homepage Welcome to WebmasterWorld Guest from 54.237.54.83
register, free tools, login, search, pro membership, help, library, announcements, recent posts, open posts,
Pubcon Platinum Sponsor 2014
Home / Forums Index / Google / Google SEO News and Discussion
Forum Library, Charter, Moderators: Robert Charlton & aakk9999 & brotherhood of lan & goodroi

Google SEO News and Discussion Forum

This 34 message thread spans 2 pages: 34 ( [1] 2 > >     
Google Data Centers - Peekaboo affect
extremegolfer




msg:3655887
 11:31 pm on May 21, 2008 (gmt 0)

For a new keyword we are targeting, our site comes top 5, but when I click search again, it is not in the first 999 results. Is that a good sign or a bad sign? If it is a good sign, when will things settle down so the rankings are consistent? If a bad sign, is there anything we can do?

 

tedster




msg:3656045
 5:20 am on May 22, 2008 (gmt 0)

I think you're describing the new Google strangeness that is being discussed in the May SERPs Changes [webmasterworld.com] thread. I don't know that anyone could predict the future of this kind of thing. My suspicion is that Google is trying an increased level of machine learning in their algo, and by rapidly changing some of the SERPs, they get a chance to measure the performance of different urls - a kind of tasting or "grazing" if you will.

Also note, quite a few people are reporting traffic getting cut by a significant percentage, but at least some times the rankings seem to be steady. I'd say that the rankings can't be steady if traffic is getting chopped. It's just that when they check the rankings all seems to be fine. By if they check at a different time, as you did, they also might find a very different scene.

There's another possibility in your case - the page is so new that it hasn't hit all the datacenters yet. That seems unlikely if you're already at #5 some of the time, but it is possible. If that's it, you should see stability in less than a week.

extremegolfer




msg:3656066
 6:26 am on May 22, 2008 (gmt 0)

Which data center do they update/test first? Is that something that's published?

Pico_Train




msg:3656083
 7:08 am on May 22, 2008 (gmt 0)

Good question there ExtremeGolfer, I don't think anyone but Google really knows that and it's probably random too!

pageoneresults




msg:3656096
 7:38 am on May 22, 2008 (gmt 0)

My suspicion is that Google is trying an increased level of machine learning in their algo, and by rapidly changing some of the SERPs, they get a chance to measure the performance of different urls - a kind of tasting or "grazing" if you will.

I've snuck this in to various topics and I truly believe Google is utilizing some sort of randomization algorithm (RANDAL). There's enough movement in those top 20 results for me to "guess" that may be what's happening. My research set is moving right now. Within a week, the two results I'm watching should move to page two which has been the normal routine now for the past few months. The first few positions appear to be "locked" while those below are in continual flux. I know, its probably one of the number topics floating about. I just prefer to classify it as a new randomization algo. Something I've been wanting to see for a long time. ;)

extremegolfer




msg:3656121
 8:25 am on May 22, 2008 (gmt 0)

pageone - to go from #5 to #7 is one thing, but to go from #5 to not in the first 999 results is not randomization.

pageoneresults




msg:3656126
 8:29 am on May 22, 2008 (gmt 0)

For a new keyword we are targeting, our site comes top 5, but when I click search again, it is not in the first 999 results.

What you are describing above is more of the natural progression of something "new". It will duck in and out of the index while it gets crunched by the various data centers. At some point it will settle down and you'll be able to get some rest. ;)

extremegolfer




msg:3656141
 8:47 am on May 22, 2008 (gmt 0)

makes sense... any idea of time frame? is it days, a week, several weeks?

whatleywah




msg:3656237
 11:15 am on May 22, 2008 (gmt 0)

I know I have been seeing this for about a month or more with at least 2 of my clients, both longish standing sites.

As for time frames, sometimes its there then next time you search its gone, other times it can be there for a whole day or two.

Looking at Stats though there is not an increase really in the traffic coming from the new keywords.

extremegolfer




msg:3656490
 4:11 pm on May 22, 2008 (gmt 0)

It seems either no one knows or know one wants to share the time-frame of the stabilization of data centers. I didn't think it would be such a secret...

jimbeetle




msg:3656502
 4:23 pm on May 22, 2008 (gmt 0)

For a new keyword we are targeting, our site comes top 5, but when I click search again, it is not in the first 999 results.

