Well I’m going to indulge myself today. I’m going to talk about something I do not understand, but I do that a lot. SEO! I try to read articles about the algorithms used by MSN, Yahoo, and Google and I get the part about LINKS, but I do not get that linking approximates relevancy values.

I can accept that links are good, but they can’t be as good as the algorithms have weighted them. When searches turn up results that take a lot of liberty with your search or keyword phrase, but you see these webpages have big rank and links, and you understand that rank is based in part on links, and that links themselves are further valued or devalued, then it gets a little far from relevancy.

I can understand some offered variations of the searchers keyword selection, because this is in part informational. I could search a phrase like “webpage templates” - which I did using all of the three search engines listed above - and get an array of results with keyword phrasing like “web page templates”, “web template”, or “pages of a website template”. But it is just as irrelevant as it is informational in many searches, sometimes more irrelevant.

If I look at the results of the three searches of “web page templates”, there are some similarities and some dissimilarities. On Google today the search for web page templates gives results whereby the first 7 results contained the word “free”, and the free was almost a part of the phrasing in most of the first seven. I don’t know why, but I do see that those first seven results offered by Google include some much looser variations of my search phrase: “Free Web Templates - Get free website templates and page layouts …”, “Free Website Templates, Free Photoshop Web Templates, Website Design”, “Offers free website templates and PhotoShop web page layouts”, “Free web templates by Art for the web”, “Free website templates from the Free templates experts”, “PageKits: Free Web Templates from . . .”, “Free Web Templates is the biggest free website template community online.” And one other similarity - they averaged 65,000 links apiece and had a PR average of about 6.25. Also #3 and #4 results were the same website - mmm. The #12 result on the Google results pages was also found on Yahoo at #30 and on MSN at #6. It had 4,150 links.

On Yahoo the “web page templates” search gave the same website the #1 position as on Google, but the word “free” did not appear in any of the of the results wording(?). And neither did the word “free” appear until the #5 result. The first seven averaged about 16,000 links if you leave out #5 which shows 9,130,000 links(?). Average PR was 4.0. But lookie here! “Offers web page templates”, “web page templates”, “web site layouts and templates”, “about web page templates”, “Simple Web Page Templates”, “Web page templates”, “Free web page templates” - less variation of my search phrase.

In MSN #1 was the same as in Yahoo and Google. The #3 and #4 positions had the same website as #3 and #4 on Google. The word “free” was mentioned in only two of the top seven positions, and not highlighted. Couldn’t find the number of links each of the first seven had, but three of them appeared in the other searches top seven as well. Now to the bastardization of my search phrase: “cool free web page templates”, “Web Site Templates”, “Web Page Template”, “web page templates”, “Web page template”, “Web page templates” and “Web Templates” - not as bad as Google - not as good as Yahoo.

Tempest in a tea cup? I like to search for what I’m looking for, not what is heavily linked and maybe in the ballpark of what I’m looking for. And, I believe I am not the ONLY one!!! This keyword phrase is much simpler than most that I need to search, and the ballpark is sometimes not even on the map. Which leads me to my Unified Field Theory - well my hunch - that a better measure of relevance (which really IS a good thing!!!) would be the click rates to websites that have the correct keywords, and further, click rates inside the website if these could be tallied by some mechanism. And you could factor out position # that might overenhance click rates and use some feature that discounts exhorbitant click rates from machines or the same IP addresses.

Okay, I am ready for the blast - how does my idea ignore the obvious. Must we Link or Die?

Arthur Browning