|Just as if G forgott some of the incoming links for a while. |
That sounds like a possibility to me. The method Google uses to build up the data-set for search results seems to involve building and then combining a number of specialized lists.
In the past, some of those lists have have had incomplete data, or at least the "mixed together" data-set only polled in some of the specialized list. When those incomplete lists get used in production, then buggy search results happen. For example, strong home pages sometimes went missing for a while and then came back.
So the idea that some backlinks were accidentally skipped during Google's earlier steps at building up a recent data-set is a very real possibility, in my opinion.
[edited by: tedster at 6:53 pm (utc) on Mar 7, 2010]
"In the past, some of those lists have have had incomplete data, or at least the "mixed together" data-set only polled in some of the specialized list. When those incomplete lists get used in production, then buggy search results happen"
Maybe these buggy search results will be less likely to happen after Caffeine is implemented.
Google may be changing, but results I'm seeing suggest its is possessive, it's a contraction for it is :)
Seems these SERP changes are none too clear cut; interesting to learn google results can be buggy
|Just as if G forgott some of the incoming links for a while. |
Or, they're getting ready for an infrastructure change on a larger number of data centers and can't access all the information they normally use for calculations for some reason... Caffeine's only live on one data center (AFAIK) and there could have been quite a bit going on (left out for a bit) in the change over that would have gone basically unnoticed because getting to the Caffeine DC was (is) a 'hit and miss' thing, but the 'lack of access to data' is now more noticeable as they are likely getting more DCs ready to change over to the new infrastructure.
IOW: It might not be an 'error' or 'miss', so to speak, but rather a byproduct of phasing out Big Daddy.
Selfishly I hope you're wrong because a number of my SERPs in competitive territory have been slowly rising...
Huge changes in my niches today (guess I should turn off personalized results first).
Ted, I think you mentioned data issues before, in one of the earlier update threads. Do you really think G could be having issues on a site-by-site basis, wrt links (forgetting/missing some)? Could this explain some of the issues some webmasters have reported since mid January? I can't get my head around it, because after 2 months, it would seem that all "lost" data would now be accounted for. While some sites/pages seem to be ranking the same as always (unaffected), others (as many have reported) are seeing -40 to -60, lost/dropped pages, and/or drastic reduction in crawl rate.
crobb305, see the November 2008 monthly thread for the Halloween update.
A particular dataset was left out and it was possible to watch SERPS update over a period of days while the Ghost set was added in, and iteratively combined with the other data.
Shaddows, I am looking through that thread now. I see your comments in there. Thanks for pointing me in that direction.
|Shaddows, I am looking through that thread now. I see your comments in there. Thanks for pointing me in that direction. |
Great thread. ;)
In my UNbiased opinion, it should be stickied and required reading for any SEO who wants to understand how Gorg works
as opposed to simply OPINING upon how Gorg works.
In my BIASED opinion, what was i thinking giving out those secrets to Gorg like that?!
Seriously?! Thank goodness, only a few people were actually paying attention and taking it seriously at the time.
"Ghost datasets" is still under TM, as "gurus" are infamous for stealing my concepts/analysis without giving proper credit, although Shaddows and Cain should add their names to the pending TM request for future royalities. :)
Big stuff going on in my little perceptional woods of Gorg.
Tests that were started 2 months ago and SHOULD have taken 3-4 days to get results are JUST NOW giving me the results I've come to expect over the past 3 years.
So, for those unfamiliar with whitenight-speak....
Expect some major NOTICEABLE changes, that everyone can see, for rankings.
currently on hiatus from publicly debating about ...err... speaking about Caff. Up to each individual to determine how what i'm seeing is related or not. ;)
Started experiencing a huge uptick in traffic last night at around 8 PM CST. Large site, diverse topic, UGC. I track referrals to the minute and can usually tell when Gorg is having issues hours before anyone else is talking about it.
This is huge for me, ~70% increase in traffic from Gorg. Given the fluctuations I've seen this past week which are similar to early Caffeine tests I'd say Caffeine is now out of the chute.
Please, Gorg, let it stay like this.
Quote: I'd say Caffeine is now out of the chute.
Will the Caffeine rollout occur gradually or suddenly? I've been thinking that Google might implement it on one datacenter at a time. Also, it might require a period of time to do an extensive and deep crawling of the web to collect a full set of data for the algorithm.
|Quote: I'd say Caffeine is now out of the chute. |
I don't see any signs. Normal ranking changes maybe, but crawl rate is pathetic, no cache updates on sites I watch since Feb 25. If that's caffeine, it's weak.
