Distributed Tagging Hell

It all started two weeks ago when I undertook the "drop-in" upgrade from Movable Type 3.14 to 3.17. Well, "drop-in" is not really the word. I had made some changes to my server configuration that I thought should not affect Movable Type and, in fact, did not for version 3.14. However, because of a change in an open source library component that is a dependency for Movable Type 3.17, my configuration broke Movable Type 3.17. As a result, Movable Type was not properly pinging the technorati server to put my tagged links in their tag repositories. There were other smaller and larger issues that I independently tracked to this same dependency.

Well, if you have read the jillions of posts I have made on tagging, you will realize I was in a panic. What!? My posts are not working in the thing I talk about so much.

A quick note to Ryan King, the ever responsive microformats enabler at technorati, revealed that the problem was pinging and that mine had not worked in two, yes two weeks of many posts.

I fixed pinging. Then, as an anti-spam measure, technorati only allowed one of my posts through (per Tantek Çelik, of technorati, multiple posts in a very short period means spammer, therefore they throttle). So, I decided to repost some of my (to me) more poignant posts. Hence, duplications in this feed and 4, yes 4 trackback pings to this post by Tim O'Reilly because of Movable Type's automatic trackback feature. Tim, I especially apologize to you for all of these trackbacks.

Yesterday, Tim O'Reilly discussed two models that are emerging in Web 2.0, one ring to bind them and many pieces loosely joined. The crowd at the conference is definitely of the "loosely joined" persuasion (with the exception of all cell service providers in attendance and many of the large companies, but they are a minority here). Movable Type integrated with technorati is the loosely joined approach. MSN Spaces and any of the recent Yahoo social software attempts are the one ring approach.

The problem with the loosely joined approach is exactly what I experienced. My republication reaction to the broken linkages which was then amplified across the loosely joined pieces is what created the duplication problems. The advantage of the one ring approach is that you have less potential for this kind of mayhem, albeit at the cost of the loss of a lot of control.