Google's New Caffeine Search Engine - part 2

[Hissingsid]

< continued from [webmasterworld.com...] >

For my top target term there are still some differences between the Caffein sandbox and .com results. It does not affect me and looking at them dispassionately the Caffein results (subjectively) serve the user better. More of my significant real world competitors appear and less of the buy their way to the top jokers.

Cheers

Sid

[edited by: tedster at 5:52 am (utc) on Sep. 7, 2009]

[steveb]

Completely different here.

[Hissingsid]

Only first 4 the same rest completely different in my UK niche.

Cheers

Sid

[Shaddows]

Same as always- broad similarities but certainly not identical.

I'd say 90-95% of listings are +/- 10 positions in side-by-side comparisons.

[MLHmptn]

Yes, I am noticing it has reverted now. Odd to say the least. Sorry for rustling the feathers but earlier it was exactly the same for like 20 queries I tested and now it's not even close again.

Just make it live G!

[mambate]

我想跟之前差不多，但是可能排序计算更加快了

[Hissingsid]

Just make it live G!

These circles we are going in are getting smaller. FWIW I don't think that what is in the sandbox is ever intended to go "live". It is just a play thing for Google engineer trial and error.

I'm personally accepting the working hypothesis that the new infrastructure has already gone live, 2 months ago(ish). It explains a lot of what I have seen recently.

Cheers

Sid

[zehrila]

I have a feeling the new infrastructure has been executed, atleast partially. I have been monitoring few keywords and now i can see the ranks have been twisted to match to those of caffeine, in addition to that, the penalised sites are not listed in those serps.

Anyways, the first serp for the keyword iam monitoring is totally not worthy of top position.

[barretire]

Has anyone seen any changes today. I know google is adding all kinds of new stuff and was wondering if any of it involves Caffeine?

[jd01]

我想跟之前差不多，但是可能排序计算更加快了

I'm with ya there, mambate...
That's exactly what I've been seeing in the results lately!

I hope they clean it up soon.

[crobb305]

I see my rankings going in opposite directions on caffeine vs regular index (rising on caff, dropping on reg), with significant differences between the two. I can't see how caff has been executed yet but could be wrong. I don't really have a grasp on what is going on with this new infrastructure.

[Lorel]

I've been watching G.Caffeine vs the old Google for my #1 and #2 keyword +city.

While rank is steady for top 3 sites in both engines, it is jumping around for positions #4+ in Old Google and the rank hasn't changed at all in top 10 for G.Caffeine in the last 4 days.

So I'm wondering if G.Caffeine is settled out.

[edited by: tedster at 4:34 pm (utc) on Oct. 5, 2009]

[jd01]

Points made by whitenight (Link Below)

"Caffeine was primarily an infrastructural change"

2 points in the first sentence.

He uses the word "was", NOT "is". or "will be".
Why?

He also uses the word "primarily", not "exclusively" or "only".
It obviously is NOT just an infrastructure change.
He has continually used the word "PRIMARILY" indicating it is/was also more.

I highly recommend the entire thread and interview with MC.
[webmasterworld.com...]

[edited by: tedster at 4:36 pm (utc) on Oct. 5, 2009]

[sirkevon]

Caffeine is going through a system maintenance at the moment. Let's access the damage when it comes back. :)

[cangoou]

It's back again - and the normal SERPs where shaken again in direction of caffein (for me at least), but the results still are not equal.

[Lorel]

I've been watching results for another site (highly competitive 3 word phrase) and Caffeine hasn't changed at all for top 30 results since 5-5-09 (except for taking out one duplicate).

However the Old Google has sites jumping up and down by 10 or more positions.

[barretire]

I am seeing the same reults in main search as I am in caffeine this just started.

[barretire]

Okay got anotherqueston for you guys that has stumped me. As i said in the previous post it appears I am getting Caffeinated results. I had some others check and they are getting the same. In other words the search looks the same on both caffeine and regular Google. here is my question. There are a handfull of terms that I am in the top 10 (some top 5) for with the new results. These are very popular terms, however following my tracker I am not getting any hits for these terms. I have read on an earlier post that with Caffeine it wil or may kill traffic programs. I use a free one through extremetracking. Could it possibly be that the hits are not showing up. I know this sounds like a Newb question but I have had a handfull of peope search and the site is coming up on page one.. Makes no sense if I am not getting traffic for these terms or am I. I am getting traffic from other terms. Looking at my click throughs on my affiliate program it seems that traffic has definately increased quite a bit but my tracker does not show this. Any thoughts? Is this caffeine possibly. I would have never noticed a difference if I did not actually perform a search every day on select terms. I noticed this about 5 hours ago.

[giggle]

barretire: I just did a query on google.com and it has the # after the .com/ which means that it's not passing the search parameters back to your referral code. Thought that they'd tested and forgotten about that one. Not happening on google.co.uk (at the moment).

edit: having said that a google.co.uk query now has that # in their URL. (I'm searching from Thailand).

