I believe I understand the model, based on cross-links, for identifying the best blogs (and perhaps postings). A problem with it is that a new blog doesn't yet have a lot of cross links and so tends to get lost in the shuffle. It may never get enough cross links to receive the attention it deserves, depending on the amount of better-established competition. How does it get cross-links?

The blog world is ultimately based on a system of gate-keepers. That's why there is "blognosing." To the extent that cross-links are the way that quality is measured, politics, "who-you-know", and skillful blognosing will always be a determinant of success on at least the same order of magnitude as the actual value of the output.

This is not because of ill-intent on anybody's part of a desire for power on behalf of the gate-keepers. It's just an artifact of the way things are done now. And it certainly isn't optimal.

In personal email, Robert Scoble of Userland responded to the above points:

"I found this isn't a problem. All new weblogs show up at http://www.weblogs.com and experienced webloggers watch there and visit quite a few of the new sites. That alone makes it pretty likely that someone will see your first weblog and if it's good it'll get pointed to

That is true... However that solution doesn't scale to the world where there are 2 or 3 orders of magnitude more Weblogs... even the most dedicated blog-surfers won't be able to randomly visit any but a trivial percentage of the new sites. And that is the world many people expect to come into being.

Also there is no reason to assume that a particular individual who is very skilled at writing things of benefit to others will be the same person who is prone to reading a lot of other blogs and linking to them... there is no reason to assume those tendencies are strongly correlated, and yet, as Robert and John Robb's recommendations for bloggers, that is the main basis for getting known in today's world. And it will play a much stronger role, even, when there are orders of magnitude more blogs, because the random discovery solution will be steadily less and less workable. Eventually, successful bloggers who are linked to a lot won't even have time to explore most of the blogs that link to them.

This will lead to an "increasing returns" effect, where the best known will become exponentially better known with a rate of growth in popularity that is proportional to their existing popularity, and the least known will find it extremely difficult to break in. The growth rate in popularity of the most popular blogs will greatly outstrip the growth rate of popularity of newcomers no matter how good the newcomer blogs are because of this statistical law of nature. This superior growth rate of the most popular blogs will continue until the market of blog readers is saturated. But then, because readers don't have unlimited time to read blogs, and all that time is already being spent reading already-known blogs, it will be extremely difficult for new ones to break in.

The same process occurs for a very wide range of phenomena, including the growth in the population of cities. This kind of distribution of popularity is known as an exponential distribution, frequently a Zipf distribution, which is a straight line on a log-log graph.

In particular, the number of links to Web sites tends to form a Zipf distribution.

Just as it is unlikely that a brand new city will emerge in the U.S. anytime soon that rivals the size of New York, it is unlikely that, once the potential audience of blog readers is reading as many blogs a day as they have time for, it is unlikely that the leaders will be displaced by newcomers in any reasonable time frame.

For more on the Zipf distribution as it applies to Web sites see http://www.google.com/search?hl=en&q=links+zipf+web&btnG=Google+Search

THAT ALL BEING SAID, obviously Robert is right in that the current blog world has a solution, and the degree to which it may be suboptimal is actually unknown -- the arguments I make above are all merely theoretical. Despite my arguments, the solution already in place may be close enough to optimal that there is not sufficient motivation to bring in another one. There's no way of knowing at this point.

In particular, in corporate installations of blogging software for knowledge management and distribution, the problems of scale will not be as strong as described above because the internal populations of bloggers will be much smaller.

So the question of whether my company's technology, which has the aim of solving the problems discussed here, will be adapted for the blog application is open at this point. We'll be talking about it more in the coming weeks.

Please feel encouraged to email me with any comments at grobinson@transpose.com. (I'm going to have to build my email address into the template for this blog soon!)
9:39:50 AM