Making it Work Perspective
Changes in Liblogs: Slouching Toward a Study
How are liblogs doing?
That’s not a simple question and there may not be a plausible answer. That’s never stopped me before. This Perspective is a note on what may be a work in progress. A few attendees at the Texas Library Association Conference heard snakk portions of it as part of my session on balanced libraries.
First, let’s define liblogs. I don’t care for the term “biblioblogosphere”—it implies too much and too little. That term might apply to a well-defined community of blogs about books themselves, one in which most participants are at least vaguely acquainted. Applied to libraries, it would necessarily include blogs from libraries and blogs about libraries.
Liblogs isn’t an elegant term. When I use it, I mean what Steven Cohen calls “libr* blogs”—blogs by “library people” (as opposed to official library blogs), not limited to blogs by MLS-holding librarians. It’s the term I used for the biggest study of the field: Perspective: Looking at Liblogs: The Great Middle (C&I 6:10, August 2006). That study considered 213 blogs drawn from “a population of around 550 active liblogs represented in the directories and wikis.” I tried to use objective, quantifiable methods for cutting the population down to size and deliberately left my personal feelings about specific blogs out of it—and avoided making personal comments on the blogs themselves. That’s partly because of certain reactions to the 2005 article Perspective: Investigating the Biblioblogosphere (C&I 5:10, September 2005), which looked at 60 liblogs drawn from a pool of 238.
Second, what does “how are they doing” mean? Possible expansions of that question:
Ø Is the liblog universe growing rapidly, stabilizing or contracting?
Ø Are libloggers posting more or less frequently?
Ø Are people commenting on posts more or less often?
Ø Are people writing longer or shorter posts?
Ø Are readers more or less interested?
Ø Are bloggers losing interest or is the field flourishing?
Ø Are liblogs becoming more or less significant as sources of ideas and discussion on library-related issues?
Ø Do liblogs have more or less impact than they used to?
These are longitudinal questions—they assume a baseline against which change over time can be measured or judged.
I believe the two earlier studies provide a baseline for a sizable portion of the liblog universe. If this project continues, I plan to establish a baseline for a much larger portion of the liblog universe.
That last question? Unknowable or at least not provable. Technorati’s “authority” number is a slight indicator of visibility among other blogs, but that’s about it. As for Google or Live Search or Yahoo! result counts, unless they’re small, those numbers are so problematic it’s hard to say much about them. Longtime readers know my opinion on this issue, but it’s a belief, not knowledge. The same is true for the penultimate question: It’s a matter of opinion.
For that matter, “are readers more or less interested?” is probably unanswerable on a general basis or, from the outside, for an individual blog. Depending on how your blog is set up, you may or may not have a reasonably good idea of your own readership numbers, but all anyone else can do is look at Bloglines and Google Reader subscription numbers, at best a vague approximation of actual readership (and at worst simply wrong because of multiple-feed issues and abandoned aggregator accounts).
As for the other questions: “Losing interest” is tough, but I believe measures can provide clues for all of them. Are they worthwhile questions? It’s fair to say that question is one reason I’m not sure whether I’ll continue slouching toward the big study.
Before I started doing real-world measures, I had guesses based mostly on reading way too many liblogs and paying attention to conversational currents. Here’s what I guessed in early 2008:
Ø The shine has worn off blogging. It’s moved from being a shiny new thing everybody should be playing with to being an easy publishing medium that’s just a tool. That might mean the universe is stabilizing or starting to contract, if you define the universe to exclude mandatory (and short-lived) liblogs created for library school courses and other web-tools learning experiences. It might mean that expansion is slower than it used to be. It might also suggest that many blogs would have less frequent posts than they had in former years, as some bloggers become less interested in blogging.
Ø The shine may have worn off reading blogs and commenting. There are too many things competing for people’s attention and time allocation is the one true zero-sum game. If that’s true, I’d expect to see slower rises in readership for blogs. On the other hand, good aggregators make it easier to follow multiple blogs, which might counteract this trend.
Ø Libloggers use blogs for serious, substantive discussions. It’s an easy, effective publishing medium: That hasn’t changed.
