Home | Blog | Brainjam | Publications | People

Why We Tag?

When ZoneTag was released early last year, we made it easier for users to tag Flickr photos directly from their phone. An incidental test, for example, quickly showed that on average, a public ZoneTag photo has twice as many human-entered tags (2.2 tags when we checked) as a public photo from another uploader running on similar phones (0.9 tags).

It was also immediately clear that, even with ZoneTag, many users do not tag their photos before upload, and some users never tag their photos at all. In fact, even in Flickr, most users do not tag their photos. Presumably, tagging is something to be encouraged because it makes the system more usable for each individual user, creates more information for the community, and so forth. Let’s imagine for now that tagging is indeed a behavior to be encouraged.

As a first step, we set out to understand the tagging behavior and motivations of people that do tag in ZoneTag (and Flickr). In a retrospective study led by Morgan Ames, one of our “rock star” interns, we conducted a series of interviews designed to extract this information. The study will be presented as a full paper in the upcoming CHI2007 conference.

The primary interview method we employed in our study was “photo elicitation”. This method is borrowed from anthropology — the idea of using photos (sometimes taken by the interviewee) to trigger and frame the discussion. Professor Nancy van House at UC Berkeley’s I-School has adapted this method for use in cameraphone studies: use the photos the interviewees have taken to trigger discussion about the circumstances of the photo capture and other information around the photos (in our case, the photo’s tags). The cameraphone study of Tim Kindberg, Mirjana Spasojevic, Rowanne Fleck and Abigail Sellen also had a foundational impact on our work - read it! We were lucky to have Mirjana comment and suggest improvements for our paper before submission.

To summarize the main findings in our work, we believe that Flickr is a vibrant, successful tagging platform because tags on Flickr have multiple, pervasive benefits. The interviews surfaced a taxonomy of motivations for tagging, which every single participant in our study had multiple motivations from the taxonomy come into play for every photo they have taken. And I will explain.

Here is a figure that illustrates the taxonomy developed by Morgan:

The taxonomy is broken into two dimensions, functionality and target/sociality.

• The Organization/Self square (upper left) is the classic reason for labeling or tagging any type of content: being able to find the content later. This is the primary reason why people (should) label their photos on Picasa or iPhoto, but rarely do.
• The next square, Communication/Self on the upper right, is akin to the reason you write notes on the margins of your printed photo albums: to help yourself recall the context of capture, or the non-obvious details about the content of the image like the names of the people in it.

The bottom (social) part of the taxonomy figure is where Flickr makes a difference.

• The Organization/Social square (bottom left) reflects the motivation of people to tag their photos so that their photos will be discovered by others. When discovered by others, the photo gets more views, maybe even “favorited”, and the photographer gets some good ol’ vibe. Now, why would people need that vibe? This question edges into social psychology and is out of our scope for today…
• The Communication/Social square (bottom right) identifies the motivation to tag in order to communicate information about the photo or its context to other people. For example, a benefit of adding the “Red Elvises” tag, to a picture taken at their concert, is that my friends would know what the picture is about, and that I attended said concert.

An interesting point is that our interviewees often discussed aspects related to Communication/Social in terms of known others (family or friends), while aspects related to Organization/Social often referred to the general public. In other words, they didn’t care much about explaining their photos to strangers, and on the other hand, also didn’t care much about helping their friends find particular photos…

You can refer to the paper (on our our publication page) for more details about the methodology; reflections on the effect of suggested tags; and implications of our work for designers of tagging systems everywhere (briefly: make sure your tags “go a long way” and serve a number of purposes). If reading is not your “thing”, come to Y!RB’s Brain Jam on April 13th to see Morgan presenting this work (our other CHI paper on privacy will also be presented), or just catch the presentation on Wednesday, May 2nd at CHI 2007.