My suspicion is that Google is trying an increased level of machine learning in their algo, and by rapidly changing some of the SERPs, they get a chance to measure the performance of different urls - a kind of tasting or "grazing" if you will.

Putting both of these together how about, just as a shot in the dark, something along this line? User searches for [keyword] and instead of clicking on a result clicks "Search" instead. Google figures the re-search means that the user didn't find a good result the first time around so presents a different set. The behavior jibes with that of the click-tracking believers (myself among them), and also just seems to be something G would do.

Again, just a shot in the dark.

pageoneresults




msg:3656541
 5:14 pm on May 22, 2008 (gmt 0)

It seems either no one knows or know one wants to share the time-frame of the stabilization of data centers. I didn't think it would be such a secret.

No, its not a secret. Its just that there are all sorts of factors involved and I don't think anyone can give you an exact time frame. Best thing to do is be patient and let gnature do its thing. Some have launched new pages and seen them settle down in a week or two. Others have seen it take upwards of a month. And others have seen it take longer than a month. And I'm sure there are a few that are going to tell you that it took longer than three months.

Which of the above answers is correct for you? You'll probably want to choose the lesser of the bunch and look at the week or two. If I told you it would take a week or two, and you waited a week or two and it didn't happen, who ya' gonna' blame? ;)

annej




msg:3656832
 9:08 pm on May 22, 2008 (gmt 0)

I just prefer to classify it as a new randomization algo. Something I've been wanting to see for a long time.

What do you mean by randomization? What I'm seeing looks like a planned randomization. The pages ranking in and out seemed to be a select few.

Also can you explain why you have been wanting to see this happen. I'm just not able to see the whole picture.

pageoneresults




msg:3656853
 9:32 pm on May 22, 2008 (gmt 0)

What do you mean by randomization? What I'm seeing looks like a planned randomization. The pages ranking in and out seemed to be a select few. Also can you explain why you have been wanting to see this happen. I'm just not able to see the whole picture.

Okay, ya ready? I've have this Tin Hat Theory that Google is now using a Randomization Routine in the SERPs, I call it RANDAL although some might like to call it RANDALG because it hasn't been real kind to them in some instances, they've been dogged.

I think it would be in the Search Engines best interest to implement this type of algo in certain sectors. There are of course many exceptions to the rule and we know that the SEs have the computing power to figure it all out, Google just happens to be ahead of the pack in this area.

There are only 10 spots on that first page. For years, some of those spots have been dominated by a select few. That isn't fair in today's indices, they are far too large for any one, two or select few to dominate, the landscape has changed. No longer can you expect to obtain and secure a top position in an extremely competitive area. Not for any long periods of time, I think we're seeing the end of that era and Google is in the midst of "making it so".

If you were to take those top 50-100 results and mix them up throughout the day in a random fashion, that would allow more exposure to websites that may deserve some play in those top positions, there are billions of pages that fall into this category. I'm referring to mainly commercial oriented results. Those old historical informational resources may not be affected as much although there may be some "slight" shifting taking place to test the results returned.

I've mentioned a specific scenario I'm watching right now with a very specific keyword. Its a very competitive single keyword and there are two sites that I monitor for that. Both have been in a constant rotation now for at least 90 days which gives me more metal for me tin. The rotation does appear to be "planned". Meaning that the two sites are usually within one or two positions of one another as they move between pages 1 and 2 on a bi-weekly basis right now. Don't quote me on the numbers as they are not exact, they are estimated.

I'm sure those that have been sitting solidly below these two are welcoming the change as they are now being exposed to those pageoneresults and enjoying the many benefits that come along with that. I don't mind the movement as I like a "fair" playing ground. There are too many of us doing this SEO thing for any one of us to take those top spots for any long periods of time. No, that landscape has changed for many and now its more a matter of getting quality content out there so you can have more of your assets spread out to capture more of the long tail which grows each day. Those money terms that many have held positions for over the years are probably seeing some changes. Maybe nothing drastic but a drop in position or two. That's all it takes to start the ball rolling. That's how it all started with my particular result set. Long standing historical domains that held top positions for years are now seeing what I'm calling RANDAL.