Put those together, and the answers to the questions in the first list might be:
Ø The universe is likely to be growing more slowly than before.
Ø I’d expect libloggers to be posting less frequently. I think some things that used to show up as short posts now wind up on Twitter.
Ø If libloggers are posting less frequently but with more substantive posts, there might be more comments per post, even with stable or slowly growing overall readership. That’s my naïve guess as to what’s happening—across the board, that is.
Ø It seems to me that libloggers are writing fewer but somewhat longer posts.
Ø While some bloggers have lost interest and others may be losing interest, the field as a whole is healthy, if perhaps not flourishing.
Ø As to the last two questions: I gave my opinion in August 2007 (Perspective: On the Literature), beginning with this forthright statement: “I believe that gray literature—blogs, this ejournal, a few similar publications and some lists—represents the most compelling and worthwhile literature in the library field today.” I believe liblogs are becoming more significant—and within the library field as a whole, their impact is growing.
The starting point: 60 blogs with metrics for April-June 2005 and 213 blogs (including some of the 60) with metrics for March-May 2006. That doesn’t add up to 273 blogs. Quite a few blogs disappeared by early 2008. (The trade paperback version of Cites & Insights 6 includes details on disappearances and changes as of December 2007.) My starting point for 2008 was 227 surviving blogs with earlier metrics; even that number is clearly too optimistic.
I wanted to expand that universe to include a reasonably comprehensive set of English-language blogs (keeping the handful of non-English 2006 blogs), with the following criteria:
Ø Not clearly defined as an official library blog.
Ø Somehow vaguely related to library people.
Ø Reachable on the open web.
Ø Established: At least one post before January 1, 2008.
Ø Not defunct: At least one post after August 31, 2007 (as of March 1, 2008).
Ø Visible: Sum of Bloglines subscriptions and Technorati “Authority” at least 9 when tested in first two weeks of March 2008. (That’s a very low bar—six subscribers and links from three other blogs are enough, for example.)
If I do the actual study, I’ll add one more criterion:
Ø Semi-active: At least one post in two of the three months March, April, May 2008.
PubSub is defunct and the Open Directory seems stagnant, but there are new ways to identify liblogs in addition to LISWiki, still the primary source for “libr*” blogs:
Ø Blogs included in the “Favorite blogs” survey done by Meredith Farkas as reported in Information wants to be free.
Ø Blogs in the LISZen source list.
Ø Blogs in the source list for Dave Pattern’s “tag cloud” of liblog posts.
Ø Blogs I discovered on my own.
According to records I kept during the process, I added 48 new blogs from the “Favorite blogs” list, 81 more blogs from LISZen, 37 from LISWiki, nine from the tag cloud and 29 from my own Bloglines list. That’s clearly wrong, since I added 315 liblogs and the sum of those numbers is 204. Where did I find the other 111? That’s a good question.
I can say that, of some 450 new candidates from those sources, roughly one-third were defunct and another 90-odd are “invisible”—they have so few subscribers or links that they appear intended strictly for a small circle of friends. I believe many of the defunct blogs are mandated blogs, where the student has deleted the blog (or stopped posting) at the end of a course or learning session—but hasn’t gone back and removed it from LISWiki.
In 2006, I thought there were about 550 visible liblogs (including a handful of non-English liblogs appearing in LISWiki). That appeared to be more than twice the number in 2005—a remarkable growth rate. In 2008, the universe of visible English-language liblogs (plus the same handful) appears larger, but only slightly: Maybe 650 to 800 with at least a hint of continuing activity. Looking at my current list of 542, I see this distribution, noting that start dates can be thrown off by changes in platform or names:
Ø Eight started before 2002.
Ø 24 started in 2002.
Ø 74 started in 2003.
Ø 84 started in 2004.
Ø 154 started in 2005.
Ø 112 started in 2006.
Ø 86 started in 2007.
First conclusion: The liblog universe isn’t shrinking—but it’s not growing as rapidly as it used to. Based on that distribution, the peak of new-blog creation came in 2005, which sounds about right.
