15 replies to this topic

[mirajewel]

Hello all,

I have had a website for about 6 six years. I never really optimized the pages but somehow managed to have a high ranking on google (top 1-5) for several of my keywords until the major google update in late 2003. My keywords then dropped down to around 50 with the google changes to their algorithm.

My site had always used a static homepage and cgi-generated interior pages. There are about 250 cgi pages and they are all indexed by google.

Here’s my problem: About a month ago, I decided to change my static index page to a cgi generated index page because of problems with the shopping cart. My google rankings dropped from 50 to about 450-500.

I then started reading this forum and made some changes that have brought the rankings up slightly (400 range). There are no html errors, page rank of 5, 35 google counted back links (about 500 total), dmoz listed, etc.

With the “-asdf -asdf -asdf -asdf -asdf -asdf -asdf” filter my rank goes to around 20 from 300+.

Questions:

Could changing from a static index page to a cgi generated index page have put me in the “sandbox” even thought the site has been up for six years and retailing the same product?

What does the ranking with the “-asdf” filter mean, if anything?

My index is spidered by google on a daily basis but nothing I do seems to have much of an effect on ranking– and I’ve read about the "sandbox" for new sites, but didn’t know if it can happen to older sites?

[Shane]

I believe the -asdf filter prevents Google from integrating semantics into the results it presents you.

For instance, if your site is about "cars" and ranked alright for that term previously, it may have been displaced when Google introduced semantic matching by sites that ranked well for "automobiles," "vehicles," etc. (i.e. synonyms for "cars").

Adding the -asdf causes Google to skip the semantic checking, since it assumes that it would add too much to the processing time of the query.

[amabaie]

Whoa! There is a long way from stemming (car versus cars, jump versus jumped verus jumping) and semantics.

Google introduced stemming sometime last fall with not so much as a peep that they were doing it. This is the first peep I hear about semantics. Where did you get that from?

[bwelford]

Actually, amabaie, they didn't introduce stemming last November. They introduced semantics at that time. If you looked carefully you would find on the Google website a somewhat subtle change in what they said about stemming. I explained what was involved in a Newsletter I wrote at that time.

[Will.Spencer]

It is most important to remember that no one outside of Google knows what the -asdf -asdf -asdf (etc...) filter does.

No speculation has been backed up with any degree of reasonable proof.

[Jill]

Welcome Will!

Jill

[OldWelshGuy]

This asdf business is the same as kkhhlkljl ;;;;;k etc. all you are doing is breaking the search out of the filters. IMO it is because longer search phrases tend to be more sepcific, so there is no need to apply a filter, added to the fact that the processing needed to filter a 10 word search is way too much, so why bother?

If this form of searching removes the sandbox effect, why have some sites reported worse resuts when using it?

OWG

[bwelford]

If this form of searching removes the sandbox effect, why have some sites reported worse resuts when using it?

Duh, ... well if some go up in the listings, then of course some get pushed down.

It's the same kind of thing as the fact that half the kids in school get below average marks in Maths. I don't believe that applies in Philosophy as far as I remember ...

[OldWelshGuy]

BW your right, but surely that would not result in a drop of like 100s of places? I mean there can not be 100s or more sites being sandboxed for the same search term, can there?

[amabaie]

Actually, amabaie, they didn't introduce stemming last November. They introduced semantics at that time. If you looked carefully you would find on the Google website a somewhat subtle change in what they said about stemming. I explained what was involved in a Newsletter I wrote at that time.

Barry, I respect your analysis, but I have yet to see any sign of somebody searching for "happiness" and results with "contentment" or "peace-of-mind" coming up...or any other similar search.

What I read on the Google site still refers very specifically to "word variations":

Word Variations (Stemming)
Google now uses stemming technology. Thus, when appropriate, it will search not only for your search terms, but also for words that are similar to some or all of those terms. If you search for "pet lemur dietary needs", Google will also search for "pet lemur diet needs", and other related variations of your terms. Any variants of your terms that were searched for will be highlighted in the snippet of text accompanying each result.

What can be said, by the example they offer, is that they are using more advanced stemming than Overture's single/plural approach, but that is still a long way off from becoming an instant thesaurus.

[bwelford]

OWG, I don't think the 'sandbox' effect means that web pages are not still in the database. We're talking here about filters or more probably different weighting factors applied to the over 100 factors in their algorithm. So I'm not surprised to see very different orders if they are applying different weighting approaches in the standard search versus the search including "-asdf".

What boggles my mind more is the sometimes relative constancy as Google tweaks its algorithms on a possibly daily or weekly basis.

amabaie, I am staggered at how they do semantic analysis anyway. I agree with you that it doesn't seem likely that they can get all terms that might appear in a thesaurus to appear in the same search. Of course, they're inferring semantic closeness from how words are used in web pages that link so this may not be as complete linking as you would see in a thesaurus.

[castlebooks]

there is a paper I read somewhere to the effect that adding kjjkkjkjjk to a seach term causes the 'modified hilltop algorithm' to fail and therefore display the 'old' results or something similar.

or, there are no expert or authority sites for the topic keyword keyword jkdfkjfdkj

This refutes the 'money words filter' theory and makes the most sense to me. Nobody knows what is ACTUALLY going on at Google so we are all guessing and the best we can do is an approximation -- so it is possible, even likely that 2 or 3 different approximations, while seemingly contradictory are all correct.

[BrianR]

Anyway, getting back to mirajewel...

First, welcome to the forum, mirajewel - glad to have you among us. And that's a great question for a first post.

I am by no means an expert in this area, but it is entirely possible that the huge drop in rankings you experienced had more to do with Google's constant tweaking of their algorithm rather than your changes to your index page.

Your best bet is to stick around and post some more, contributing where you can, and when you've notched up 20 posts, put your site up for a site review - you'll get lots of specific, helpful advice.

Meanwhile, there's tons of advice on this topic in some of the other, more specific categories in the forum.

BrianR

[hoboseo]

Hi,

I was studying this "-akdjd" effect for spanish terms last months. I think when you use "-kjkad" google don't use semantics in the results it presents you (only speculation, of course).

but... two questions:
What do you think about the "-adadd" effect in Yahoo?
and why do you think some search terms have differents results for capital letter and lower-case letter (in Yahoo! too)?

[john alphaone]

Barry, I respect your analysis, but I have yet to see any sign of somebody searching for "happiness" and results with "contentment" or "peace-of-mind" coming up...or any other similar search.

You can see the effect of the semantics if you do a tilde search. It's quite limited - nothing seems to match 'happiness'. However as an example if you search for '~barn' you will see that the word 'farm' is matched in bold text. That's clearly going farther than stemming.