McDar Experiment

bpal,

It has been my own personal experience to put up a link to a new page and even though I find evidence in my logs that Google has spidered and even cached the page with the new link on it, Google has NOT necessarily followed that link to the new page.

Yes, how deep Google is actually spidering is a factor.

 

NEW PAGE UPDATE:

The "New Page" sleeping-bags.htm
04/07/2004 "New Page" went live

Search for "Sleeping Bags"
Position#n/a allinanchor:#n/a - 04/08/2004
position#104 allinanchor#32 - 04/09/2004
Postion#232 allinanchor:#n/a - 04/11/2004
Postion#193 allinanchor:#n/a - 04/12/2004
Postion#136 allinanchor:#n/a - 04/13/2004
Postion#133 allinanchor:#n/a - 04/14/2004
Postion#108 allinanchor:#n/a - 04/15/2004
Postion#104 allinanchor:#30 - 04/16/2004
Postion#92 allinanchor:#32 - 04/17/2004
Postion#29 allinanchor:#13 - 04/18/2004 [11 datacenters #30, 4 datacenters #27, 2 datacenters #22]
Postion#28 allinanchor:#13 - 04/19/2004 [ 5 datacenters #27, 2 datacenters #22]
Postion#29 allinanchor:#15 - 04/20/2004 [16 datacenters #25, 6 datacenters #27]
Postion#25 allinanchor:#7 - 04/21/2004 [All datacenters]
Postion#18 allinanchor: #7 - 04/22/2004 [10 datacenters #19]
_____________________________________________________
note: these pages are found using ( site:www.compar.com +sleeping Bags )
number of Bob's PR6 and PR5 links found:
19 - 4/13/2004
24 - 4/14/2004
40 - 4/15/2004 (most datacenters report 40 but some as low as 31)
37 - 4/16/2004 (7 datacenters report 36, 7 report 42, the rest - 37)
37 - 4/17/2004 (8 datacenters report 36, the rest - 37)
38 - 4/18/2004 (10 datacenters report 37, 16 report 32, the rest - 37)
48 - 4/19/2004 (7 datacenters report 47, the rest - 48)
50 - 4/20/2004 (9 datacenters report 49, the rest - 50)
50 - 4/21/2004 (7 datacenters report 49, the rest - 50)
51 - 4/22/2004 (12 datacenters report 48, 5 datacenters report 50, the rest - 51)
__________________
note: this page is found using ( site:www.komar.org +sleeping Bags )
Alek's PR 7 Link:
Not reported by Google 4/17/2004
Reported by Google 4/18/2004 (all datacenters)
Reported by Google 4/19/2004 (all datacenters)
Reported by Google 4/20/2004 (all datacenters)
Reported by Google 4/21/2004 (all datacenters)
Reported by Google 4/22/2004 (all datacenters)
__________________
note: this page is found using ( site:www.ski-france-ok.com +sleeping Bags )
Foxy's PR5 and two PR4s
0 - 4/20/2004
1 - 4/21/2004
1 - 4/22/2004
__________________
# of Links from my posts to this thead
69 4/17/2004 [Edited to remove 20 posts in this thread that did not have link]
75 - 4/18/2004
90 - 4/19/2004
93 - 4/20/2004
97 - 4/21/2004
104 - 4/22/2004
__________________
Caryl's Links [links posted 04/07/2004]
2 PR6s & 5 PR5's (no way of telling which are accounted for)
__________________

 

I don't have a problem with those results. The fact of the matter is that "sleeping bag" singular, is not in any URL, or anchortext. So Google is reporting that correctly.

However when it comes to determining which sites are relevant for a search for "sleeping bag" there is nothing to stop Google from determining that a site with many reference to "bags" must also be relevant for "bag".

The point, I think, is that the algorithm and ranking tools that Google uses for SERP placement are not necessarily the same or limited to the tools they give us for research purposes.

 

Bob,

I agree! Yesterday someone reported seeing that the results in one datacenter for a search for link:www.domain.com was returning "old" data.

Indeed, when I did that search for my site, it did return the backlink count prior to the last update (April 7). But, if you did a regular search on the same datacenter, It returned the "new page" in it's current position!

It is important to note the the "new page" did not go live until April 7th, so if ALL of the info was old, the site would not have come up at all.

This again, points to evidence that Google is using data, other than what we can view, to perform it's algo on and determine results.

Well, speak of the devil...
You can see this for yourselves, right now on datacenter 216.239.53.99 !!!

... A search for link:www.mcdar.com is showing only 124 Backlinks (at this moment) that is OLD yet, the new page is coming up in the regular search for sleeping bags, at today's position #18.

 

I think this is a very important concept and worth a little more pontification on my part.

Google's raison d'etre, or mission in life, is to return the most relevant pages for any search. The problem we all have is that we have no idea how they calculate relevancy. They do not provide any tool called "allinrelevancy:". The only evidence we have is the SERP itself.

So why do we expect that the things we can measure, or even things they report, to have a 1 on 1 relationship to the SERP ranking? First I'm sure that Google uses many many factors. When we do an allinanchor or an allinurl or whatever we are only seeing a fraction of what Google may be using. Take stemming which we discussed earlier in this thread. There is no tool provided by Google to measure stemming, but clearly they are using it.

Now lets discuss the most misunderstood measurement that Google gives us -- PR. We have proved above that Google is using links to evaluate and position a site in the SERPs long before they report the PR figure to us. And yet there are people who impatiently wait for the PR update, because they are sure that then, and only then, are all those links taken into account.

