THIS BLOG POST WAS NOT MACHINE GENERATED

On Helping Bloggers Overcome Writer’s Block

By Casey Dugan, Researcher, IBM Center for Social Software, Cambridge, Massachusetts, USA

At Lotusphere 2009, I heard from customers that their employees were reluctant to author blogs. Later in the year, we surveyed IBMers about this issue, and discovered that would-be bloggers often abandoned their blogs because they didn't know what their audiences wanted to read.

Because of this, Werner Geyer and I decided to create a tool to help increase blogger and reader interaction called the Blog Muse -- a tool that lets readers suggest topics for bloggers to write about.

By way of background, Werner and I are Researchers at IBM's Center for Social Software in Cambridge, MA working on projects such as SocialBlue, IBM's internal social networking tool (formerly Beehive). We've been interested in the area of recommendations, and in particular, engines that recommend content which users can create on social media sites -- such as questions they should answer on their SocialBlue profiles, or suggestions for expanding their social network.

Last year, we ran an experiment with Blog Muse that showed it was possible to increase the quality and popularity of blog posts by giving readers a greater stake in the topic-brainstorming process.

In preparing to build our prototype, we first looked into what was happening on IBM's Blog Central. We saw that some of the issues customers were mentioning could also be found inside IBM: only 3% of our employees have ever authored a personal blog entry. Further, it seemed that many of those who started blogging eventually abandoned it: 80% of personal blogs inside IBM have less than five entries. Other papers published about blogging externally and blogging inside companies show similar findings about abandonment and low participation by employees. Previous research on BlogCentral showed that while employee blogging has numerous benefits for individuals and the organization, only a small fraction of users can realize those benefits.

We then sent a survey out to 700 users of Blog Central, ranging from users who had never authored an entry to the most active bloggers in the company. We asked them questions about why they started blogging or had stopped, and whether they would be interested in getting recommendations of topics to write about. The more active bloggers were divided on receiving topic recommendations; some said they would welcome suggestions while others said they already had a surplus of topics to write about. One even went as far as saying: "This would be similar to writing paid reviews for consumer products."

However, this greatly contrasted what we heard from less active bloggers (such as those who had never written a blog entry, those who posted a single entry, or those who posted very occasionally). From these users we heard things such as:

- "In the beginning, I had no idea what to put on the blog"
- "Yes, I would absolutely use blogs more often if asked to give my feedback or to offer suggestions," and
- “If the interest number was high I would write on the subject of large scale interest. Instead of me building a following the following is prebuilt.”

Our survey helped us design Blog Muse. The topic suggestions we make to bloggers come from the "audience" -- or blog readers who are given a voice by being able to share with bloggers what they want to read about. From both our survey and previously published research, bloggers often complain that one reason they stop blogging is because of a lack of feedback from their audience. Few receive comments or ratings, there wasn't an indicator for readership; they didn't know if anyone was paying attention or finding value in what they were writing about! That is why we wanted to create a tighter link between bloggers and their audience, using Blog Muse.

The interaction cycle we envisioned, shown below, is that readers would share with bloggers topics they were interested in reading about. Our tool would then forward these requests to potential bloggers, and if any of these bloggers then wrote about the topic we would then notify the requesters. We also allowed users to vote on others' topics in order to gather large audiences together, all of whom wanted to read about the same thing, as we thought this would be an even greater incentive for bloggers to write about a given topic.



More results from our survey with blog users and their reactions to suggestions for topics to write about were published and presented at CSCW 2010 in February. The paper can be found here: http://www.research.ibm.com/social/papers/407n-geyer.pdf

So, we deployed Blog Muse in May 2009. It was built as an iWidget for the Lotus Connections homepage as well as a Firefox plugin to add Blog Muse features directly to Blog Central. A screenshot of the Blog Muse iWidget is shown below. We tracked the usage of our tool by over 1,000 IBMers and made a number of interesting findings. Our intuition that audience matters was correct: we found that users preferred topics requested by other users over those that had previously been written about & found that topics with votes were 6 times as likely to be written about than those without votes.


But, surprisingly, we did not see a significant increase in participation for writers! It seems that we couldn't get bloggers to blog more often by giving them topics to write about, which might indicate some displacement went on; in other words, bloggers wrote about topics we suggested rather than something else they were planning on writing about. Despite that, we were able to offer these bloggers something: more feedback and a tighter connection with their audience. We found that entries written through Blog Muse suggestions received twice as many comments, three times as many ratings, and more hits than others written on Blog Central during the same time period.

We published these results and they will be presented at CHI 2010 in April. You can read the paper with more of our findings here:

http://mysite.verizon.net/vzeodu7a/mysite/download/chi2010.pdf