You may be wondering what that strange “Blank post” was all about. The science blogging crew has been having a discussion about the “Most active” box you can find in the right sidebar, and RPM challenged me that I could put up an empty post titled “Blank post” and it would get 10-20 comments. I proved him wrong—it got over 30 comments in less than an hour and a half. That’s just wicked, people.
It also made the #1 most active post on scienceblogs. Man, my job just got a lot easier…who needs to write anything?
I think my case is made. Using traffic activity to determine what links you’ll put up is a perfect example of a positive feedback loop, and it’s also why ranking systems like Technorati that base your position in the hierarchy on how many links are made to you or how much traffic you get contribute to the perpetuation of that same hierarchy. It stabilizes rankings and can mean that the best are hindered from rising to the top, and cushions slackers so they can coast. Despite the fact that I’m benefiting from this, I don’t like it; here I am with my socialist leanings, wearing a top hat and monocle and reaping my ill-gotten harvest with no effort.
So I’ll throw it out to you. What information would be most useful to you in that box, that would also be easy to implement, and that would subvert the dominant paradigm by distributing links more liberally?
P.S. This blank post idea is interesting, but no, I hadn’t heard of it until it was brought up in the blank post thread.
It must be admited that Jason’s nice article on Orr’s review of TGD got into the infamous box.
I like the idea of the Most Active box. It shows which posts have the most active conversations on them at any given time. I like this function, but maybe Seed could tweak it some.
PZ, you always dominate it because you get as much traffic as the next three highest put together, so you always seem to dominate this list.
Therefore, I suggest that Seed keep the Most Active box the way it is, only limit it to a maximum of two posts (or maybe just one) from each blog.
Here’s my proposal: A carnival of the unpopular. Use stats to track which blog posts are least active in a particular category. Periodically review them, and then post a list of the inactive posts that are of high quality. Unfortunately this isn’t something that can be implemented robotically, like the present contents of the infamous box.
In passing I will note Afarensis did good review of the ‘Neandertal Cannibalism’ paper which made the news some time ago, and has finally been published. They survey 1330 skeletal specimens!
How about weighting blogs inversely to their traffic? That way a lesser traveled blog could make an appearance on the list if there are a few comments to a post, while you (PZ) would have to generate interest comensurate with your traffic load to get on the list. Top 5 Most Active Relative to Traffic
A quick look at the ‘most active’ box indicates that the most important of activity is having the name “PZ Myers.” So obviously, if other ScienceBloggers wish to increase their traffic, they should change their names to match.
Yeah, the whole generating-traffic thing can be a head-scratcher. It does seem as if there are the celebrity blogs, and then there’s Everyone Else. Usually when you get a link from one of the big blogs, it can make people aware of you more effectively than anything. When PZ linked to us over at the Atheist Experience that one time, it tripled our uniques in a day. Thing is, how does one keep those people coming back and commenting…?
How about a quality ranking system, like on Amazon? There could even be several categories (educational, thought-provoking, good polemic, etc.).
Content. PZ can lead readers to your watering hole, but he can’t make you post squid porn.
Easy. Keep the Most Active box, but don’t count people who click on a link from it. Just count, as activity, people who click from elsewhere. That breaks the feedback loop.
Well, you could always make it completely subjective and have “the box” list the author’s five most gratifying posts based on his or her impressions of the comments received. Since I get so few comments on my blog, that’s how I rate my posts *shrugs and shakes head sadly*.
This would side-step the current paradigm, yet provide a measurable standard (although a relatively asthetic one) by which to qualify the post ranking.
Just a thought.
By the way, I just had to post a permanent link to you guys on my little publication. This is a great blog. Useful and frivolous at the same time. What could be better?
Thanks,
KC
Thing is, how does one keep those people coming back and commenting…?
Content. PZ can lead readers to your watering hole, but he can’t make you post squid porn.
There’s more to it than content. It seems like people feel freer to comment where lots of other folks have commented. Getting comments on a new/not widely read blog can be tough. (Another positive feedback loop.)
Not as wicked as if you had wagered some money on it.
