| 11:06 pm on Apr 6, 2010 (gmt 0)|
I am seeing something on the Caffeine datacenter that I have not really seen too often.
One way in which we have identified when were were looking at the Caffeine dataset on the Caffeine datacenter or on Google.com is by looking at the number of competitive sites that show up for a search.
Usually, the Caffeine dataset has a smaller number than the regular dataset.
Now, I am seeing more competitive websites than I am used to seeing for searches that I do on Google.com and also on the Caffeine datacenter but now the Caffeine dataset seems to have a lot more competitive sites for a search (sometimes 10 million more) than when I do the same search on Google.com.
Could this be because Caffeine is adding more datasets to its index and when Caffeine is rolled out the number of competitive sites that show up for a search will be more than what we now see on Google.com when it is showing the regular dataset?
| 11:26 pm on Apr 6, 2010 (gmt 0)|
Could it be that this the last inning if the rollout, and google is finally rolling in more and more data?
| 1:47 am on Apr 7, 2010 (gmt 0)|
Actually yes I do. Why it prevents misinformation from getting into someones head. I wanted to make it clear and let them know there will be an aglo change. As I said I already know of 1 Site speed. This is very important to the discussion because as I said when sites tank/rise we can learn from them and apply.
|Fair enough... Do you think our discussion helps anyone else out a bit? |
I know without a doubt when Caffeine is complete we will see a ton of post seeking information, then we discover from this, or at least I try to discover others may have a different agenda.
| 1:55 am on Apr 7, 2010 (gmt 0)|
|I wanted to make it clear and let them know there will be an aglo change. |
And I wanted to make sure people realized it may not be the algo change that cost them their rankings, but as you point out, it could definitely be... I think maybe we've both given people something more to think about, and I try to learn more about what goes in to rankings constantly, but what I keep coming up with is as time goes by rankings are going to be tougher to achieve, so IMO depending on search engines solely for traffic could be a 'Florida Failure Plan' which is one of the biggest things I got out of those old threads that has stuck with me through the past 5+ years of this game...
Anyway, good point on the fact the algo will probably change and be the cause of at least some issues for people if they're not 'solid' in what they are doing and 'ahead of the game', and the speed of the calculations will probably also increase (Caffeine), so as the roll out is completed not only will 'discovery of issues' related to the current algo be applied faster to more sites (pages), so will any algo changes.
| 4:37 am on Apr 7, 2010 (gmt 0)|
wow not see this before, a new url structure for Cached pages,
that must of had something to do with the URL only entries are people seeing them come back?
| 4:48 am on Apr 7, 2010 (gmt 0)|
That was first spotted here two weeks ago by MLHmptn - [webmasterworld.com...]
When I think about it, that could well be part of the Caffeine infrastructure. Sending the cached link through a hostname instead of the bare naked IP address could help with load balancing and giving faster responses to cache: queries.
| 4:56 am on Apr 7, 2010 (gmt 0)|
now that I'm looking a little more closer i also see the delivery times of the pages being quoted in Hertz instead of Seconds as of old.
so I'm assuming that I've finally stumbled across the Caffeine infrastructure on live.
| 5:18 am on Apr 7, 2010 (gmt 0)|
Google already uses google translate with the googleusercontent URL [translate.googleusercontent.com...]
| 7:19 am on Apr 7, 2010 (gmt 0)|
I was wondering if you can help me figure out where my sites are ranking on what set of results (i.e caffeine, non caffeine) as everything seems to be all over the place this morning
is it possible you can post the i.p addresses of the google uk results which you believe are caffinated?
many thanks :)
| 7:21 am on Apr 7, 2010 (gmt 0)|
I too have a question: for those that are following this, what % of SERPS do you believe is caffeinated so far?
| 8:14 am on Apr 7, 2010 (gmt 0)|
i'm getting caffeine fed results in the UK from this IP 184.108.40.206 or certainly from this range on a large % of the time.
I think this fits with what MC said that he would expect a european IP to start seeing caffeine on and off next.
I still think a full roll out, might be some time off, they seem to be taking this very slowly.
All those in the US, you've been seeing caffeine on and off for a while now haven't you?
| 8:24 am on Apr 7, 2010 (gmt 0)|
"Could this be because Caffeine is adding more datasets to its index and when Caffeine is rolled out the number of competitive sites that show up for a search will be more than what we now see on Google.com when it is showing the regular dataset?"
