Forum Moderators: Robert Charlton & goodroi
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]
Well, the observation is that almost every SEO seems to rank better on the caffien-DC then in the normal SERPs. The Q is: What does that mean?
Its a good question, worth some testing to see if any of the above are true. Anyone else got ideas?
Who said "only SEOs are doing well" on sandbox.goog?!
It's obvious who LIKES sandbox.goog as they are speaking up loudly in this thread.
It's also subtly obvious who doesn't like the SERPs if one follows this thread back to it origins.
Personally, I haven't spoken one way or another whether my sites or clients sites are doing better or worse on sandbox.goog.
My one and only comment about the "quality" of the original sandbox SERPs was that it appeared many minor Over-Optimization Penalties had been removed.
And I can certainly see enough "SEO'ed sites" doing worse in the results.
Remember, the first page is a ZERO-SUM game, so somebody is doing worse, whether they speak up in this thread or not.
Let's start with more solid premises before jumping to anecdotal conclusions.
it appeared many minor Over-Optimization Penalties had been removed
That could line up with a faster infrastructure. On the legacy system, phrase-based spam thresholds were only recalculated on a rather long cycle, probably because those calculations are computationally intensive. If that cycle has been speeded up for Caffeine, then there could be fewer dolphins caught in the tuna net.
That could line up with a faster infrastructure. On the legacy system, phrase-based spam thresholds were only recalculated on a rather long cycle, probably because those calculations are computationally intensive. If that cycle has been speeded up for Caffeine, then there could be fewer dolphins caught in the tuna net.
Yes, exactly.
For those interested, please reference the earlier Caffeine thread about why I believed the yo-yo issues would also be eliminated.
Same reasons.
Relative to the main SERPs, I'm barely up or down in the Sandbox. However, looking at the results, there are less amateur sites in the mix, and sites seem to be rewarded for what I would call "structure".
I've already mentally moved away from thinking the sandbox impacts SERPs, or that direct inferences from one can be applied to the other, now or in the future. As far as I'm concerned, the Caffeine architecture was "on display" in the sandbox, and will be eventually underpin the main results, once the infrastructure investments are in place.
As an off the wall suggestion, the dataset fold-in could have been G figuring out how to utilise both architecture/infrastructure set ups (caffeine and decaff) within one dataset, rather than as discrete entities. This would allow incremental datacentre upgrades without any perceivable disruption to main SERPs.
sorry not been following too closely as i've had my own little disaster to prevent in the last couple of weeks.
Has there been a major change, from the data sets i'm seeing on google and the sandbox i'm seeing the caffeine results way behind the live ones.
it appears a very incomplete data set, from my view the caffeinated results are approximately 5 days behind the live SERP's.
if they were the same then i wouldn't be that baffled but from cangoou and whitenight's comments i'm not seeing the speed of the infrastructure or the three sets of data. (damn i hate it when i miss things)
i can only assume that the live results i'm viewing, are the sandbox results that you guys are seeing, would that be a fair assumption.
Vimes.
it appears a very incomplete data set, from my view the caffeinated results are approximately 5 days behind the live SERP's.
lol, no it is confusing.
At some point, We need to all agree on the same terms.
the 3 dataset roll-in was on the live SERPs (which goes with my theory that the Caffeine infrastructure is mainly already in place)
And yes, the dataset on the sandbox.goog does seem to be behind right now.
We need to all agree on the same terms.
How about this
Sandbox = test area
caffeine = the suite of changes requires for infrastructure upgrade
infrastucture = architecture + dataprocessing + ranking algo
architecture = physical DCs, specifically the hardware
Thus,
Caffeine is live in the sandbox.
Since the fold-in, caffiene is unquestionably in evidence in main SERPs. I have pages ranking, TOP 3, minutes after publication
I doubt Caffeine is live on all DCs, as there has been insufficient opportunity to get the architecture in place to implement the required infrastructure.
Sandbox and SERPs diverge for reasons unknown, although I suggest algo or base data differences.
Sandbox seems completely irrelevant since the fold-in (apart from esoteric testing, of course.)
For one term I watch #3 has moved to #1 and the previous #1 and #2 have moved down one slot on all 3. Similar consistent across all 3 changes in other terms I watch.
It looks to me that the sandbox is live and has same index but slightly different filters. Could simply be geo-filter. Geo filter on .co.uk is not working to create a somewhat different serp as it usually does in my UK niche. It looks like .co.uk geo-filter is not being applied.
It looks like the Google engineers all came back from a good Friday lunch and one too many Coors and decided to have a play.
Cheers
Sid
I've been tracking a 3 word phrase in both old Google and Caffeine (Keyword, Keyword, City) for over 2 weeks and printing out the first 20 results of each and I do not see the same results in both. The sites are mostly the same in both but positions do not match at all and I checked to make sure one wasn't just delayed beyond the other either. Also the city map does not show up in Caffeine.
A couple days ago after doing several searches, it looks like Caffeine is a MUCH closer match on Google.
All in all, things are looking ok. I can't speak from any penalty point of view to know whether anything like that would have been going on during the last few months of watching as I don't believe we have had any penalties on our site. So, strictly reporting from a 10 year old, ~40k page website, I think Caffeine is matching on Google more now than it was during the last few months.
The big thing that I have noticed during all of this is that for our primary keyword phrase, a few sites have popped into the first page results that really have no business being there. They are both more local sites where the rest of us are a bit more general. For example if it were travel, the main competitors on the first page might be travel for the US. Whereas two of the new results are maybe for a particular town or a particular destination in a state. Extremely localized which we have never seen at that level. This started in Caffeine and has migrated now over to Google. Neither of these sites are spam so at least that is a bonus. ;)
One of our long time competitors has been replaced in position by one of these sites which I actually think is a little sad because despite them being a competitor, the information was more relevant than the newcomer. Also, another old competitor that had dropped several pages nearly 2 years ago, has popped back on to the first page which I believe to be in better position than the several pages down. I prefer them not to be higher than us but that's part of the game.
Those are my observances from the past few months.
None of the above seems to have the same effect on the rank in caffeine as the same sites have different positions in both engines.
Just an example of how fast the new infrastructure is.
Cheers
Sid
Thats for a completely new page for a new product on a ecom site (i.e not a blog or forum or other "freshness-affected" site). It came in above several other ecoms and displaced the manufacturer.
As its not QDF, I took that to mean internal nav structure is currently a key componant of the algo.
For reference, a highly trusted review site is in at #1. I expect to take it within a month as we gain links.
Well, the Q then is: Why is the www2 sandbox still online?
Good question.
Since many here NOW agree that Caffeine is live (ahem...still waiting on some props here from certain ppl),
I have a feeling the "sandbox algo"
(formerly known as the Caffeine algo/dataset) will become the traditional holiday dataset.
Besides being a massive FUD distraction (also remember this was released around public Bing! excitement)
Again, it's a different dataset/algo. Caffeine infrastructure, in one form or another, has been LIVE since August.