Google Datacenters Watch 2006-04-06

[bobmark]

< continued from [webmasterworld.com...] >

Reseller,

Having looked at these dc's in some detail, I have to disagree with you (tho glad you're making a comback!).

Because of major changes to a site I have in the OCT-FEB period, I can pinpoint the vintage of the index quite easily (and confirm the site specific results with others I am familiar with). What I see on these dc's is an index made up of ancient results (circa AUG 2005) with some additions from FEB or so.

If this truly is the result of the carnage of the switch to BD then Google has accomplished nothing in the FEB-15 to current timeframe.

If as you say, Google knows exactly what they're doing, what they are apparently trying to achieve is to augment the AUG-05 index with a few new pages. If you mean Google intended to "break down" the entire index and rebuild it "live" - unlike in the past where more or less finished updates rolled onto all dc's over a 5-7 day period - then I might agree. What I can't see is the reseller frindly ® dc's being anything but yet another interim step. The difference is, I see the process as scambling to try and fix an unanticipated scru-up, you see it as an deliberate plan.

Either way, I'm counting the days until Microsoft rolls out their new product as Google is ripe for the taking ... takes awhile for the public to turn against a SE but there certianly is precedent (remember when you couldn't turn on your TV without seeing ads for Lycos?).

From a webmaster point of view, the ideal world is MSN, Google, Yahoo with 30% market share each.

[edited by: tedster at 10:32 pm (utc) on April 6, 2006]

[selomelo]

After the full propagation new toolbar PR, I see a slow but gradual reshuffling (recalculation by the new Mozilla?) in index accross the datacenters. As a result, changes in serps are barely noticeable, as if arising from the so-called everflux, as contrasted to more disruptive changes in the past. It is as if Google is adding parameters bit by bit to its new algo.

An actual observation: Some long "buried" keywords for one of my sites started to climb after the last TB PR rollout. First at +#400, then at +#300, then #280, #230, #180, #167, and now at #128 and climbing!

I do not know where it will stop, but it is very interesting to observe. :)

[frakilk]

reseller -

I guess my pages could be classified as thin affiliate as they are part of a datafeed store I added to an existing community. Here is the interesting part... the pages that form the community i.e. forum pages, blog pages, are untouched in rankings. In fact traffic has increased somewhat since the start of the year. However the datafeed store pages have been strangled by the mighty hands of Google. Traffic to these pages has all but dried up.

It seems that as it currently stands Google will kill off any thin affiliate type pages even if they belong to a quality site.

[g1smd]

The wierd stuff is back: [72.14.207.99...] and [72.14.207.104...] at least.

[reseller]

frakilk

"reseller -

I guess my pages could be classified as thin affiliate as they are part of a datafeed store I added to an existing community. Here is the interesting part... the pages that form the community i.e. forum pages, blog pages, are untouched in rankings. In fact traffic has increased somewhat since the start of the year. However the datafeed store pages have been strangled by the mighty hands of Google. Traffic to these pages has all but dried up."

The issue here sounds more than affiliates, IMO. I guess possible duplicate contents might played part or was the main reason for being "strangled by the mighty hands of Google" :-)

I have ran a test on one of my test-domains for around six months. I'm sure now that duplicate contents is a killer, not the affiliate links, nor AdSense spots.

I might write more details about my test in a future post.

[reseller]

g1smd

"The wierd stuff is back: [72.14.207.99...] and [72.14.207.104...] at least."

I'm not so sure yet, but right now (9:53 pm GMT +1) when run my keyphrase and apply &filter=0 on my "friendly" following DCs, I end up at serps with the top 10 sites the same as those of the 2 DCs you mentioned when run the same keyphrase on (without &filter=0).

[64.233.171.99...]
[64.233.171.104...]

[64.233.185.99...]
[64.233.185.104...]

[72.14.203.99...]
[72.14.203.104...]

Thoughts?

[bobmark]

"It seems that as it currently stands Google will kill off any thin affiliate type pages even if they belong to a quality site. "

Reseller may be right, frakilk.

One of the top ranked (frequently #1) sites in an area I monotor is the definition of the "thin affiliate." ZERO original added content, page after page generated off the original supplier database.

Unless something changes, they sure don't seem to be suffering.

[g1smd]

I can very easily spot the wierd stuff, as for one result, instead of "1 to 15 of about 15" I now get "1 to 50 of about 45000" (and most of the 50 results are pure supplemental junk that has never been seen for this search at any time in the last four years).

Some difference, eh?

[bobmark]

I got a feeling we're gonna see a major "adjustment" next few days, g1smd.

Can't believe Google is staying with any of the current flavours of BD indexes until the next update (one contains tons of ancient pages, the other was pared down by dumping anything that had incorrectly been listed as supplemental).

I think Google's gonna try to merge the results of the heavy crawl of the past 2-3 weeks with the pared down index soon (course they have been batting about .050 on attempts to merge the circa AUG BD index with anything else so far).

[reseller]

g1smd

Could it be that they upload the "unhealthy data" as it is on [72.14.207.99...] and [72.14.207.104...] to be used for later testing?

[kamikaze Optimizer]

Wierd stuff everywhere now, totally weird.