Forum Moderators: Robert Charlton & goodroi
the SERPs were different in the past few days for sure.
changed mid day today.
yep i think its started, 74.125.95.132 appears to be the cache server for live results, from where I'm looking anyway.
hold on to ya hats this weekend.
Vimes.
If you are, can you post for which ones?
Wonder what this will look like in Canada... Does anyone have a Canadian datacenter IP address?
I do, but to be honest, all I see is what appears to be the regular shuffling of results, most movement within 1-3 positions. As expected I don't see any monumental changes, at least in the sectors that I watch, save for those 1-3 position movements (that we were accustomed to seeing pre Holiday SERPs)
What I am noticing, however, is a range of previously filtered websites released from filters. These are websites that were filtered back due to what I would deem excessive SEO on the part of the webmasters.
I emailed Google, Matt Cutts and his team for confirmation - so stay tuned.Update: A Google spokesperson told me, "we have nothing to announce today regarding Caffeine." Basically, it is not live yet and when it does go live, they will let us know.
[seroundtable.com...]
[edited by: tedster at 8:43 pm (utc) on Jan. 7, 2010]
[edit reason] added the link [/edit]
And yes, there is an algo change at least in the one keyword I tested (international results topped an existing term, same domain but .cctld. They have many more links, better site so it makes sense)
That doesn't prove an algo change. The rankings could have changed because Caffeine includes deeper crawling, and so additional data could have been factored into the calculations, thereby causing the other site to move up.
Happened on or near 11am today.
Am wondering if Caffeine was live in LA for a couple of days, then G turned it off.
What exact data could they be looking for by turning it on and off?
"Everytime I try to get out, they pull me back in!"
Too many suppositions to clarify, so I'll respond to no one in particular.
-- Caffeine Datasets are flickering on and off.
-- Yes, it is a very DISTINCT SERPs from what is currently being shown
(at the exact moment I'm writing this, which could be "on" by time you are reading this)
-- The Caffeine datasets HAVE BEEN "live" in a sporadic fashion since Dec 24th around 10pm(PST)
(Gorg's definition of "live" and mine are obviously different. If a Caffeine dataset is sending me traffic, it's 'live' regardless if it's the final permanent version or not)
-- A DATASET change and an ALGO change are TWO COMPLETELY DIFFERENT THINGS - please see my long and repetitive explanations re: what Caffeine "is", by following this thread all the way back to it's origins.
------------------------
Now, bonus points for the following posts.
1. 209.85.225.103
2. 74.125.95.132
3. 66.102.7.18
4. 64.233.163.19
5. 74.125.19.106
6. 66.102.7.104
What exact data could they be looking for by turning it on and off?
Are they "testing" or simply screwing up?
With the Caffeine datasets being LIVE (ie sending actual and large amounts of traffic) for periods of time, one wonders what the hold up is....again.
That doesn't prove an algo change. The rankings could have changed because Caffeine includes deeper crawling, and so additional data could have been factored into the calculations, thereby causing the other site to move up.
At the risk of frustrating you even more....could you perhaps C&p your original post?
To "CainIV": I agree with the filters being lifted part. Saw several sites over the past few days that matched your description rise to the top. Ones with exact keyword matches were #1 as well.
ty
Hey White, these forums are a bit confusing (threads being sliced and moved/etc)....I wasn't able to find your explanation regarding the difference between a dataset and algo change.
[webmasterworld.com...]
which leads to the first Caffeine thread.
(sorry can't copy and paste, it would take up too much space and annoy the mods)
And I'm sure i was ranting in another thread about Caffeine Algo change v Dataset change.
(Think it was in MC's interview thread?)
Yes, it is a very DISTINCT SERPs from what is currently being shown
If you can, please clarify the qualifying factors you see that make that particular SERP different than any other standard ranking shift on post holiday SERPs
Walkman:
And yes, there is an algo change at least in the one keyword I tested
Not entirely sure if I am understanding what you mean. Are you seeing different websites that you are accustomed to, in new positions in the SERPs, and attributing this to Caffeine? If so, can you give us qualifying factors why you believe those results are tied to the infrastructure release.
Marketingmagic:
I am currently drawing from 72.3.232.139 which is, from all accounts, not a suspected ip that might be delivering those results.
If you can, please clarify the qualifying factors you see that make that particular SERP different than any other standard ranking shift on post holiday SERPs
I'll try... with the usual caveat -- that I often see things in the "matrix" that others simply don't see. (or won't acknowledge)
Of course, we have to go back to past posts to fully elucidate.
IN Nov and Dec, various SEO "Guru" X,Y,Z were seeing Holiday SERPs and calling them Caffeine IPs, when in fact, there were both datasets on different IPs.
(I hinted at this back then, and to the careful observer, it could be seen)
So now we have the similar situation in reverse:
Caffeine Datasets, Post-Holiday Datasets, and "Normal" datasets.
