Data Mining: Text Mining, Visualization and Social Media

PubSub top 1000, Part 2

Bob Wyman has posted a follow up to my post on PubSub's 1000. I've now figured out the analysis I want to do on the data, but will have to hold off as I am about to go on vacation. Putting that aside, I've been thinking about the middle ranks. As Bob points out, the middle ranks fluctuate in terms of linkRank. I hypothesize that there is some subset of all sites for which a given site has a stable linkRank. This set is effectively the community that the ranked site is a member of. In other words, by having linkRank run agnostically over all sites, the upper, stable sites act whimsically on the linkRank of the lower sites, making them appear unstable when in fact from the community point of view they have a very stable rank.

Think of it this way. If you are a middle rank blogger with a solid community blogging on a specific topic, your in links will be relatively stable - coming from peers in the community. And then one day you blog about something that catches the eye of an A-list blogger. This causes your link rank to fluctuate wildly until the popular topic that you posted on goes beyond the attention span of the blogosphere.  There are two important ideas here:

• We need blogospheres plural, not a single blogosphere - these may be fuzzy, mercurial things, but they make a lot more sense than this top heavy being which we might term blogosphere 0.0.
• We need to think about influential ideas and influential blogs/bloggers as two separate elements of  any model backing ranking of any sort.

Technorati Provides Blog Search

Technorai's blog posted about their new find-a-blog service:

By default, the blogs are presented in order of authority, which means highly-linked blogs appear first. So each of these Blog Finder pages is like a mini Top 100 for any topic you can imagine. You can also sort each tag by how recently the blogs have updated, or alphabetical by title.

This looks like a response to all the uber list complaints, and interesting in light of my earlier post on the opening in the blog space for just such a service as well as discussion on what tags really should be used for. I'm not convinced that ranking by tag authority is the right way to go for the ideal solution (a brief scan of some results suggest this is a light-weight solution), but getting something going in this area is something of a milestone.

TalkDigger Update

TalkDigger - everyone's favourite blog tracker tracker - has updated with a useful facility which gives link count output as plain text. I asked Fred for this feature and he put it in over the weekend (thanks Fred!)

It allows you to simply cut and paste the results of a search for simpler inclusion in blogs.

http://talkdigger.com/index.php?inline=true

http://fgiasson.com/td/index.php?inline=true 

Here are sample results for a search on www.instapundit.com:

bloglines: 0, technorati: 0, pubsub: 2,020, icerocket: 0, feedster: 9,217, blogpulse: 14,113, msnsearch: 600,939, google: 80,900, blogdigger: 0.

(Fred might want to be a little more lenient with timeouts...)

 

The InLink Discussion

Jason Calacanis' post on Technorati's top 100, Bob Wyman's response all make for good reading, but Mary Hodder's recent sketch of a ranking system based on a mixture of features trumps them all. Hodder gets back to one of the key issues of the discussion which is: link based ranking is a proxy for influence. This is something that Sifry has always been careful to articulate.

The notion of a proxy for influence is important because it leads to: is it a good proxy? and, is influence really the thing that we are interested in? Wyman's response is interesting in that it offers some great insights into how PubSub think about the various features (a subset of Hodder's features). The notion of decay with respect to the time that a link was created and with respect to the number of links from a single site is as insightful as Hodder's comment that a link in a blogroll now is far more selective than one from last year (when there were fewer blogrolls to deal with).

My take:

• A ranking like Technorati's is not as sinister as some make it out - it is what it is and we can take or leave it.
• The user needs that a ranking of blogs satisfy are as numerous as the methods we could use to rank them. Hodder calls for a transparent system, but a system with the number of features that are listed in her post (19) would be transparent but inscrutable.
• Transparency is not independent of simplicity and simplicity can not serve all use cases.

The fundamental problem is that there aren't any good models of the blogosphere that have been adapted to ranking or other sorting problems. In addition, the articulation of the desired function has been lost in the high dimensionality of the things that we could measure or observe. This is why the simple link count is such a great whipping boy - everyone understands it and no-one likes it.

So rather than a complex transparent feature set, I would rather see a number of simple ranking models. For example, blogroll based ranking, site ranking, link ranking, media mention ranking, and so on.

UPDATE: Chip Griffin posts very similar sentiments.