The Setting

The course was BIT320: Databases and Information taught at the University of Michigan's Ross School of Business in Fall, 2004 to juniors and seniors in the undergraduate program. BIT320 introduced relatively novice students to n-tier architectures, Oracle, Linux, XML, XSLT (a language for transforming XML), and XML syndication formats such as RSS. The majority of the grade for the course was based on two projects using these technologies that students themselves had to structure.

Suffice it to say, the course in its iterations has both proven challenging to students and, remarkably, increasingly popular, now ranking very well against other undergraduate electives. The goal of using blogs in the class was to get a view on what students were learning and where they were encountering problems in a relatively unstructured and technically challenging learning task.

Overview of the Blogging Environment

Students used Typepad for their individual blogs. We aggregated students blogs at an aggregation site. The aggregation site published a view of our blogosphere in both XHTML and various XML syndication formats including RSS, atom, and OPML.

I chose Typepad for the student blogs because it provided them with a standards compliant platform that had excellent user interface support. Further weighing in Typepad's favor was that, at the time (and I think still), it was the only major blogging service that provided support for all of the major XML syndication formats: RSS 1.0, RSS 2.0, and atom. This gave us necessary flexibility because, right from the start, we sometimes needed to switch syndication formats due to arcane technical incompatibilities. Graciously, Typepad agreed to let students blog for one semester for free.

Andy Seidl of MyST Technology Partners put in the basic architecture of the aggregation site using their very high quality back-end semantic processing systems. With a little bit of elbow grease, it is now possible to replicate virtually everything we did using ReBlog, Movable Type, and a few other plugins for very low out-of-pocket software cost (but a lot more hand tooling than required in MyST). I have Movable Type templates and modifications to the basic ReBlog package that I am willing to share.

The general design of pulling posts from many blog sites into one aggregation site is now being repeated in many arenas for search engine visibility (e.g., PR and CEO blogging). The key difference between these efforts and our site is that we were focusing on creating a learning community of individuals that interacted with each other, not on increasing the web visibility for our participants. This latter did happen as a side effect.

Key milestones in the semester

As I described here, Andy Seidl and I initially put a lot of thought into how to enable community interaction. More important were the lessons we learned as the semester progressed. It was somewhat like assembling the airplane as we flew it. I describe these lessons as milestones, but they were not necessarily sequential. I list them here in order of urgency as we faced them.

Milestone 1 — Getting the conversation going

Current wisdom in corporate blogging is to listen to the ongoing conversation before diving in, and I think that is good advice. The problem we faced however was that there was no conversation to begin with. We had to get it going.

Now, some might wonder that as the person teaching the course, I could have just mandated participation, and indeed I did mandate twice weekly participation. I also provided broad guidelines that encouraged students to post on the general topic of information business, questions they had in class, or responses to other students. But, mandates and guidelines only work if people have a specific idea of how to proceed (the “where do I start exactly?” question), otherwise they just avoid the issue. So the guest bloggers I mentioned at the start of this post were one very specific how-to. I also provided examples of posts with my own “student” blog (blogonaut).

Most importantly however, I developed a whole feedback system for students, so that they knew their posts mattered and were hitting the mark. To do this, I highlighted student posts on blogonaut, and responded as appropriate. Also, since the class was comprised of non-technical people who might not be as branched into the Internet as the average blogger, I highlighted salient student posts when we met in class and talked about how what they were doing was good and/or responded to issues they raised in front of everybody. In that way, students saw that they and their peers were capable of making a contribution just as the syndicated guests.

I dealt with things I was less happy with by not even mentioning them for the most part or quietly having a conversation with the particular parties if things seemed to be getting out of hand. Only once, after the fairly sharp sting of the 2004 election results took a while to wear off for many in the class, did I feel the need to mention generally that we were getting off topic. I focused on the positive because many people are naturally inhibited in writing, and most students worry too much about what the professor will think about them. I wanted them to surge forward without getting hung up worrying about how they looked.

Milestone 2 — Tracking the class blogosphere

Now that students were saying things, how could they become aware of what was going on in the class blogosphere? Many might think the answer is obvious in early 2005: just use an RSS feedreader such as Newsgator or web-based aggregators such as bloglines, MyYahoo, or MSN. Have everyone subscribe to everyone else's feeds. However, in August 2004, there were impediments to each of these, particularly given the budget of under $100 that we had for the whole project (not including labor which people donated). MyYahoo had just started a beta service for RSS, and it was buggy. MSN did not do RSS. Bloglines was hard to figure out for the neophyte (based on my own attempts to use it, chock that up to operator issues if you will). Newsgator required purchase and installation in the school's corporate client image. Not only was that hard organizationally, but it ignored the fact that students use many computers, not just one with all of their own personal settings. Newsgator has recognized this issue and is working on a server-based product.

