| 2:25 am on Sep 10, 2009 (gmt 0)|
I'm seeing a more updated version of the current index on Caffeine now, looks like they have included a few more of the live filters.
| 12:35 pm on Sep 10, 2009 (gmt 0)|
Anyone else seeing de-capitalisation of page titles on the caffeine index.
some page titles are having the domain brand name altered and placed in lowercase, on some pages but not across the whole website, looks to be a filter of some sort but its unclear why some pages are getting treated differently than others.
No change in this websites SERP's on either index but looks like there is a cross over between "live" and caffeine for this filter as i saw it first on Caffeine and now its appeared on to the DC i'm viewing.
| 12:48 pm on Sep 10, 2009 (gmt 0)|
No decap here in a UK niche.
| 12:48 pm on Sep 10, 2009 (gmt 0)|
I am seeing a change in the caffeine serps as of this morning. They appear quite different then yesterday in my sector. Anyone else noticing this?
| 10:52 pm on Sep 12, 2009 (gmt 0)|
I've had this as my usual Google frontend for a few days now. Guess I've been picked out for a live user test somehow.
Is it better? Dunno. I'm not sure if I'm satisfied with the SERPS or not, as my searches those days have been very "niche", so I didn't expect a lot of relevant results anyway.
| 12:58 am on Sep 13, 2009 (gmt 0)|
|I've had this as my usual Google frontend for a few days now. Guess I've been picked out for a live user test somehow. |
Wow, that's the first I've heard of that. Just so I understand you correctly, are you saying that when you go to google.com and do a search - those results are always the same you see when you do the search on www2.sandbox.google.com/
| 1:56 am on Sep 13, 2009 (gmt 0)|
Just did a test search: I got slightly different SERPS, but not more different than if you sometimes hit another server (eg. when using a different geolocation IP).
This time the www.google.com gave me six more images and one more video than "sandbox". The top 3 listings are identical, then 2 videos on "sandbox" compared to 6 images+three videos on "google.com".
Between the two, only 3 listings are changed in the top 10 - one change is due to images/videos replacing a text listing.
The "Show options" tool to the left stays the same, and the drop down search box with the large text is also the same.
This minor difference is probably related to hitting another datacenter. Search quality is very high, both for text, images and video, meaning that results are not only relevant, but spot on.
Edited for clarity. First version wasn't specific enough
| 3:02 am on Sep 13, 2009 (gmt 0)|
< moved from another location >
Okay now, i have been reading Matts blog and he clearly mentioned that there would not be a major shift in serps when the caffeine update is rolled out.
|Most of the search result rankings should be the same, because whatís changing is infrastructure under the hood of the search engine. But Caffeine is a radical revamp (more like a complete rewrite) of large parts of our system, so not all the rankings will be unchanged. Thatís part of why we wanted to ask for feedback before things are fully deployed. |
However, one of my site which was slapped on 4th June 2008 is ranking very good in caffeine update and seems like all my hardwork to fix it shall be rewarded with the update, but my question is since Matt said that there would not be a big shift in Positions, i wonder if my site shall sustain the positions which are being shown in cafeeine?
Note: My site was not fully slapped and none of my page got deindexed or Was deduced in PR. My home page stood number one for its keyword even after the slap, it was just the internal pages which just stopped ranking.
[edited by: Robert_Charlton at 3:34 am (utc) on Sep. 13, 2009]
| 5:24 am on Sep 13, 2009 (gmt 0)|
I'm seeing much more recent cache dates in both live results and also the sandbox results. Most of the sites in my field are showing Sept 2009 cache dates now. For a month they were only showing the index pages as up to date and lower pages were approximately 2 months old cache dates.
| 5:45 am on Sep 13, 2009 (gmt 0)|
|since Matt said that there would not be a big shift in Positions, i wonder if my site shall sustain the positions which are being shown in cafeeine? |
My understanding was that Matt was writing about the "big picture". It is inevitable that some urls will see a different treatment - and if you are still seeing improved rankings on caffeine at this time, I'd think you have good reason to hope -- along with others who are in the same situation.
