If you have no interest whatsoever in an Inside Baseball post on the blogosphere, Horseman, pass by.
A little over a month ago, N. Z. Bear made a change in the EcoSystem. Now, I have a high regard for N. Z. Bear: I think he’s one of Nature’s noblemen. The EcoSystem, his ranking system for blogs, is a fantastic contribution to the blogosphere as a whole and a labor of Hercules to maintain. IMO it gives the blogosphere a greater sense of community. If N. Z. Bear did not exist we would be compelled to invent him.
As best as I can deduce the change had several components. The first was a long overdue re-calibration of the phyla of the EcoSystem The second was a change in the algorithm by which the rank of individual blogs was determined. Blogs are now ranked not by links (or, more precisely, not merely by links) but by the blog index. This blog index is shrouded in mystery (mysterious are the ways of the Bear!) but I gather it’s calculated based on both the number of links and on the ranks of the sources of the links.
I’m beginning to wonder if this change was prudent. Hypothetically, one’s rank in the EcoSystem is a measure of one’s influence and, consequently, should go up and down as one’s influence wanes and waxes. That’s feedback that I, for one, have found useful.
I’m hungry for any form of feedback. Emails. Comments. Trackbacks. Whatever. It’s a combination of measure of how I’m doing, reward, punishment. And one’s EcoSystem ranking is another form of feedback. The problem is that, unless I misunderstand what’s going on, the use of the blog index actually makes the ranking a poorer measure of actual influence than it was previously. Let me give an example.
Let’s say you have a blog, A, that’s a highly influential blog, and B, C, and D, which are substantiallly less influential blogs. B, C, and D all link to A and to each other. A links to B. That will cause B’s influence to rise (and C’s and D’s, as well). Is that correct? Do C’s and D’s influences actually rise as a result of that single link?
Now extend the example a little. Change the previous example a little. A links to B, C, and D. The influence of B, C, and D will go up substantially because of that link and the mutual links amongst themselves. My intuition is that that is really not correct.
One final adaptation of the example. Let’s say that B’s link from A for some reason disappears. The influence of B, C, and D will go down a lot, without any of them having done anything in particular and probably without any actual reduction in influence. That’s why my hypothesis is that the new ranking is farther divorced from actual influence than the old one was (and the old one had problems of its own).
So, what characteristics should a good blog ranking system have? First, I think it should be transparent. Transparency gives the opportunity for playing the system but lack of it leaves more openings for outright errors. Transparency will, I think, confer legitimacy. And open the ranking system up to improvement through suggestions from the blogosphere.
Second, it should be based on both links and traffic. There are problems with both links and traffic as measures of performance: not everybody links and RSS causes a misrepresentation of traffic as does the existence of large group blogs i.e. communities like Daily Kos.
Third, the ranking should be a trend ranking. By that I mean that real influence (it seems to me) changes based on whether on the rate at which a blog is becoming more influential or less influential. Second differential issues, if you will.
Well, those are my thoughts for right now. You?