[cangoou]

I still see massive changes here (Germany) in SERPs compared to 2 days ago. The SERPs don't match the Caffein-SERPs at all. So I have 3 result sets:
1.) The set from 3 days ago (which is still showing up sometimes)
2.) The set now showing (most of the time)
3.) The Caffein-set
.. and it's all mixed up, like positions:
Site 1: 8, 16, 4
Site 2: 30, 250, 8
... and so on. No idea how to make tests in such an environment.

[whitenight]

You're doing great observational analysis cangoou (and barretire).

Test something small and non-critical or just watch some pages that are the most divergent within those datasets
and figure out what you can.

Soon you'll be coming up with your own ideas to test that prove or debunk your intuitive thoughts.

Maybe Shaddows will come out of the shadows (sry, couldn't resist) and add his analysis.

[cangoou]

Thanks for the flowers whitenight ;-) I was wondering how to make conclusions out of such observations. For example, if a domains jumps from 80 to 20, it could be:
- That it got some very good links
- That it got out of a penalty
- That some kind of "this-site-is-too-young" filter was released
- That the sites before 80 were all put in a penalty (okay, thats not very likely, but you know what I mean: Got my site one up because I did something right or got the competitors site one down because he did something wrong?)

There are a lot of levels which could explain the observations - or am I just thinking to complex?

[barretire]

I just want to update: I found (actually with much help) that my tracker code was not working properly. therefore now that it is fixed I am able to track all traffic compared to my earlier post. However I will say that I have seen alot of bouncing going on> I myself am seeing very simialar results to the sandbox DC however I seem to be the only one.

[whitenight]

I myself am seeing very simialar results to the sandbox DC however I seem to be the only one

Not the only one. ;)
Been seeing it off and on for over a month now.

I just get tired of debating the board about stuff that's
ultimately there to help them.

It's becoming more obvious now, so we just wait til others catch up.

[whitenight]

- That it got some very good links
- That it got out of a penalty
- That some kind of "this-site-is-too-young" filter was released

Ok, now you just need to figure out a way to test these theories.

Each test leads to new insights til you narrow down some key aspect that will help your sites.

[barretire]

I keep seeing the same. It also makes sense that Google is realeasing all of these gadgets, (many of them I consider worthless) It seems it would make sense to do this after releasing a new infastructure and not before. Just my thoughts.

[Lorel]

I'm seeing differences on both DB for two sites I've been tracking (NW US). First few sites don't change positions but the others bounce around.

[srik79]

Here in India, going to google.com gives
[google.com...]
Does that mean its caffeine?

search results are same as regular google.

using the google toolbar gives
[google.com...]

[whitenight]

Glad there are people on the other side of the world watching.

Google just rolled out a completely new dataset across all DCs in about 4-5 hours!

People in the west are going to wake up and wonder what happened.

It's was nearly seamless, with the 3 datasets cangoou pointed out, being weaved into each other like a basketball drill.

That was the Caffeine infrastructure in effect and it was quick and painless.

I suspect there will be some more movement as I still see some "ghost datasets" not incorporated
but when they do weave those in, it will happen quick!

Go to sleep or out to lunch and one will miss it completely.

[Shaddows]

Missed the fold-in, but sandbox and main SERPs still don't match. I've always thought they were not too disimilar, and they remain comparable.

I've been trying to resolve the different interpretations posited in this thread, especially by HissingSid and cangoou. I think Sid had it right:

FWIW I don't think that what is in the sandbox is ever intended to go "live". It is just a play thing for Google engineer trial and error.

Specifically, either or both of the base data and ranking algo are different to the live version, although somewhat similar. The sandbox seems to be a test of the store, fetch and batching processes. Thus, there will never be a "roll out" or converging datasets.

I'm sceptical that the live version is on the same infrastructure, for two reasons. The first is that the invitation to Sandbox makes no sense if its raison d'etre has already been fulfilled. The second is that the physical infrastructure overhall of the actual datacentres would be a massive project- there should be definitive (or at least anecdotal) evidence of this.

I have absolutely no knowledge of large scale distributed files systems. Could the mechanisms used for load-balancing be re-purposed to push all traffic to a few upgraded centres, to mimic a complete DC upgrade?

[JackPage]

I honestly haven't paid that much attention to caffeine since it was announced but I have seen something very interesting in the past few hours now that I've sat down to look at it.

Over the past four days I've been on a bit of a link building campaign for two sites. Both are a couple years old, and about a year ago they both fell out of number one spots for various keyterms. I let it go because I was just messing with them part time and had other stuff going on. Four days later, I'm already seeing some strange affects from this link building in Caffeine.

Site one:
Caffeine - Few hours ago it moved to third page for it's keyterm from way back in the SERPS somewhere. Right now, it's moved again all the way to number four on the first page. It's been no where near this position in a year.

Current Live Google - No change

Site two:
This site has never ranked for the keyterm I targeted four days ago before.
Caffeine - Page 1, 8th position

Current Live Google - No change

Neither of these sites have been under a penalty before, just lost a lot of links over time and suffered the affects.

Granted, I'm not as full time into studying the search engines as many of you, but I've never seen sites rise that fast like that. I didn't pick up links that were such quality that they'd bounce a site like that that fast either, least not in the live engine. Least I'm pretty sure I didn't.

Whatever's going on, here's hoping that the Caffeine results I'm seeing become live at some point.