No one knows for sure what's going on, either with penalties that are not applied or, in some cases, penalties that seem to show up only in Caffeine. We also don't know how long until the new infrastructure becomes the only infrastructure.
| 8:40 am on Sep 13, 2009 (gmt 0)|
|when the caffeine update is rolled out. |
Where did Matt call it an "update". It's not an f-ing update it is a fundamental infrastructure change under the hood.
In my niche there's a site home page that I saw drop from SERPS a week ago. That page is still in the Caffeine sandbox.
Whether they have installed the Caffeine infrastructure or not at this time I think that what is in the sandbox does not include all of the hand spam penalties and that they will be added back at some time when Caffeine does go live. If the hard work you have done would get you back in the pre-caffeine index then there is nothing to suggest that it will not get you back into the new system. Just don't expect this change to do it for you.
| 1:54 pm on Sep 13, 2009 (gmt 0)|
tedster: I hope it treats me well :D
Hissingsid: Should not your nick be DissingSid? Anyways, i think your argument is valid. Interesting thing is, i don't see any other slapped sites for the same keyword, just mine!
I am not keeping my hopes high tho.
| 5:17 pm on Sep 13, 2009 (gmt 0)|
I'd say Hissingsid is correct - it's important NOT to call Caffeine an "update" because that may confuse many people.
| 7:26 pm on Sep 13, 2009 (gmt 0)|
|Hissingsid: Should not your nick be DissingSid? |
Nah! I'm just a miserable and cynical old git.
I've noticed an upturn, on this forum, of people who state things with such certainty you would think they have hacked into the Googleplex intranet and others who just don't seem to have got it. In both cases I've tried to do my bit to point out what I hope is a helpful alternative without upsetting anyone.
Since none of us really know what is going on it is best IMO to try and debate what might be going on and interpret what little information we have in such a way as to help each other to prepare for what might happen.
meanwhile tedster will keep us all on the right track.
| 9:40 pm on Sep 13, 2009 (gmt 0)|
I'll explain this one LAST time.
(My ad nauseum rant from Day #1 on this forum was and IS,
Stop listening to ANYONE. Build your own databases of 100s of keywords, track those keywords for 3 months. Yes, you'll get no sleep over that period but you'll UNDERSTAND Google better than any Goog employee does)
The Caffeine INFRASTRUCTURE is already live.
How do I know?!
Cause I've tested it.
An INFRASTRUCTURE change does NOT affect SERPS per se.
To OVER simplify what the INFRASTRUCTURE does - how quickly pages get crawled, updated, and ranked.
The Caffeine ALGO UPDATE is what people are seeing on sandbox.goog.
I really don't care WHAT a person CALLS it.
I've called it the "long national nightmare update" as no one seemed to want to "officially" give it a name as THIS board has had the privilege of doing in past updates.
Calling it the "Caffeine UPDATE" is as good a name as any.
But it is OBVIOUSLY a different ALGO is on sandbox.goog than goog.com.
THAT's why the SERP results are slightly different.
An UPDATE doesn't have to be a huge cataclysmic shift.
Sometimes it is, sometimes it isn't.
But it's obvious TO ME, the RANKING ALGO on sandbox.goog. IS different than what's currently running on goog.com
If one wants or needs Goog Employee #1, #2, or #3 to "officially" tell one this, then that person is the first person I would tell to STOP visiting SEO boards for 3 months, get your databases together, and then one will see how FUD-dy the information one receives from Goog PR is.
As always, I originally WARNED about what's was going on from MY OPINION. It's up to you to TEST whether I or anyone else is closer to the truth or not.
Waiting for Goog to say "this or that" is a recipe for nights of restless sleep, -950 penalties, yo-yo issues, and slowly losing rankings over time.
So again, what's showing on sandbox.goog. is a different ALGO than what's on Goog.com.
That's an UPDATE.
and has little to nothing to do with Caffeine the INFRASTRUCTURE change(that's already been implemented)
Take it or leave it, but one would be wise to TEST my theory as it's going to affect people greatly if they are "searching" for what changed AFTER the fact instead of NOW, when one can more easily see it.