That requires some sort of rating system. Some blogs are unpopular for good reason, such as John A. Davison‘s.
I’m sure there’s a business opportunity in that. Blogfans R Us, or some such.
Sorry about that, but I had a late lunch today.
I’d like to see blog authors nominate post of their own — likely ones they feel are important for whatever subjective reason (perhaps ones they’ve invested a lot into writing) — and have what is now the “TOP FIVE/MOST ACTIVE” (what does that even measure?) box change to list five randomly selected from those nominations (updated hourly or on page refresh or …). That way I’d be getting pointers to posts that the authors themselves feel are especially worthy/interesting. You could call it “NOTABLE POSTS” or some equally innocuous phrase.
Or maybe a “SciBloggers Recommend” box, where bloggers point out posts at other of the growing family of ScienceBlogs which they think are worthy of attention.
Damn! I knew I misspelled aesthetic in my earlier post and forgot to go back and change it before submission. That’s just sloppy.
KC
If you post links to the most active/most popular threads, as you noted you’ll simply get a positive feedback loop. Threads that succeed in attracting attention will get even more attention, and obscure but potentially-interesting ones will be excluded.
But clearly putting up links to only the least active threads isn’t viable either. There are far too many of them, for one, and the least active/popular are often that way for a reason.
Choosing some other metric has its own problems, as the metric will either a) be related to popularity, b) be so specific as to not be useful to many people, or c) be arbitrary.
So: why not list some randomly-chosen threads? If people are interested in the titles, they’ll click through. If they’re not, they won’t.
(Additional: I know the whole “socialism” thing was a joke, but you’re still all getting free rides from Seed.
Equally distributing popularity and attention is not only ludicrous, but impossible to implement, and expecting it is just dumb. If unpopular ScienceBloggers want more attention, they’ll have to ‘earn’ it.)
How about the most divergent articles – posting on a subject that correlates least with other posts in a given period. What no one else is talking about, sort of.
I think you may be guilty of generalising too much from a single data point. Try making just blank posts for a month or so and see the effect over time…
Your point about perpetuation has some merit, which is why we changed the Technorati ranking system over a year ago to have a 180 day integration window, meaning that you have to keep getting people to link to you to stay highly ranked. This has given a higher turnover in the top 100 and other blog tagged rankings.
As you might expect if you’re familiar with Duncan Watts research, the rankings follow a power law distributions, however if you’ve read his work more closely you realise that the changes in rankings do so too.
“easy to implement” is the stopper.
If you are a robot, it is no problem. We just just write a program which emulates you. If you are smarter than a slime mild, any successful program would want to write its own blog.
Technology aside, it’s like keeping in shape by hiring a kid to jog around the park in your place.
Content still matters. The blank post does not indicate that you can maintain popularity without writing anything. The post’s popularity is itself based on the popularity of your blog. The novelty of the blank post is inherently attractive to commenters at a very active blog with lots of visitors. People will post comments just to attract clickthroughs to their own site, or because they know they will be read by a wide audience – that is why the “open thread” posts work so well on your blog.
If you tried submitting 5 blank posts in a row I bet they would not all make the top 5. And if you started submitting several blank posts per day over a longer period, or if you posted nothing but blank posts for a week, interest in them and in your blog would likely wane further…except that it is no longer possible for you to run these tests since you have thrown off the results by announcing your intentions!
As was mentioned above, it probably makes sense to filter out any referrals from the “Most active” list when determining which topic is most active, if possible. That will at least remove the “positive feedback loop” problem.
But that still leaves you with the list being little more than the most popular topics on the most popular blog. Perhaps it could show the topics that are most active compared to the average for that blog? Or maybe there is no good “automatic” solution.
PZ,I wonder how many of your curent and former students contribute. I think it shows how much influence and reach you have when those people are contributing. I am not one of your students unless you consider an interested Geophysist a student. I like the biology and biochemistry discussions a LOT.
Nothing. I come here to read your posts and occasionally the comments. I rarely read and never click through the sidebars with the possible exception of your random quote generator.
