An
impor­tant part of the redesign was an auto­mat­ic key­word generator.
Posts were run through a script that auto­mat­i­cal­ly pulled out keywords
from the text. My 2003 arti­cle, Going all the way with Mov­able Type gen­er­at­ed the fol­low­ing tags, which appear as links after the post:

• choate
• non­vi­o­lence
• curly quotes
• per­son­al blog
• php
• org
• move­able­type
• mov­able type
• hmtl
• dash­es
• tex­tile
• garbage
• blogs

Fol­low­ing the links takes you to similarly-tagged arti­cles. At least
that’s the con­ceit. When you fol­low a tag’s link you’re sim­ply doing a
site search for that key­word. A lit­tle htac­cess rewrite mag­ic is making
the result look like it’s a sta­t­ic cat­e­go­ry page.

“Fine and well” you’re think­ing, “big deal.” Well, here’s what’s
cool. There are 225 entries on the Quak­er­Ran­ter blog. Google’s just
gone through and indexed the site and is now claim­ing it con­tains 1300 pages.
Each tag is being indexed as its own page. Every time I men­tion any
inter­est­ing term, it becomes a page that Google index­es and deliv­ers to
its searchers.

Which brings us to today’s cool piece from the access logs. In
Decem­ber of 2004 a rather inno­cent post on Quak­er Ranter became the
cen­ter of a mini-whirlwind on the polit­i­cal blogs when it mentioned
that I had got­ten a call from a CBS News pub­li­cist inter­est­ed in Non​vi​o​lence​.org.
All polit­i­cal blogs get pub­lic­i­ty calls from news and opin­ion think
tanks try­ing to sug­gest (or plant) sto­ries but no one’s sup­posed to
talk about it. I only men­tioned it because it was so unusu­al. One of
the blogs denounc­ing the lib­er­al con­spir­a­cy my post revealed was the
some­what slimy Lit­tle Green Foot­balls. After a few weeks the
denun­ci­a­tions died down. 

But this morn­ing, some­one looked up lit­tle­green­foot­balls in Google and came to my site. Because of my auto­mat­ic key­word gen­er­a­tor, tags, and static-loooking links, I’m now the num­ber two entry, on two three-year old posts, now relo­cat­ed to a days old quak​er​ran​ter​.org. Cool. 

This mix­ing and match­ing of con­tent and rich manip­u­la­tion of data is some­times lumped togeth­er in the cool bu zzphrase folk­son­o­my.
Note that none of what I’ve done is a trick­ing of Google. Every tag is
real­ly going to a page with that con­tent. These are “nat­ur­al” and
“organ­ic” search results in the lin­go of SEO. I’m just pre­sent­ing my infor­ma­tion in mul­ti­ple for­mats that appeal that the widest array of audiences.

For what it’s worth, I don’t think I deserve #2 sta­tus for
“lit­tle­green­foot­balls” and I don’t think Google will keep it there for
long. It’s a bit odd that they have ele­vat­ed that par­tic­u­lar term so
high and no oth­ers tags seem so stratospheric.

Positive Results:

As of Feb­ru­ary 2007, Google index­es 3,540 pages
on Quak​er​Ran​ter​.org, a blog of only 239 posts. In Decem­ber 2006 30% of
my Google vis­its were to one of the “tags” page. Recon­fig­ur­ing the blog
in this kind of tag-intensive way has more than dou­bled search engines
vis­its, again in a very nat­ur­al and organ­ic way. Adding tags has simply
made what I’ve writ­ten more acces­si­ble to search engines. Very cool.

Negative Ramifications:

Short­ly after installing this new sys­tem, my servers started
peri­od­i­cal­ly crash­ing (about once/week). The prob­lem would be multiple
MT-Search process­es over­load­ing the memory. 

My guess is that a search engine spi­der came along and started
index­ing all of the tags. Each link ini­ti­at­ed a search query in Movable
Type. The built-in search for Mov­able Type is just not able to handle
this vol­ume of traffic.

I installed Fast Search to solve the prob­lem (tip of the hat to Al-Muhajabah). It took awhile: Fast Search required a MySQL upgrade at my host. After that I need­ed to install these plu­g­in fix­es.
Then it was fine-tuning the htac­cess files. It was been more work than
I ini­tial­ly expect­ed and the tag results now for­ward to a fun­ny URL that Google does­n’t love as much.