Worth It?
So is tagging worth the effort? Yes, it adds another layer to get to the information and aggregate information. These tools and services around tagging, including Technorati Tags, are another means of finding and sharing information. Tagging in general works well when the terms being used for tagging are used for aggregating like items. Tags can be helpful when the work used for the tag is even in the information or object being tagged, it is more helpful when used to on information and objects where the word does not appear. It is often said Google will show you what you want, but folksonomies (including general tagging) will show you what you did not know you wanted. Tagging in this sense leads to planned serendipity.
_Thomas is obsessed with information structure (among many other things) and makes a living feeding his passion of building usable and reusable information delivered to Web browsers of all kinds. Thomas has been practicing his craft for dog years and has spent the last 10 years working in and around the web as a webmaster, information architect, interaction designer, director of internet product development, and web program manager. He can be found at www.vanderwal.net and www.personalinfocloud.com._
Folksonomy tags are only useful if you are satisfied with finding _any_ result. They are less useful when you need a little more exhaustive overview.
The problem with the folksonomy philosophy is that it assumes the taggers do a better job than the writers of the site.
While in theory tags can classify a page with words that are not used on the page, to create a sense of ‘aboutness’, in reality people use the most generic words that can also be found in the title or on the page. So these pages could also be found using Google, more effectively even.
You only have to look at the right bar on del.icio.us to see how little effective these tags are.
I would argue that blogs do a better job in classifying other sites, because they use more and more varied words to describe the sites.
But what makes del.icio.us interesting is that users are grouped, so that ‘graphs of popularity’ are created. It is a continually evolving Zeitgeist, and it does a good job in spreading the topic of the hour.
Thanks for this great post on tagging. I also have become a believer that tagging is incredibly powerful. I am not yet sure if it will be that great for finding new information but it is far superior when you are trying to retrieve your existing bookmarks.
I have been bookmarking and tagging all of my sites on www.blinklist.com. Not only can you save and access links very quickly, blinklist also offers a the unique (kick ass!) feature that breaks up your feeds by having them listed in 3 channels: Most popular, recent, and favorites ala gmail stars.
Taggin is also super powerful for sharing information among relatively small groups of people since you can share common tags to aggregate links and documents of interest by blinklisting them together.
I have applied the same principal to an application I created http://www.similarthings.com It uses tags to describe and also organize pages. Not bookmarks but actual pages on the free site you create. I loved using del.icio.us to organize my bookmarks and I figured it would be a great way to organize a website too.
One way to eliminate what Cory Doctorow calls “metacrap” is a google-suggest-like dropdown for the tags. I think del.icio.us currently implemented something like that. I need to do something similar.
The sites also use tags in the rss and atom feeds as well as ping technorati so it helps feed the tag pool.
The cool part (much like del.icio.us) is once a lot of people start using it. It will become a place to find “SimilarThings” people are writing about or creating on the system.
Clay Shirky, as often, has something interesting to say (free registration required, as it’s bizarrely not on his own site).
There’s an interesting interaction of this cyberspace stuff with meatspace in the form of location based tagging from mobile devices. But, if I were a pub owner and someone rubbished my pub in a post what would happen to my trade? What if the post were unfair or a simple lie? It wouldn’t stop it showing up on people’s phones as they walked past.
The tagging phenomenon that exists for purveyors of information has many promising advantages for people. Unfortunately, by handing the power of creating personal taxonomies and classification to the end user, will we be better off?
Certainly one of the beautiful things about tagging is that it now allows users to disseminate, locate, and share information in an efficient manner. Moreover, we tend to rely on the classification of information by persons we learn to trust.
Therefore, if it does not already exist, I suggest that as the social software scene grows and grows, there should be a rating system that each of these social software services enables so that the information consumer can rate the taxonomist. For example, use Amazon as a model. They use a rating system so customers can assign a “was it helpful” ranking to layperson reviews of products and services. Then as a consumer [in this case of products and services] I can decide whether or not to trust this review based on the feedback this person has received. Why can’t we use a similar system to help consumers of information?
I liken the tagging phenomenon to a filter of sorts. Why scan the entire Internet when I can apply a filter and search for information that has already been classified for me? However, since now all laypersons using social software wield this power, could this filter become distorted? You bet. However, if we can rate this growing group of taxonomists (myself included), then we are provided with an even richer piece of metadata that will aid us in our quest for information retrieval. I propose we introduce a new term and feature set to social software - “Thumbing” - a term that describes a “thumbs up” or “thumbs down” approach to rating these social taxonomists.
Finally, perhaps when we tag, these social software sites might have a “me” and “you” demarcation so we can tag with both ourselves and others in mind.
