Some people are up in arms, because Google appears to play with the way servers log referer information. Instead of using the established way of tucking on name & value pairs to a URL string by using question marks and ampersands they now use a hash sign. Problem is, the '#' in a URL is traditionally a local jump target, so servers only log what's to the left of the hash sign in a URL.
Why is that important? Because there are still people who tune their pages - after they've been published - to contain phrases or terms used to find those pages. Problem is, this approach leads to reduced link popularity, simply because if you start adding keywords and phrases in order to accommodate how people search, you at the same time risk that people start pointing to their forehead when reading your stuff. When that happens, it's more difficult to get linked.
Allowing keywords and phrases to come from external links is a far better approach to extending the vocabulary of a piece of prose than working every conceivable term into the fabric of one's writing. It will, over time, still "contain" all those strings you'd normally pluck from from your referer logs, but once people arrive at the page in question, it will still be worded as you intended. Anchor texts still work as always - unless you're looking at a high profile page where the results have been adjusted by Google.
Searching for "ibm thinkpad t43 cmos battery" - I bought a T43 to be able to work in the garden from next year on, which in itself is a long story - reveals the effect of combining bugs of some sort, or some incredibly stupid spam. When selecting "Pages from the UK only" the results reveal several domains [plural] beginning with 2008 [2008latest5.net/, 2008flashin5.net/ etc] in top positions with nothing showing up in the browser window.
It's possible that the content served by the domains in question comes from a server which keeled over last week and therefore doesn't manage more than an opening title tag and a partial title text. But then a title with more than 1000 lines is somewhat excessive, even if the closing tag is missing.
What I don't really understand is what they're trying to achieve with this sort of rubbish, especially when there is more than one domain involved, and more than one subject. In fact a site search reveals many subjects from healing to cheap travel.
I find it even more revealing though, that Google not just manages to position this, but that it gets indexed at all. Don't they have a trash can?
WTF do these people think when they talk about "soft 404s and hard 404s? Why not call those what they are? Errors. Problems. Laziness. Ignorance. There's an HTTP Status code for most of them.
Danny Sullivan has an interesting article on Knol the day after its debut. Looks like there is an "unexplained" ranking boost[1] for pages in Knol, even those not linked from Knol's start page. We can't but accept Google's desire to replace Wikipedia's untold good ranking and ad free pages with something that registers in the till - but does it have to be so obvious?
[1] http://searchengineland.com/080724-140223.php
Google now offers a way for you to see who can access your Gmail. What can one say...
In a just published patent application Google can be seen to be considering the monetisation - what an ugly word - of your friends. You still can't buy friends. But soon, thanks to Google, you may be able to sell them.
© Copyright 1998 - 2009 Klaus Schallhorn.