[edited by: whitenight at 9:47 pm (utc) on Sep. 13, 2009]
| 9:45 pm on Sep 13, 2009 (gmt 0)|
What signs do you look for to know whether the new infrastructure or the legacy infrastructure was used to generate the results for your query?
| 9:57 pm on Sep 13, 2009 (gmt 0)|
|What signs do you look for to know whether the new infrastructure or the legacy infrastructure was used to generate the results for your query? |
As I wrote to you to help your clients site, (seems like years ago now)
When the last update "ended" I noticed that Goog was picking up onpage changes INSTANTLY.
In the past, changes would appear on the 'lesser' DCs first and then propagate to the main Goog.com.
(and it would take up to 72 hours or more)
Now, it was occurring in reverse.
Changes were shown on Goog.com first and then propagating to the 'lesser' DCs.
(All within 1-24 hours)
note - these changes were happening on sites that RARELY change content, not blogs or other frequently updating sites
Obviously it's more than that, but that's as good an example as any.
I often compare what happens when one gets their databases and studies it non-stop for a few months to the scene in Matrix,
where they are looking at the green Matrix codes fly by and they instantly "know":
"That's a red-head crossing the street"
"That's a businessman shopping at a store"
To explain in detail, would take many many pages and of course, one has to be willing to "test" stuff that may or may not even be related to one's own sites.
I test just for the sake of testing, so that I DO know what's changed in the algo.(since Goog rarely changes the "infrastructure", this change is even more apparent, to me)
Eventually, its going to affect one of my sites, somewhere.
| 11:12 pm on Sep 13, 2009 (gmt 0)|
Thanks for that, whitenight.
And just to keep the record clear, that particular discussion was not about one of my clients but rather an early example of the yo-yo that someone else pointed out to me. It was a fascinating study and I got quite interested in observing it, but I have not had a client caught in a yo-yo.
|Now, it was occurring in reverse. |
Changes were shown on Goog.com first and then propagating to the 'lesser' DCs.
(All within 1-24 hours)
That is quite a difference, isn't it? I've been noticing that the IP addresses serving results on google.com (detected with the ShowIP Firefox add-on) do not always show the same results as when that IP is queried directly. I haven't yet checked for how rapidly newly spidered versions of pages show up, however.
| 11:31 pm on Sep 13, 2009 (gmt 0)|
"Changes were shown on Goog.com first and then propagating to the 'lesser' DCs.
(All within 1-24 hours) "
Are you suggesting that this a phenomenon we should b seeing, after testing various DC's, because it is not one I am seeing (at the moment)
| 11:57 pm on Sep 13, 2009 (gmt 0)|
Don't know what you're testing or expecting to see, but from how you phrased the question, I don't think we're talking about the same thing.
| 6:33 am on Sep 14, 2009 (gmt 0)|
wow any one seeing comments being added to the SERP's, under some results i'm seeing.
Comment by test, Dec 16, 2008 - "good"
competitive terms and not so competitive .
All on live Dc's rather than sandbox.
| 6:34 am on Sep 14, 2009 (gmt 0)|
hehe scrub that i was logged in to a very old test account
false alarm :)
| 8:35 am on Sep 14, 2009 (gmt 0)|
whitenight, OK, sometimes it needs a rant to clear the air and get to the centre of the issue. I find what you are saying very appealing and admit that it does explain a lot of what I have been seeing. I also understand why Google might tell us that they have not gone live and point us to a test "sandbox" for us to comment on.
So to summarise my understanding/interpretation/extrapolation of what you have said:
1. The new infrastructure is live and that is what we are seeing on google.com.
2. www2.sandbox is also on the new infrastructure but has test changes to the algo on it. Or certainly has a somewhat different algorithm on it. I'm not convinced that this is a vision of the future though now that the penny has dropped (in my tiny brain) re 1. above.