This could be nonsense. Maybe liblogs have continued to double each year, such that there are now at least 2,200 active English-language liblogs, but two-thirds of them aren’t in LISWiki or noticed elsewhere. That could be true—but I doubt it. I’d believe 1,100: I can believe that half of the liblogs with more than four or five readers still manage to fly under the radar.
If you subscribed to all of the liblogs in that “visible universe” (542 of them), could you skim the posts without being completely overwhelmed?
I tried that absurd experiment, adding every blog Bloglines could find a feed for (and keeping a handful of official library blogs already in my Library folder, for a total of 551 feeds).
The answer is Yes—although I can’t imagine why you’d want to. Somewhere between 60 and 150 blogs were updated on typical weekdays, with an average of around 250 posts. I needed about 45 minutes to skim them and read the 30 or 40 that interested me. After a week, I deleted three very high frequency blogs. For the second week in mid-March 2008, I counted an average of 221 posts per day.
Just for fun (it’s a meaningless comparison), I compared posts per day per blog for this larger universe with the two earlier studies. For 2008, the average was 0.41 posts per day per blog. For 2006, the average was 0.49 posts per day per blog—a difference but not a big one. For the 2005 sample, the average was 0.92 posts per day per blog, but that was a handpicked set of high-profile blogs.
First, I defined the preliminary universe for any 2008 study. Then I began assembling 2007 metrics, using March-May, to do two things:
Ø Establish an overall baseline for one-year change evaluation if I do the big project.
Ø Look at changes between earlier years and 2007, for the 227 blogs that were in earlier studies and still around.
This time, I looked at number of posts, number of comments, total length of posts—but also number of images (videos, photos, drawings, whatever) in the posts. I also made a couple of other changes for clarity and ease:
Ø If a blogger doesn’t allow comments, I noted that in a validity column and left the comment space blank instead of entering zero.
Ø If a blog’s archives have fully or partially hidden posts, I noted that in the validity column and just counted posts (and comments, if comment counts were visible), leaving the length and picture columns blank.
Ø If a blog’s archives weren’t available at all, I noted that in the validity column and left all columns blank. For the purposes of this preliminary note, I excluded the blog altogether.
Thanks to changes in my work environment and improvements in Word2007, the process of building metrics was much faster than in previous studies, so I could do the first part of the 2007 metrics before deciding whether to do the full study. The rest of this article offers notes from those metrics.
“But wait,” you might say, “you’re mixing 2005-2007 comparisons and 2006-2007 comparisons.” That’s true, but when I looked at the subgroups (43 blogs in the 2005 study that weren’t also in the 2006 study, “all the others”), it appears that changes on a blog-for-blog basis are reasonably comparable.
The universe for these comparisons is 201 blogs. The other 26 were either unreachable or had unreachable or uncountable archives.
The total posts in the earlier study period (whichever it was) for these 201 blogs: 11,948, or an average of 130 per day. Total posts in 2007 for the same blogs: 9,618 or 105 per day—20% fewer. That’s a reduction in overall posts, but not a large one.
For what it’s worth, the earlier average posts per blog (for blogs still active) was 59 and the median was 38—but the standard deviation is 76, which makes “average liblog” a meaningless term. Similarly for 2007: average 48 posts, median 28 posts—but a standard deviation of 76.
Want a truly strange pair of figures? I calculated 2007 posts for each blog as a percentage of 2005 or 2006 posts. The average of those percentages is actually 112%—but the median is 76%, close to the overall 80% figure. The standard deviation for the percentages? 159%.
I thought it was worth looking at changes in post counts on a more granular level, taking into account extremes:
Ø One extreme is blogs with very high posting frequency. I used two posts per day in 2005 or 2006 as a cutoff (that is, 184 posts—but no blogs had between 176 and 194 posts). There are eleven such blogs, ranging from Attempting elegance (which had a different name in the earlier study) with 195 posts to beSpacific with 723, the other nine ranging from 206 to 371 posts. Seven of those eleven are within a change range I’d consider “roughly the same”—from three-quarters as many posts (75%) to four-thirds as many (133%), three having slightly more posts in 2007, four having slightly fewer posts. The outliers: Collecting my thoughts had 36% more posts, Out of the jungle had only 39% as many posts (88 instead of 225)—and Attempting elegance, a very different blog than its high-frequency predecessor, had 7% as many posts (13 instead of 195).