I could be so far off base and become a WebmasterWorld raving lunatic but I don't think that is the case. I'd be willing to bet some C Notes on the tin. ;)

Some will say that what I'm seeing is the constant flux within the data centers. Okay, I don't have a problem with that. RANDAL is at work in all data centers. Not just one.

tedster




msg:3656869
 9:49 pm on May 22, 2008 (gmt 0)

I'm with you, p1r, except I wouldn't call it "random". I'd say the URLs involved have very specific characteristics that flag them for special treatment. I think it's more like advanced machine learning, and my guess is that it's done for a dual purpose:

1. Find under-rated sites that are helpful for Google's end users - if they really are, then Google's metrics will show it when they cycle up to a high ranking for a test period.

2. Promote suspected spam in an automated fashion, and then at some point Google's metrics will have enough user data to measure their true worthlessness.

The factors laid out in the four April 2007 patents [webmasterworld.com] could be the foundation.

In a recent Google blog post, VP of Search, Udi Manber [googleblog.blogspot.com] talks about the team of statisticians that Google now employs. and we all know of their emphasis on scalability. The current strange new world in the SERPs might well be the sign of a new machine-learning algo - another steps toward AI.

pageoneresults




msg:3656879
 10:00 pm on May 22, 2008 (gmt 0)

I like Machine Learning Algo or MALALGO. ;)

I think its gotten to the point where Google realizes just how severe the situation is with manipulation in the top positions. I'm sure they are responsible for some of it in an effort to counteract those forces at work that are constantly at battle with their algos.

I firmly believe that they are working diligently and successfully to upset the "traditional" SERPs that we've become accustomed to having some control over. We'll still have that control but it won't be constant. We'll get our time at the wheel at which time we'll need to make the best of it and then someone else gets to drive for a bit.

It really is absolute genius if this is what they are up to. And, I've read those patents you referenced and that we've discussed over the years, that was more ammo for me tin hat theory. I like that tin hat expression, it protects me from making a fool of myself if I'm too far in left field. "Hey, I had me tin hat on, whatdya expect?" :)

extremegolfer




msg:3657005
 1:39 am on May 23, 2008 (gmt 0)

Page1, we had discussed that in my office several years ago when I was worried my SEO guys could not keep me on the first page. We all concluded it would make a mess out of Google. Imagine calling your friend and saying search for XYZ, look at the #1 result. Yet the friend sees something totally different than the expected #1 result. What's the conclusion going to be among the 2 friends..."man google sucks"

annej




msg:3657024
 2:14 am on May 23, 2008 (gmt 0)

Google's metrics will show it when they cycle up to a high ranking for a test period.

How would this be tested? By click rate and bounce rate?

at some point Google's metrics will have enough user data to measure their true worthlessness.

That explains some of the items I see being rotated in and out of the top few positions.

I think I'm beginning to wrap my mind around all of this.

extremegolfer




msg:3657030
 2:30 am on May 23, 2008 (gmt 0)

Site worthiness can be measured several different ways. 1)Google Desktop 2)Google toolbar 3) Google Analytics 4) Links sent in Gmail 5) The javascript that is tracking Google search result clicks

However, all of the above does not apply to my current situation of #5 one minute to nowhere in the first 999 results in the next search minute. Then you have all the possible filters they can apply. If this then apply this penalty, if that then apply this other penalty. Then when that penalty is somehow lifted give one data center that has updated itself, then give the other data center that has not updated itself and on and on.

outland88




msg:3657110
 6:44 am on May 23, 2008 (gmt 0)

Imagine calling your friend and saying search for XYZ, look at the #1 result. Yet the friend sees something totally different than the expected #1 result. What's the conclusion going to be among the 2 friends..."man google sucks"

Advertising research with magazines showed years ago that randomizing the placement of ads to give everybody an equal shot at a top page basically hurt the sales for all. In other words you were better off having a stable position somewhere than having top position that varied. How many times have you gone to show somebody or buy something based upon where you remembered it was in a magazine or newspaper?

