Elijah Wright (presenting with Lois Scheidt and Susan Herring) on the weblog genre. There's a popular view, but it represents an unrepresentative elite. Most of the bloggers mentioned in mainstream media reports of blogging are male. "Blog Research on Genre" project (BROG), with a goal of empirically characterize the "typical blog."
[Ack! Isn't this like trying to characterize the "typical person"? Or the "typical woman"? Is there any value in an "average" representation? Why do we need to see blogging as an undifferentiated label??]
Defined blogs as "HTML document with entries in reverse chronological order." (No mention of authorial voice here...so would a software revision list be a blog, under this definition??)
Used a random sample from blo.gs for their analysis. Used web content analysis,through lens of web genre characterics. Coded features of blogs, quantified results.
Hypotheses: (1) Blog content tends to be external to the author (news, links); (2) authors are typically well-educated adult males, (3) blogs are interactive, actively soliciting comments, and (4) blogs are heavily interlinked.
- Of sample, 70% were personal journals. (Numbers gone too fast to see the range for other types.)
- 91% are 1 author, 54 male, 60 adult, 57% students.
- Gender and age of blog author varies according to blog content. (Shouldn't variables be reversed there?) Second most frequent profession mentioned was "unemployed."
- of blogs allowing comments: 43. seems to be related to the default settings in blogging software.
- 70% include external links (excluding "badges" for software developers, e.g. MT or blogger). This means 1/3 of all weblogs have no links !
Blog content is mostly personal, and often intimate. authors are roughly eqully split between male and female, adult and teen. Adult males create more filters and k-logs (in fact almost all are created by males), females and teens create more personal journals.
Conclusion: Blogs featured in contemporary public representation are not representative.
They acknowledge that sample size is small, and is English-only. However, more recent samples seem to reinforce conclusions. Present several interpretations, but I find these overly speculative. You don't know why people do things until you ask them, or at the very least do more qualitative inquiry into the phenomenon.
Jump right to predictions:
- Increasing mundane use.
- Increasing contentiousness
- Increasing commercialization
- Increasing non-blog use of blog software
I am reminded of a line I heard from Pat Cadigan (a sci-fi author) at an ALA conference, when she warned against "the danger of predicting the future in a straight line."
So, they ask, what then is "new about blogs?"
- Ease of update means more interactive webpages
- Creators can be itneractive yet maintain control
- Blurs distinction between traditional HTML documents and text-based CMC
That last one is where my interest lies. Blurring of boundaries (just search my archives for "boundaries" to see previous references to that theme). Susan Herring puts up a graphic showing a continuum of web pages to CMC.
More conclusions:
* blogs may ultimately be transformative, but not in favoring a specific content, audience, or quality
* rather they create new affordances that will be open to a variety of uses (cf email)
* important to look at "typical" blogs as well as intersting unusual ones
* look at socio-political, social-psychological, and technical implications
I asked if they had concerns about creating a "typical" profile of a diverse population--response was that they realize they need to break it down more.
Also asked if they might consider longitudinal studies--does content change over time? Go from externally focused to internally, or the reverse?
(Update: Cameron Marlow has a wiki page with his notes on this session.)
(Another update: Elijah Wright has posted the PPT presentation from this session, so you can check my #s and find the ones that I missed!)
good questions.
being researchers, they probably looked at you as though you were from mars.
i like the latter one especially. interesting
I also saw this panel. I thought I heard them say (into the non-working microphones) that they only sampled first and last posts of blogs? Did I hear them correctly?
If so, I would also want to know if they sampled first and last "days" or just "posts" ... often I will post a "news" post full of links, followed by a more personal note with no links in the same day.
I also felt that breaking blogs down into sub-genres (personal, political, technical, etc.) is somewhat counter-intuitive - my somewhat limited blogging experience leads me to believe that such strict delineations are rare except in intentionally "focused" blogs (read: corporate).
By the way, thanks for taking notes and posting them - the soeakers flipped through their slides faster than I could write with my pen ;-)
Jason, I didn't hear that. I hope that's not true, since it would make the research pretty pointless. I was under the impression that they'd done content analysis of a larger body of material.
Just downloaded the ppt from the presentation (see update to the entry, above), and it appears that the criteria for inclusion in the sample was a minimum of two entries, but that they collected over a three month period (March to May) and excluded from the results sites that didn't update for a two-week period.
I'll have to review the ppt when I have more time at the computer, but I agree - if it's true, that makes it pretty pointless... (so, I'm hoping I didn't hear correctly)
Jason - we counted some features for each blog as a whole, some for the home page, some for first and last titled entries on the home page (everything they included), and some for first (= most recent) entries only, depending on the feature. The categories we broke the blogs down into were taken (roughly) from Rebecca Blood's book; we adopted them as a working hypothesis. We coded type on the basis of the predominant content found on the home page of each blog. Where no one content type was clearly predominant, we assigned a code of 'mixed' - this only accounted for 9.5% of the blogs. Taken together, the three main types identified by Blood -- personal journal, filter and k-log -- accounted for 86% of the blogs in our sample, suggesting that the categories (especially the personal journal category, at 70%) are useful constructs.
A fuller description of the results and methodology than we had time to present at AoIR can be downloaded from http://www.blogninja.com/DDGDD04.doc. (This paper was written for a different conference, so it's not identical to the AoIR presentation, but it reports on the same content analysis.)
We agree that Liz's questions are good ones, and we plan to investigate both sub-group and temporal variation in the future.
There seems to be some misunderstanding about sampling methods. Specifically, our data was collected as a random sample. This means that any one individual's daily practices should not untowardly affect the statistical makeup of the sample as a whole - after all, we're collecting the statistics of the *corpus*, not the statistical makeup of an individual blog. This is a standard methodological tool for the social sciences and CMC research in particular.
I hope this helps clarify, at least a little. It is unfortunate that we had to rush through our slides so quickly, but that's what full papers and posted slides are meant to counteract.