I think you summed it all in your question, Gouri.
| 8:29 am on Apr 7, 2010 (gmt 0)|
should I be adding a en&gl=uk&sa=2
to the search string to get something a little closer to how the UK specific results would be displayed as?
If so, this is what I've been tracking for a while :)
| 8:31 am on Apr 7, 2010 (gmt 0)|
MM - just add &gl=uk to the end
| 8:48 am on Apr 7, 2010 (gmt 0)|
thanks leeds1 for confirming
this is what I've been doing since december on here, there appears to be more variant between the current serps and those on the caffeine set than ever before which I find a bit confusing, up until this week I felt (on the majority I've been tracking) that they were almost identical?!
| 10:01 am on Apr 7, 2010 (gmt 0)|
adding &sa=2 in the end donates that you have clicked the search button in footer of google instead the button with top search box
And yes I can confirm that on [220.127.116.11...] caffeine is appearing and disappearing with each click on search button.
Also California IP's has the most odd results. A Major shuffle not just one two results replacing each other.
And yesterday ..it was again a major role our then reverted similarly like 1st April roll out.
| 12:07 pm on Apr 7, 2010 (gmt 0)|
Do you guys think all the datasets that are going to be added to Caffeine have now been added to it or are we going to see further additions made?
| 1:04 pm on Apr 7, 2010 (gmt 0)|
I think what we are seeing in the Caffeine datacenter now are the Caffeine dataset that we have been seeing for the past 4 or 5 months alternating with a new dataset over the past couple of days showing a larger number of competitive sites for a keyword search.
| 1:38 pm on Apr 7, 2010 (gmt 0)|
The SERPs we are seeing right now is a hybrid rollback, some data is about a month old, some data is new, so, we are not done yet. Clearly, they have not rolled in all the data they have.
| 1:52 pm on Apr 7, 2010 (gmt 0)|
Are you referring to Google.com or to the Google Caffeine datacenter?
| 2:05 pm on Apr 7, 2010 (gmt 0)|
Here are some INFO from an 8 year old Ecom Site.
Just checked the SERP, 15 sets of main Keywords that the site should rank for: we're still in the top 3 for all of them, which changes monthly, but still in the top 3, always.
The MAIN Keyword combination is 4 positions higher that a week ago on page 1.
The traffic from US and CANADA is 32% below normal starting exactly April 2nd.
| 2:18 pm on Apr 7, 2010 (gmt 0)|
TheMadScientist I think our posting did much to help improve discovery in this new Caffeine era we are going into. Even though our opinions were different we did it in a way that improved the overall discussion IMO.
I am seeing more and more of Caffeine so it seems Google is real close to the full rollout.
| 2:50 pm on Apr 7, 2010 (gmt 0)|
OK, I'll bite, what was learned from all the hours spent watching IP addresses, everflux & the rollout of caffeine? I have been watching this thread for many years, and I am still wondering what is learned from watching G in the middle of a change. I just watch the thread, I do not spend any time watching the serps on different IP addresses. Not with proxies, not at all, as I cannot see the value of my time being spent watching, a process in motion, expecially with the load balancing going on, so you never know what you are looking at, or where it came from.
| 2:59 pm on Apr 7, 2010 (gmt 0)|
The best place to test right now is indeed 18.104.22.168...
I oscillate between 2 SERPs with repeated searches. One is old (2 month-ish), one is newer, but the new one is a "hybrid", a mix of old and new... The new one is not quite up to date, and is not as advanced as what I saw April 1/2. I would say the new one is about 2 weeks old, but some portions are up to date. Compared to April 1/2 SERPs I saw, which I would say were about 70% up to date, I would call the new SERPs at 22.214.171.124 about 40-50% updated.
-shrugging- those are my observations, in my niche, for the varied SERPs I watch. ; )
| 3:00 pm on Apr 7, 2010 (gmt 0)|
Here's a new observation for me about an internationally customized SERPs. This was a surprise:
I'm talking about google.de results here - as seen from Germany and from the US east coast. The site has an English language page that came up for inspection. From Germany it ranks on page 2 and in the US it ranks page 1.
The keyword phrase is the same in both German and English. From the US, all the pages that outrank it are in German so I assumed that in Germany even more preference would be given to German language and German hosting.
Not so! From Germany SERP page 1 is all English results, and from the US it is mostly German! Very strange "localization" there.
| 3:06 pm on Apr 7, 2010 (gmt 0)|
Thanks for that posting on what you have been seeing on the Caffeine datacenter recently.