Thatís why I donít believe the results are random or willy-nilly. Theyíre time sensitive and traffic oriented and youíre likely involved in either Adwords or Adsense and I can be wrong. If you watch closely itís basically the same principle used in both those programs. You'll both settle and periodically bounce IMO. Google didnít buy all those companies to know ďnuthinĒ.

soxos




msg:3657228
 10:46 am on May 23, 2008 (gmt 0)

Golfer, is this a new page & new keyword? If it is, than rather than think it is a penalty, think of it as new page bonus - where the page is on probation. It is very hard to get such new pages to stick in their highest position. Usually they land somewhere between the two and work up gradually over say 6 months.

pageoneresults




msg:3657230
 10:56 am on May 23, 2008 (gmt 0)

Imagine calling your friend and saying search for XYZ, look at the #1 result. Yet the friend sees something totally different than the expected #1 result.

That's been happening for quite some time now. The result sets are not the same for everyone. With all the various targeting and personalization taking place, there is a good chance that your friend is not going to see XYZ in the same position you are seeing them.

Imagine calling your friend and saying search for XYZ, look at the #1 result.

I don't think you'll be able to do that for much longer. You might be able to say something like "take a look at those first ten, do you see them?" No? Okay, go to Advanced Search, select the 100 Results Per Page option, now do you see them? ;)

extremegolfer




msg:3657602
 5:21 pm on May 23, 2008 (gmt 0)

Golfer, is this a new page & new keyword?

No, same old homepage. New keyword.

extremegolfer




msg:3657673
 6:54 pm on May 23, 2008 (gmt 0)

The serps I'm watching have the same first 10 results. Except once in a while for the target keyword my site sneaks in there in spot #5 and bumps #10 to #11. However, I hit search again and my site is not in the first 999 results for the target keyword.

Ive been seeing that fairly consistently, except for today. Since 8A my time, my site has not come up at all for the target keyword. Any comments as to possible reasons and/or data center unsolved mysteries?

annej




msg:3657685
 7:22 pm on May 23, 2008 (gmt 0)

If you have the personalized search system that Google does that may cause you to get new serps on your second search when you get the 999 result.

Your sites popping in and out of the the 5 position is happening to a lot of us so that seems to be normal they way things are going lately.

jimbeetle




msg:3657733
 8:06 pm on May 23, 2008 (gmt 0)

The only thing I don't like about the randomization or machine learning theories is that they don't address the question, "How does this help the user at the time the search is performed?" That's Google's stated number one concern.

I think its gotten to the point where Google realizes just how severe the situation is with manipulation in the top positions.

But would they go to randomization of any sort to fight it? We've already been there, done that. Wasn't randomization of SERPs in order to thwart SEOs one of the nails in Alta Vista's coffin?

reseller




msg:3657751
 8:31 pm on May 23, 2008 (gmt 0)

extremegolfer,

Try to test your new keyword you are targeting on these two DCs:

[72.14.207.104...]

[64.233.167.147...]

extremegolfer




msg:3657758
 8:36 pm on May 23, 2008 (gmt 0)

Reseller,

Although [72.14.207.104...] had a higher number of results, I was still not in the first 999 results for either one of them. Are these 2 data centers the most up-to-date?

reseller




msg:3657781
 8:53 pm on May 23, 2008 (gmt 0)

extremegolfer


Are these 2 data centers the most up-to-date?

Not sure. But I can see them at the moment displaying different serps for the sectors I watch.

I assume you don't see your site at the moment at top 5 at any DC, right?

CainIV




msg:3657805
 9:08 pm on May 23, 2008 (gmt 0)

This thread is all about the reason I love WebmasterWorld and the community. It's also the reason why when we surf the tech and news sites, they are quoting and referencing the site so much.

AI, or the move towards it from a calculation standpoint makes perfect sense. We can see this trend and follow it though the Google time line throughout the years.

Right now in the SERP's, for much of the genres I watch, I see a mix of unstable and unpredictable websites popping in, and full complex and robust websites.

Strangely enough, it seems to be the middle of the road websites - one that did not fulfill all of the quotas - that took a hit.

This 34 message thread spans 2 pages: 34 ( [1] 2 > >
Global Options:
 top home search open messages active posts  
 

Home / Forums Index / Google / Google SEO News and Discussion
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Home ¦ Free Tools ¦ Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About ¦ Library ¦ Newsletter
WebmasterWorld is a Developer Shed Community owned by Jim Boykin.
© Webmaster World 1996-2014 all rights reserved