This is an interesting list of advice how to encourage more comments.
No, that’s a negative feedback loop.
:)
Blnk pst.
Frthr vdnc tht ths s mndlss, mnnglss ch chmbr whr ppl r nly ntrstd n hrng thmslvs tlk. . .n ntrst whtsvr n cntrstng pnn.
hoody’s post.
Further evidence that this is a mindless troll.
hoody:
I can’t remember you ever doing anything but whine about how nobody listens to you. In fact, I don’t even remember you ever explaining why we should listen to you, or, for that matter, offering us anything actually worth listening to in the first place.
I don’t think that quite qualifies as a “contrasting opinion.”
Further evidence that this is a mindless, meaningless echo chamber where people are only interested in hearing themselves talk. . .no interest whatsoever in contrasting opinion.
Hoody, your mom’s calling. She wants you to turn off the computer and go to bed. She’s paying the electric bill, after all.
“Recent posts” (SB-wide) above, under, or instead of top 5
Batting average — rate the average response (what % of readers voted for it, whether reader stayed on page for more than 1 min, whether reader commented) rather than counting responses
Collaborative filtering, related posts, or ScienceBlogs-wide tags — “other posts you’ll like,” not “the posts more folks clicked.”
Emphasize the derivative — growth (expected future popularity?) not current popularity. Low percentage traffic growth -> off the list.
Visited links disappear — always 5 posts you haven’t seen
One thing that might help would be to normalise the activity, by dividing the activity rate by the average for the blog. What should happen is that the more popular blogs get a relatively constant rating (because of the Law of Large Numbers), so the good posts on the less popular blogs will get a boost.
If you want a deeper explanation of normalization, ask Bill Dembski.
Bob
I usually get a good fifty percent increase in traffic to my blog–that’s not much of a feat seeing as I only get about 12 visitors a day, but I’ll take what I can get–after I post a comment here. I should really comment more often.
Hmm, PZ’s three entries on the top 5 were the blank post (first), this post about a blank post (third), and “Cheaters usuall don’t prosper”. I’m sure there was a deep irony in there somewhere, but fortunately it’s been overtaken by the ice age.
Bob
How ’bout a “Random Scienceblogs” box which presents one (or five) headlines & opening sentences from the set of front-page postings currently offered by all of Seed magazine’s online menagerie, structured so that no two viewers see the same selection?
Well, if you want to do it automatically, you could implement the bureaucratic solution: as other folks have indicated, the Amazon/Daily Kos-style commenter ratings of posts. I know for me, after I’ve scanned Kos’s front page, i always take a look at the top-rated diaries over at the sidebar–often a lot of good stuff there. I also like the “diary rescue” feature they’ve implemented, and it seems like that too would be pretty easy for sciblogs to implement.
Being Olde Fashioned myself, I like the human touch, so I come here first, and then try to follow links PZ points out. I’ve found much awesomeness that way, such as the squdtacular Deep Sea News. I think as long as you remain generous with your links PZ you’re not in danger of turning into a moustach-twirling blog fatcat.
You could call it “NOTABLE POSTS” or some equally innocuous phrase.
I like that idea. It’d make people actually post,hoping that the Great PZ Myers of Minnesota will notice them.
It’s like a gold star glued to your forehead.
Brilliant! (sorry, but I just cannot quite duplicate the accent in those Guiness commercials)
Ok, then PZ, how about throwing a little liberal love my way? [/end of shameless blogwhoring]
I like the idea of weighing the blogs inversly to their average traffic. Or put something in so if the amount of increase in traffic only results in being posted on the front page if it has some cut off level of statistcal significance to the background traffic.
I figure it can’t be done. IBM did lots of work when they developed the “links to” concept to show importance. And once we all learned that we devoloped systems to “use” it and so made it, well, less useful. The same will happen for anything else we develop (I predict). People will work to the testing criteria and so will devalue the test.
I suspect the best measure of an entry’s usefulness is the average length of page views. If I go to an entry and it doesn’t interest me I leave pretty quickly. If I go to one packed with valuable info then I am likely to stay.