Solution 1 lasts about 1 week

Our initial approach was just to have students go to a web page where all of the blogs were listed. They could then click through all of the blogs and see what was happening. This solution was simply not efficient, and lasted about one week.

Solution 2 lasts a little longer

Desperate, our next approach was to have each student sign on to MyYahoo and get a beta RSS-enabled front page. We tried to create an OPML file (an XML format that, among other things, makes it easy to subscribe to multiple RSS feeds easily) that students could just import into MyYahoo, but we found that we could not get MyYahoo to work with the file consistently (the file worked with a number of other readers). So, we tried to have students sign up for each RSS feed individually and provided a button for doing so for each student blog. There were 31 students in the class plus over 10 other feeds students needed to access, thus this approach also wound up being too inefficient to be practical.

Solution 3 is a “River of News” feedreader

It took us another two weeks to get to the point where we had a usable aggregation view. The answer was simple. We just created a page that was like a river of news feed aggregator (thank you Dave Winer) that just sorted all student posts in reverse chronological order as they came in. The aggregator held four days of information. Through the end of the class, about 80% of students used this option for keeping up with the class.

Solution 4 is an open source RSS feedreader

During our travails, Mark Shuster, a student, came up with the suggestion that we try the Sage RSS reader that operates in Firefox. About 20% of the class wound up using this and loving it. It is the solution I now recommend as a first option. The combination Sage + Firefox was free and sage could be installed into Firefox by typical users in the school system under Windows. Sage worked for us because I had already managed to fight the organizational battle to get Firefox installed, and students could freely install Sage and Firefox at home. This combination still had the Newsgator-like limitations that I noted above, but I could tell students to just download and install without having to worry about the licensing.

Milestone 3 — Tracking Conversations

As Fredrik Wacka recently pointed out, tracking conversations in the blogosphere is an issue everyone is facing. The simple question raised by students was, “How do I know someone has responded to me?” We wanted to avoid comments on individual sites because that would take us back to everyone having to follow individual blogs to track conversations, already rejected as inefficient. Students could follow the river of news aggregator for a response, but they could miss the response because they could not know a post related to their previous post without reading it.

As it turned out, one of the most useful features we found for the real-time maintenance of conversations was Typepad's support of the trackback feature along with email notification for when students received a trackback. Nonetheless, actually getting trackback to work was difficult, and I doubted its generality. Again, a student saved the day. Elizabeth Balten figured out how to get trackbacks to work between two blogs in a way that her peers understood. Further innovation led to the use of the Typepad bookmarklet with the trackback box enabled and enabling email updates to student bloggers when their posts received a trackback. This worked for us.

That said, trackbacks are far from perfect. They probably work best in a small group where everyone knows each other and will not be offended by errant or slightly off-topic trackbacks. Further, there is the problem that when individual posts trackback to each other, each post must track back to every other post in the conversation. It starts to become like an email mailing list with everyone receiving a copy of the new trackback. More importantly, when the subject you are talking about consists of multiple posts by multiple people, what is the subject exactly? Can you say it in a word? These questions become particularly relevant as you start to scale beyond our class size.

Shelley Powers' proposed tagback system may help. I am currently trying to come up with systems where people can create new subjects from the bottom up, publicize new subjects, and associate resources with them.

Milestone 4 — Why blog instead of using technology X?

This question came mainly from academics who had invested in some previous computer supported cooperative work (CSCW) system. Nonetheless, it also came from students and is a reasonable question. Should we adopt new technologies because they are new? As I hope this tale illustrates, adopting new technologies is costly.

My answer is this. By design, blogging allows individuals to raise topics of interest and create threads of conversation without having to ask anyone's permission. That was an explicit design consideration for this course; I wanted to know what was going on with students. Bulletin boards tend to be top-down and are owned by one person. Wikis force you to go through a social filter. Others can edit your pages or even delete them.

Second, because blogging also produces XML-based feeds, it is very easy to aggregate all of the individual contributions in one place while still maintaining individual attribution.

Third, the XML-based feeds in blogs allow me to join people and resources to my group vs. having to get them to join me. Note, I did ask permission of everyone whose feed I aggregated into our site, but they did not have to go through a sign-on process and explicitly produce content for the site. By localizing content creation, blogs make it possible to ask permission and get a coherent stream of content.

What's Next on the Docket

You might be wondering at this stage what impact this all had on the class' performance and how students perceived it? My next post will excerpt a survey of the class that asked this question along with other measures of impact. The survey was conducted under my supervision by Jude Yew, a recently admitted doctoral student.

the next post is now up and relates quantitative analysis bearing on the patterns of interaction in the learning blogosphere.

[The tagback for this post is LearningBlogosphere. To tagback, just copy the link into your post. I'll eventually pick it up in Technorati. Feel free to trackback also.]