Added: I have to add that I am not in a position to dispute those who have been monitoring closely. I certainly do not have a data set, just making an observation about the crawl rate. I am really hoping to see some improvements soon. :)
No caffeine as far as I can tell
small ranking adjustments. if that's caffeine then ... "meh" ... is all I have to say about it ... probably not much will change in the UK SERPS ... not much ever does ...
We are seeing more crawlers on our site..
I am not seeing caff results, but something is happening for sure, traffic is on the up tick.
Google had gotten excellent at cleaning up 301s. But now Google seems to be not cleaning up 301 URLs these days hardly at all. The new URLs appear in the index and rank, but the old URLs stay in the index too and just sit there and sit there.
Steveb, it is interesting that you mention that. For anyone who missed my earlier posts (in January), I will recap: another domain pointed their CNAME or DNS to mine (I believe in a malicious attempt to sabotage my rankings), and my content was ultimately indexed in Google on their domain. Their domain displayed a PR5 in the toolbar, and was showing MY backlink profile. It caused such confusion that somehow THEIR domain showed up alongside mine on my DEDICATED IP (discovered using the Domain Tools reverse-ip lookup). I set up 301s in mid January, but their urls are STILL indexed with my content. I eventually changed to a new dedicated IP, but my site is still being filtered in Google, presumably due to canonical issues. It is a big mess.
I don't mean to hijack the thread with old news, but I am simply agreeing with you on the VERY slow response in handling 301s, canonical issues, etc. This behavior is consistent with the extremely slow crawling that I am observing and stale cache.
This thread finally got me to sign up. ....
I've witnessed some peculiar traffic trends recently on my site. It is an aggregator site that collects thousands of updates per day. It has seen a pretty decent uptick in traffic that shouldn't be there for this time of year.
I think either caffeine has something to do with it, or Google had an internal update for my site.
Thought I'd share.
[edited by: tedster at 4:23 am (utc) on Mar 9, 2010]
Welcome, xposebrant. I am hopeful for some changes soon, myself. :)
|Google had gotten excellent at cleaning up 301s. But now Google seems to be not cleaning up 301 URLs these days hardly at all. The new URLs appear in the index and rank, but the old URLs stay in the index too and just sit there and sit there. |
That's true. I am seeing the same thing. I set up 301 almost 2 months back and old page is still there. When I see the catch date of old page, it's of Dec. 09. That page is not crawled since 3 months even though I have submitted the sitemap.
I am seeing search results for "widget" (single high traffic word) that has domain.com/widget but the page has no content on it related to "widget" and also one other in top 10 that has a "this page no longer exists for.." message (not a 404).
I am seeing some good things as well but for a few single term high traffic keywords I see REALLY bad results.
Don't mean to cause a mass panic or hysteria...
but who else is seeing Caff datasets on -
Cause I DO :)
and it means something fun :)
always nice when u stop by gford.
please do so more often.
Whitenight & gford, I do see some minor ranking changes AND finally an updated cache on a few sites that I watch (but those cache dates are still from March 1, despite the fact that they appeared today, March 9 -- that is sloooooow). Also the 301'd urls that were stuck in the index for months (that I mentioned above) are finally gone.
[edited by: crobb305 at 6:08 pm (utc) on Mar 9, 2010]
Yeah it is quite interesting after a bit more research. Long tail keywords seem to be stable or improved but I am still really dismayed at the singled word search phrases, which I could show a few examples..
steveb crobb305 and seolearner - glad its not just me :)
we've got a couple of site rebuilds with 301'd URLs in process at the moment, one is 5 weeks in (was showing PR5 before rebuild and usually recached pretty fast) and is only 30% redirected, the other is 2 weeks in and theyve cached all the new pages and maybe 3 redirects so far.
shockingly slow compared to how it was last year.
kevsta, do you still see those 301'd urls appearing in the dataset that Whitenight links to above? After 2 months, mine are gone from that particular dataset, but continue to show up across all other datacenters, and Google.com
whitenight - I see the same thing on comcast, but if I use comcast through a couple proxy servers I use then I do not. I guess these results will not be shown to all.
Seems those serps on comcast also seem to be missing datasets.
Sites are completly gone in them as in not being indexed when in fact they are and with a descent link profile.
Perhaps they just have not folded in all the data yet if that set is in fact caffeine.
nice catch whitenight.
I see caffeine on comcast as well. Do you often check comcast so you know it wasnt there before?
I usually check aol, no caffeine on there. Caffeine has been steady on all the caffeinated proxies I have been tracking though...still do not see it on my home or office locations.
Could be a good sign, hopefully it will be released sooner rather than later