I’ve been researching tagging my photos, with metadata, so far I’ve not completed a work flow I am happy with. I see this as something that needs worked on, partly because I realized that the filename is a tag and the sequential nature of the autonamed photos from my camera is not conducive to easy searches, ala google. So ideally I really want to tag each photo, rename it, and then submit it to flickr etc. My computer is Linux, so I’m focused on that OS, but I have read that Picasa adds some of these options to Windows. Any other suggestions?
Nice post. Some follow-up to various comments:
tags from page creator vs. tags from taggers. I believe the latter carries more weight. As a creator of a page, I can, for whatever reason, assign all kinds of incorrect tags to my pages (think SE spamming, SEM, etc.), but if the populus says that my web page is really about X and Y, then my page is most likely really about X and Y. If enough people tag my pages with X and Y, then that carries more weight. See below about recommending tags.
Yes, suggestions are a good thing, and help with what I described to Arthur above. See how Simpy does it:
What you are describing is something that, again, Simpy offers through the feature called Topics. It’s about watching links and tags of people we think know a good deal about certain topics. Topic filters let you cut through link streams of multiple people, and pull out only links that you want to see from a certain group of users.
Something didn’t like my Flickr URL, so here is another try:
Simpy screenshot of Recommended Tags at Flickr
Bob (& Rob too)–
In part I don’t think tagging is all that different that what has preceded it, word of mouth, but with the exception the pub owner now knows what is being said about their pub more quickly.
The other side of the coin is it is important to take the tagged comment in context (currently we are missing a form on context in many tagging systems — the understanding of the person tagging). With the pub scenario, on erroneous negative tagging will surely be drowned out in many other positive comments. But the better piece of information is understanding who has made the comments and posted the tags. Or social networks have to get more granular and hence more powerful before we can make sense of the negative pub tag, which may have come from the community grouch that never has anything possitive to say.
To a greater degree Rob’s thumbs up requires not only the voting to be usable for the person giving thumbs up or down, as often our collaborative voting voices are utterly worthless to many who do not have the same perspectives. It was the looking away from this perspective that folksonomies made sense to me as could filter voices to their most granular levels, the individuals, and find people whom I could surf their tagging that could get me to the information, or a helpful tag, that no other system could. The failure of other systems and collaborative binary voting is it removes the ability to get access to the long tail or the alternate voice. I found the great value in del.icio.us when in months of searching for the terminology that knowledge management used for a concept I knew what information architects use so I could find the vast years of research I knew existed that is utter void in the IA community. In this quest I found one person who tagged with the IA term and the knowledge management term.
In a folksonomy (people tagging content the find for their own reuse (I have been watching for many years the attempt at tagging for others often hitting the many, but never being useful for the whole)) we easily see the similarities and variance of individual vocabularies. What works for one person may work for others, but rarely does it work for everybody.
Look at Jon Udell’s collaborative filtering performed with del.icio.us data for a good next step forward based on granular capabilities using the person, tag, and object reference points. The possibilities are only beginning to be tapped, but we must embrace the complexity and the granularity to get to the great value.
The tagging concept has been built in to Google’s Gmail service. Gmail’s label system essentially replaces the normal hierarchical series of information silos (personal mail folders) with an unstructured list of emails each tagged with a flat, user-generated set of “labels”.
I find it a useful way to organise my emails (and attached files) and retrieval is quite good especially when combined with the google search facility. But would it make my emails, as a collection, more useful to someone else? I’m not so sure. I’d like to think I organise these things logically, but it’s all done very quickly and without much consideration.
I wouldn’t expect anyone else to make sense of the various acronyms, shortcuts and personal descriptions that crop up there.
Thomas (Vander Wal):
You bring up valid points. Much searching is done to find those gems on the “long end” of the long tail, and I certainly would not want to limit this side of the coin. I also agree with you when you say that vocabulary differs between users. However, trust is earned and can be part of an algorithm, or another option, when searching for information.
I bookmark articles of interest for my del.icio.us account and then feed some of these bookmarks to my blog. The trigger tag is “links” (or it could be “to_read”) - yet that tag has no relevance for anyone but myself. If I only used that tag and used limited descriptors for the bookmark, might others find it? On the other hand, if I include other tags to describe the content, perhaps I might attract the affinity of others’ and find folks with similar interests.
My point is that my bookmark may or may not include a variety of tags to describe some web content. It can still be part of the “long end” of the tail no matter how many tags I use and no matter how descriptive they are. However, if I used more descriptive tags and if I were rated more highly, you might find that “gem” that only a few others found, in part because you trusted me - and this trust might translate into a algorithmic filter of sorts.
OK/Cancel is a comic strip collaboration co-written and co-illustrated by Kevin Cheng and Tom Chi. Our subject matter focuses on interfaces, good and bad and the people behind the industry of building interfaces - usability specialists, interaction designers, human-computer interaction (HCI) experts, industrial designers, etc. (Who Links Here) ?