It is very informative.
| 3:11 pm on Apr 7, 2010 (gmt 0)|
|OK, I'll bite, what was learned from all the hours spent watching IP addresses, everflux & the rollout of caffeine? |
You mean apart from the sheer visceral thrill?
In a word, screenshots. Data for later analysis as you compare different routes to a converging result. A few weeks down the line, you might be able to work out what penalties are added in where, which sites got temporarily abandoned or promoted (interesting if they go/ come back after the "rollout" finishes) and any other tidbits.
I have to say I'm not an avid DC-watcher, even in the midsts of turmoil. As steveb observes, its never the whole story there. Best to watch the public version- at least then you know what you're watching.
| 3:13 pm on Apr 7, 2010 (gmt 0)|
WW_watcher - one of the main things I have learned, and indeed, I am in a learning stage here, I am a "newbie" in this regard, one of the main things that always strikes me is: who moves, and who doesn't. Wow. Fascinating. Clearly, my goal is to become one of the sites that never move. Some sites move in from nowhere, then dissappear, and some move around a few spots. It is like there are 3 classes of sites: static, semi-static and the wild cards. I do not want to be a wild card. I want to be a static. So, when the SERPs jump around like this, it gives you a really good way to identify who falls into which group, then you can study them. What are they doing, what aren't they doing. Because, wildcards can jump into the top 3, then fall off the map, so just because a site is in top 3 doesn't mean gorg loves it. It's like driving a plane, once you reach altitude, and are coasting, nothing much ever happens - the real #*$! happens when you land and when you take off - those are the moments of chaos. So, same here, once the SERPs settle down, it is hard to notice much, but when the SERPs are changing, you can see a lot. The SERPs I watch, some I automate, and some I watch manually, and the ones I watch manually I have been watching for years, so, you become quite aware of who is who and where they sit. So, when you watch SERPs in transit, you make a mental note of all that, and later on, it becomes interesting to see who is in motion. I come to really "know" those manually SERPs I watch. It is the good, the bad and the ugly. I do not want to be a site in motion. So, I have a lot to learn from both the sites that are in motin (sorry!), and those that aren't. I know what you're saying, watching gorg algo changes or infrastructure change can be like watching paint dry, and who cares, it isn't settled anyway, so we're wasting time, but... when you consider that the intention is to see HOW things move and WHO moves an then, the magical WHY, you can learn a lot from that. I post here that "I see the SERPs in transit!", "they are old", "they are new!", but that is pedestrian stuff, becuase that is the public stuff I can share in passing - I can't really post details of my real observations on individual movements and why because they will only relate to me... Hope that makes sense. This thread mostly reports the surface of things, but I think what some of us are doing is analyzing the details of WHY we're seeing what we see, at least, we should be, and trying to learn and adapt from that. ; )
| 3:29 pm on Apr 7, 2010 (gmt 0)|
|I have been watching this thread for many years, and I am still wondering what is learned from watching G in the middle of a change. |
One of the easiest things to learn is what they're not exactly looking for and what the final results will not be, which can be an interesting study in and of itself.
One of the funniest things I used to be able to do back when McDar was a site and you could 'watch the update happen' was project a 3 to 7 days ahead of time when the update thread was going to end... I could usually be fairly accurate about the 'level of ranting' and the number 'G's broken posts' on a given day by looking there before I looked at the thread here too. :)
| 4:18 pm on Apr 7, 2010 (gmt 0)|
I find it very difficult to tell in which state an update or folding is as well. I've seen 6 sites of 6 completly different areas/keywords popup from around 14-16 to 4-6 and going back againafter a day - but every site on different days in the last 2 weeks. What does this mean? Which filter has been applied to these sites when you couldn't not correlate the events because every keyword seems to has it's own timing?
| 5:12 pm on Apr 7, 2010 (gmt 0)|
|What does this mean? Which filter has been applied to these sites when you couldn't not correlate the events because every keyword seems to has it's own timing? |
Screenshots again. Then, when it settles down, compare to the mid-update, and the pre-update SERPS. Find whats changed- see if the bounderies of helpnow's classes have shifted to include/exclude new groups.
Once you do this for a couple of shifts you will see patterns mid-shift. Then, you can have "hunch-test" pages in situ for the anticipated system, allowing you to get a jump on the competition.