It's been all over the web: Google engineers have developed a refined image search algorithm that will, once it scales to current requirements, replace what's in use now. It'll also replace, as Searchengine Watch points out, the many hours of free labour of those who lent a helping hand with the Google Image Labeler.
Was it an oversight? As John Gruber pointed out yesterday, HuddleChat, one of the first Google Apps introduced and causing immediate controversy, was a clone of Campfire, an older chat application created by a small software house out there. Later in the day he says: Even if you think it's OK to copy someone else's application feature-for-feature, the big fear for developers with something like Google App Engine is that you're trusting Google with all of your source code. Why should small indie web developers trust Google when the first example app is a Google rip-off of a small indie web app?
Google's Marissa Meyer hints in an Interview at the possibility that Google might identify your friends through Gmail or third parties. Probably not, to help you meet those friends again.
http://venturebeat.com/2008/01/31/googles-marissa-mayer-social-search-is-the-future/
The one thing that's really surprising after last week's Pagerank update, which led to a lot of sites apparently engaged in link selling to lose big chunks of Pagerank, is the utter silence from Google.
Which says a lot, really, as in the past on similar widely reported occasions someone somewhere at Google in the end said something, even if it was termed a personal opinion and not "official" policy. This time there was nothing.
After a lot
of sites were left out in the cold in the wake of last week's Pagerank fix the same
changes have now affected link selling sites in many countries. There are, however,
some sites whose PR has been increased although they have been selling links
ads for years.
People who have avoided dodgy links will be smirking gleefully. But I fear that some shady characters might be rubbing their hands: time to hyperpromote forum posts and blog comments with entries such as "I couldn't understand some parts of this story, have to check elsewhere". The signal to noise ratio probably takes another hit.
Quite a few people believe the way Google's Pagerank algorithm juggles value has been changed. A number of sites selling text ads appear to have lost quite a bit of PR. Downgrades are between 2.0 and 3.0 on the one to ten Pagerank scale for popular sites including autoblog.com, forbes.com and the washingtonpost.com.
Others appear to have stable PR values even if they've been selling ads through one of Google's mini competitors or on their own for ages. Could this be another high profile demonstration like the [very temporary] banning of BMW last February to signal that links aren't a commodity?
So we read again that links should not be sold, and that the powers that be may or may not devalue or punish sites which sell links like others sell merchandise. Nothing wrong with that, and nothing new really, since Google started to be concerned with this very problem some years back. I didn't hear any alarm bells ringing at the time when Google announced their interest in programming solutions to "Detecting common templates in pages, and separating out the common structure from the individual content, where all the sins take place.
To me, and probably some others, it was obvious, they were going to try to differentiate between "real" page content and all the bits we humans routinely ignore because of their irrelevance.
Links aren't - and not just in theory - that different from degrees. One you bought at a degree mill isn't really that prestigious. Applies to web as well: if lots of people really praise your site, it not only will be recognised as what it is [an opinion expressed through linking], it will also be as close to eternal as it can be on the web.
After some email exchanges several years back I always regarded Urs Hölzle as someone whose manners showed an enormous amount of potential.

AP now confirms that my first impression was probably right.
© Copyright 1998 - 2008 Klaus Schallhorn.