3. Changes to pages, I assume on heavily spidered sites, are seen in the index within 1 to 24 hours.
If we accept that the infrastructure is live then we can move the discussion on to algo changes on the new infrastructure and how we can use the instant response to tweak pages and perhaps identify dangers in doing this. Will tinkering cause a penalty?
| 8:45 am on Sep 14, 2009 (gmt 0)|
|how we can use the instant response to tweak pages and perhaps identify dangers in doing this. Will tinkering cause a penalty? |
:) Now you're asking the right questions.
No, you should be able to test on-page stuff far quicker and determine whether the rankings go up, down, or no effect sooner.
Don't like what you see, change it back to original, if conservative or make a different change, if impatient. Rinse, repeat.
note - title changes are still "a major change" and should be respected as such.
| 9:21 am on Sep 14, 2009 (gmt 0)|
I'm not quite sure I'm understanding this right. So Google has a new file-system - what has this to do with the speed of changes taking effect? Of course the new file-system might be faster and more flexible - but to get a greater speed at changes you have either to ping google to tell that something has changed or google has to increase the speed of crawling. I still find subpages that hasn't been visited by googlebot since april this year, so I guess a change here wouldn't get into the index in 24h (unless Google isn't showing up as googlebot anymore). What did I get wrong?
| 9:26 am on Sep 14, 2009 (gmt 0)|
"note - title changes are still "a major change" and should be respected as such."
I absolutely agree. Title change = Subject Change and can drop you from connected markets. I have found in the past that when I change a title I also have to make alteration to the balance of associated body text and header importance. This probably seems really obvious to the experienced out there, but it really took me months to fathom which elements were being rated by G.
As whitenight says it would be wonderful to be able to say "nope that didn't work" and switch back to the original version without having to wait for a week or so. I suspect that this would lay in the hands of the private seo software developers rather than in Google tools etc.
The present test process I use at the moment is using a range of standalone sites for research. I try different changes on each and have an un-changed site as a control. I follow the Sandbox and watch for tweaks in the G dance tool, but never make any commercial changes until I see definite results on AOL. This obviously can takes months and I think sometimes it's just worth developing a new site to cover new keyword markets.
Sorry for waffling. Really enjoying this topic.
All the best
| 9:46 am on Sep 14, 2009 (gmt 0)|
|So Google has a new file-system - what has this to do with the speed of changes taking effect |
Its not just a new file system it is a new infrastructure which includes a new file system. It is what they do with it that is important.
It could be argued that previously googlebot brought back stuff over time and google sat on it, it was incorporated into a non live index and periodically the index was churned, eventually the the system decided that the mix was right and the data was pushed out to the data centres and went live.
With the new infrastructure as googlebot brings back its spiderings they are added into the live data. You could postulate that there is now only live data and no update and push system.
PS I could be wrong!
| 9:47 am on Sep 14, 2009 (gmt 0)|
Prompted by a Sticky, I'm noticing a quite a number of dubious penalties that were only applied in main SERPs have now been lifted- i.e normal listings more resemble Caffeine from a penalty POV
Sid, you were watching some penalty discrepencies that you felt ought to be penalised. Is there still a difference between Caffeine and main, and if not, are they both Caffeinated or decaf?
As an aside, now I'm watching the page discovery rate and reranking rate on Caffeine, main SERPs feels antiquated. How slow it must have been pre-Big Daddy!
| 9:52 am on Sep 14, 2009 (gmt 0)|
|With the new infrastructure as googlebot brings back its spiderings they are added into the live data. You could postulate that there is now only live data and no update and push system |
I think the main thing is that batch processing no longer happens. In the past, major variables used to get crunched in one go, then another variable, etc. Now, as soon as data is discovered, everything updates to take account of it. There is no work-through.
| 4:00 am on Sep 16, 2009 (gmt 0)|
Suddenly tonight I am noticing caffeine has damn near the same results 1-10 to whats live on Google.com. Is anybody else noticing that maybe the algo on g.com is finally the same algo applied on caffeine? I've checked a few queries and they are all the same.
| 7:06 am on Sep 16, 2009 (gmt 0)|
Completely different here.
| This 238 message thread spans 8 pages: < < 238 ( 1  3 4 5 6 7 8 ) > > |