Ø The other extreme is blogs with too few posts in the earlier studies to make changes meaningful. I used nine posts as a cutoff. Thirty blogs fall into that infrequent-posting category. Every blog with more than 3.5 times as many posts in 2007 as in the previous study fell into this category—nine of them, none of which had more than eight posts the first time around or more than 50 in 2007.
Ø Of the other 160 blogs, 56 are “roughly the same”—the 2007 post count is between 75% and 133% of the 2005 or 2006 count. Sixteen had significantly more posts in 2007, including nine with at least twice as many posts. That leaves 88 with significantly fewer posts in 2007, including 46 with less than half as many posts.
It’s fair to say a lot more people did significantly less posting to their blogs in 2007 than did significantly more posting. Will that trend continue for a larger universe and from 2007 to 2008?
Some liblogs don’t allow comments. Others make commenting difficult. Still others, for whatever reason, don’t attract comments. And in some archives with hidden posts, comment counts are also hidden.
For this note and comparison, I’m ignoring any blog that had no comments on posts during March-May either in 2005/2006 or in 2007. That omits some blogs with comments in one year but not the other and leaves 141 blogs with comments in both years.
The overall figure is clear enough: There were significantly more comments in 2007 than in the earlier years—about 37% more. Liblogs were, across a broad sample, more conversational in 2007 than in earlier samples.
For 2005-2006, the average comment count per blog is 57 and the median a mere 24—with standard deviation around 99. For 2007, the average is 48 and the median is 31, with a standard deviation of 62.
One extreme—blogs with loads of comments—could be 800 or 400 in 2007. Two blogs had more than 800 comments: A fuse #8 production (1,689) and Annoyed librarian (813). In the first case, comments more than doubled from the previous study. In the second, where the blog barely existed in time for the earlier study, there were nearly twentyfour times as many comments in 2007 as in 2006. Four more blogs had between 416 and 457 comments: Text & blog, Collecting my thoughts, LibrarianinBlack and Slaw.
Many blogs rarely have comments, but 93 more blogs had at least 10 comments in 2007. Roughly half of those had 50 or more comments, including 22 averaging more than a comment a day.
To consider changes, it’s necessary to look at the extremes for 2005-2006. Viewed that way, there are three “many comments” blogs: A fuse #8 production, Zenformation professional and Slaw. 104 more blogs had 10 or more comments.
Within that middle group, ten blogs had at least three times as many comments in 2007. Another ten had at least twice as many comments. Another eighteen had substantially more comments—at least one-third more than in the previous study. Twenty-one blogs stayed roughly the same (75% to 133%), while 44 had significantly fewer comments overall. In all, leaving out extremes, 48 blogs had significantly more comments, 44 had significantly fewer.
If a blogger posts ten items a day and gets one comment per post, they’ll wind up with a lot of comments—but a blogger who posts once a week and gets 20 comments per post has more extensive conversations.
Let’s look at comments per post. Across all 142 blogs with comments in both quarters, the “average post” had 1.0 comments in the 2005/2006 studies—and 1.6 in 2007, a significant 60% difference. Looking at individual blogs, the average of average comments per post was 1.2 in the earlier studies, 1.7 in 2007—still significant.
Three blogs had very high conversational intensity in 2005/2006—more than four comments per post: Zenformation professional, Vampire librarian, and InfoTangle. All three have lots of comments per post in 2007, although Vampire librarian declined slightly. At the other end, 46 blogs averaged less than one comment for every two posts in the earlier studies.
Leaving out the extremes, five blogs more than tripled the average number of comments per post—and they’re not all within the group that tripled overall comments. Another six more than doubled conversational intensity. 29 more increased comments per post by more than a third. 31 blogs show relatively minor changes and 21 had significant drops in average comments per post—in five cases dropping to no more than one-fifth the figure in the previous study. That’s 40 with significantly more comments per post, 21 with significantly fewer.