Some people are seeing Post-Holiday Datasets and calling them Caffeine. But they are neither the "normal" datasets, nor Caffeine.
For my "testing/control" sites, the Caffeine datasets are obvious.
I receive traffic (and quite a lot) for terms that ONLY rank on Caffeine and not on the other 2 datasets.
ie. some extreme cases would be #40+ SERPs on other 2 datasets and #2 SERP on Caffeine dataset.
---------------------
On top of this, we still have the fact that SEO GURUs X,Y,Z have NOT figured out that Caffiene is 2 different things with Gorg.
There's the Infrastructure update
and a Dataset rollout.
(which should be named something else, but i'm tired of getting people to see this...as if Goog Employee #1-#3 doesn't say it, it supposedly isn't "true")
IMO, The Caffeine infrastructure update is a complete mess right now.
(which is probably why they are hesitant to rollout the dataset now)
As several threads indicate and I observe,
If the old infrastructure is called "Big Daddy 1.0"
And Caffeine is "Big Daddy 2.0"
(aka superfast indexing, etc)
Then, again, after trying to roll out Big Daddy 2.0 in mid Dec, it was such a disaster that they reverted back to Florida 3.0!
Which is the nonsensical indexing and updating that's been occcuring for the past month or so.
So again, when asking Gorg Employee #1-#3 about "Caffeine" and it's status, one doesn't even know what the reply is referring to.
And they aren't going to spell out this fact to an SEO, no matter how "friendly" they are.
So now we have the similar situation in reverse:
Caffeine Datasets, Post-Holiday Datasets, and "Normal" datasets.Some people are seeing Post-Holiday Datasets and calling them Caffeine. But they are neither the "normal" datasets, nor Caffeine.
For my "testing/control" sites, the Caffeine datasets are obvious.
I receive traffic (and quite a lot) for terms that ONLY rank on Caffeine and not on the other 2 datasets.
ie. some extreme cases would be #40+ SERPs on other 2 datasets and #2 SERP on Caffeine dataset.
Whitenight,
Can you please tell us what we can look for to identify a dataset as Caffeine?
yea white see what you mean on those DC's
1. 209.85.225.103
2. 74.125.95.132
3. 66.102.7.18
seeing a huge difference in the page returns for site operators on large media websites. i wasn't getting these results yesterday.
Vimes.
Whitenight,Can you please tell us what we can look for to identify a dataset as Caffeine?
At the moment of this post (which again may change by the time someone reads it), it isn't on any "public" IPs.
and yet, they are still filtering traffic thru it.
(lol, seems Gorg really hates when every SEO can spot their testing DCs.
So even now I'm getting intermittent traffic from Caffeine datasets but they are hidden from any of the public IPs/datacenters)
It's kinda funny.
1. 209.85.225.103
2. 74.125.95.132
3. 66.102.7.18
4. 64.233.163.19
5. 74.125.19.106
6. 66.102.7.104
Even for webmasterworld the .* query numbers are just 50% what they used to be.
that I often see things in the "matrix" that others simply don't see. (or won't acknowledge)
Well we definitely appreciate the education. :)
I receive traffic (and quite a lot) for terms that ONLY rank on Caffeine and not on the other 2 data sets.
Again, how do you know it is Caffeine? Saying that you see something on one data sets that doesn't rank on two other data sets - how does this confirm to you that this data set is Caffeine? It seems to be somewhat circular logic - a tree is a tree because I call it a tree, but without any previous reference, how would one know this 'new' set is x,y or z?
This was my point with many of the last posts in this thread.
Because the data set contains rankings that were previously not seen, or are different from post Holiday SERPs, wouldn't justify as Caffeine 'by default'
As indicated, i have seen Caffeine "flicker" on and off with the below public IPs.
I have seen many different data sets on those ip's. What distinguishes them from Caffeine?
Thoughts?
Yes, it's tree because I say it's a tree....
... Plus, I made a rather big stink back in Oct about keeping a eye on the Caffeine datasets,
which again, TO ME, are highly different than the other datasets that have floated around thru the past 3 months.
This was the reason why...
All i can tell you, is i don't usually post ANYTHING, unless i'm 99% sure of my data.
(as always, or you can check my track record on this type of analysis)
So... you'll just have to believe i saw a black cat do the same thing twice. ;)
... Plus, I made a rather big stink back in Oct about keeping a eye on the Caffeine datasets,
which again, TO ME, are highly different than the other datasets that have floated around thru the past 3 months.
I know, and this is why I asked select members of this board that I know are very good at what they do.
Of course I am a big Matrix fan, so that answer definitely will suffice :)
It's NOT caffeine.
What's NOT caffeine?
The 3 different datasets? Today? 3 hours ago? Yesterday?
the 1 flickering dataset? Today? 3 hours ago? Yesterday?
The 10+ datacenters/Ips? Today? 3 hours ago? Yesterday?
The current infrastructure? Today? 3 hours ago? Yesterday?
You're going to have to clarify what you are referencing in this thread.