Overall? There was more conversation—but not for every blog or every post. The first is encouraging. The second is certainly not news.
Here, comparisons include 162 blogs. The nature of archives in other blogs precluded calculating text length without doing an unreasonable amount of work.
The total length was less than in the previous study, but not by much: 89.5% as long.
For overall length, using earlier figures, there are two extremely long blogs (more than 140,000 words in 2005 or 2006), A fuse #8 production and Out of the jungle. The first is still extremely long, down just 3% in 2007. The second dropped to typical length in 2007. Another four blogs were very long, between 53,000 and 83,000 words: Slaw, Gypsy librarian, Collecting my thoughts, and the blog now called Attempting elegance. One of those grew considerably longer, one stayed about the same and two got considerably shorter.
It’s hard to say what constitutes an extremely short blog, but I’ll suggest two thousand words over three months as a cutoff. Twenty-three blogs had fewer than 2,000 words in the earlier study—including nine with fewer than 1,000 words. Three of those nine wouldn’t be included using my current criteria, as they had only one post during the quarter.
In the middle? Eight blogs more than tripled in length, another seven more than doubled, and another 18 grew by more than a third. Forty-three stayed roughly the same (from 75% to 133%). In all, 33 grew significantly longer and 58 got significantly shorter—including 17 that had less than one-fifth as much text in 2007 as in the previous study.
Length per post is a more meaningful number than length per blog. Overall, posts were slightly longer. For the 163 blogs where I could verify length in both samples, the “average post” was 232 words in earlier samples, 262 words in 2007. That difference is only 13%, not very significant.
In both samples, one blog stands out for true essay-length posts: InfoTangle, which averaged 1,463 words per post in the earlier study, 1,713 words per post in 2007. In 2007, one other blog exceeded the thousand-word mark: Zenformation professional, where the average post grew from 634 words to 1273 words.
I’m not sure it makes sense to exclude extremes when looking at percentage changes, although it may make sense to exclude blogs with very few posts. Looking at the whole sample, here’s what I see:
Ø Five blogs had at least a tripling of the average length per post: blogwithoutalibrary.net, Reflective librarian, Librarian’s rant, Filipino librarian, and Information wants to be free.
Ø For fifteen more blogs, average post length more than doubled.
Ø Forty-one more blogs showed significant increases (34 to 99 percent).
Ø Fully 61—more than a third of the blogs—stayed about (the average post in 2007 was 75% to 132% as long as the average post in 2005 or 2006).
Ø Relatively few bloggers became significantly more concise: 37, including 17 where average posts were less than half as long.
Ø There are six blogs where the average 2007 post was less than one-third as long as the average in an earlier study—but in five of those cases, one of the quarters has so few posts that the results don’t mean much.
Ø Overall, that’s 61 blogs with significantly longer posts, 37 with significantly shorter posts.
For a fairly broad range of liblogs in March-May 2007, as compared to a similar period one or two years earlier:
Ø Bloggers post somewhat less often.
Ø People leave more comments in general and substantially more comments per post.
Ø Blogs are slightly shorter but posts are slightly longer.
Ø The universe of liblogs is probably not growing as fast as it was in 2005 and 2006, but may still be growing. It’s certainly neither collapsing nor shrinking rapidly.
Will I do the full study? I’m still inviting comments; go to Walt at Random (walt.lishost.org) and search for “liblogs waste”—or go directly to walt.lishost.org/?p=751. Comments may help me decide whether or not to do the big study, and will certainly help me decide how to do it—what to include in the book that might result. Comments will be more useful before June 8, 2008.
Cites & Insights is sponsored by YBP Library Services, http://www.ybp.com.
Opinions herein may not represent those of PALINET or YBP Library Services.
Comments should be sent to firstname.lastname@example.org. Cites & Insights: Crawford at Large is copyright © 2008 by Walt Crawford: Some rights reserved.
All original material in this work is licensed under the Creative Commons Attribution-NonCommercial License. To view a copy of this license, visit http://creativecommons. org/licenses/by-nc/1.0 or send a letter to Creative Commons, 559 Nathan Abbott Way, Stanford, California 94305, USA.