Then there are people who are convinced that SERP placement is directly related to PR. It isn't, and it can't be, and here are the reasons why. PR is a mathematical number calculated from the shared PR of all interlinked sites. It is indicative of the number of links a site has, but it has nothing necessarily to do with the relevancy of the site.

Remember the second paragraph in this epistle -- Google is indexing pages in their SERPs based on relevancy. PR doesn't measure relevancy. Therefore PR cannot be a prime driver of SERP placement.

 

The whole bank of datacenters (in - 216.239.53.xxx) are showing OLD backlink info right now.

I wonder if they are getting ready for this weekends PR/Backlink update!

(April 25 will be 18 days since the last update and put it dead on Googles recent PR/Backlink update schedule)

 

Bob,

You may have missed my point! A datacenter is only reporting X amount of Backlinks, in this case indicating OLD Cached info., yet ranking a page that was NOT in the index at that point in time.

Google must store link info separately and searches within a datacenter are not all accessing the same data. Thus a regular search is using one dataset and a search for link: is accessing another dataset.

 

No I don't think I missed the point. I think that is exactly what I was saying. Google uses a lot of stuff that we don't know about or have access to.

It's like two parallel universes. All we can do is guess what is in the Google universe and there certainly isn't any one tools or measurement that is going to tell us how they position a page in the their SERPs.

 

Absolutely Compar

This parallel universe is showing up here more and more - just yesterday my mind went, after posting this

, that what google is doing is keeping two systems running with one overlaying the other and that is why sometimes we see the "old" results as an update is occurring in the new overlay as McDar reported.

On the other issue of PR and effect we know it doesn't have any effect from the positions and timing shown in this experiment - but I have noticed that the downward movement stopped at around 29 for a few days and then started moving again after new links were put in [by me] last Monday - now that suggests that a program of "building" links on a daily basis would have a more positive effect than a "dump" or perhaps an initial dump and then trickling on [as has been done here] is the best way to push the page to the top!

What does ya think?

 

Well it is very hard to tell for sure. But I would agree that adding links should be an ongoing and perpetual task. Because your competitor is probably adding links daily and if you don't you stand to be overtaken.

Even the initial "dump" as you call it wasn't picked up all in one day. It took several days for them all to be found, and infact I have added 3 or 4 more pages to my InfoPool since the beginning so even links from my site can be said to be trickling in. The new pages haven't got any reported PR yet but that apparently hasn't prevented them from being found and making a contribution.

And then there is the daily add of the links in Caryl's signature from this forum. So there has been a constant stream of new links added daily from the very beginning.

The interesting thing of course would be to stop adding any more links and see when and where this page plateaus out.

 

I agree basically with what you said in your full message, but I would be interested in where you see our experiment as being over complicated. We had a bit of that when Alex was obsessing over the exact time of being crawled etc, but short of that we changed the title on the page and then got on with link building.

I think this experiment has been the model of simplification. Where people have been looking for siver bullets, and one to one cause and effect realtionships, we have tried to explain that there are none. We have defused much of the urban myth about PR. If this experiment is anything, I think it is KISS. Keep It Simple Stupid.

 

Shawn,

Could you be more specific regarding the statement above? Was your comment regarding this whole experiment or a specific topic being discussed?

Thank you,
Caryl

 

Ummmmm ... the whois data for ccapplication.com looks ever so slightly familier ....

BTW, in another thread you mentioned a lotta Email spam going to your register contact information - looks like you have a new Email address - nobody@localhost.com - how much spam Email have you been getting to that?

alek

P.S. If you classify those as terms as "mildly competitive" (I looked 'em up on Overture's and Adwords), then my stuff doesn't even qualify as doggie scraps.

PPS. I think your last sentance is pretty darn good advice ...

 

Uh huh

Sometime back in the last century [thanks compar] I started out by just plonking pages down and building site content to where I got to 2500 pages [doesn't have that now] - I grabbed from places - I created - I went like the clappers without once thinking about SEO because I realised that my clients [all of one type] needed an all in one guide. This worked so well that I ended up with the knowledge that it is quantity is one way that will drive a site to the top.

However when I needed to be more specific to different people, that became more difficult and like the new millenium I had to find new ways and I turned to SEO, just like you Shawn in August/September of last year when you said on another forum that you were new to SEO [can't find the quote], and I found some interesting and contradictory results when compared to what other SEOs were saying.

So in some ways this is an extension of that drive and, although we waver around a little the essential truth is still there and what makes this Forum [obviously led by the boss] different in that people are willing to experiment with ideas.

So I agree with Compar, Mcdar, Alek that in fact I think this definitely kiss



BTW I have been running alongside this experiment another experiment [its called a double blind trial] using Weber BBQs - I have not detailed the experiment so far as I have been mirroring this experiment without outside links [only my own] and so far I am at 58 but I am No1 +2 for allintext and allinanchor but not showing on allinurl yet [which is a bit strange] but I seem to have stalled at the moment - so when we finish this one we will, perhaps see where the other is and make adjustments

 

Err.... read just got to 17+18 on Shawns KWT but it looks like it is just on

216.239.39.99
216.239.39.102
216.239.39.103
216.239.39.104
216.239.39.105
216.239.39.147

and this forum that raced to 6 has disappeared on this same block but is still there on the others

 

Sorry didn't look closely at each listing on 216.239.39.99

Your right

 

I have tried to update my hosts file but it does not seem to help. Can anybody check and see if the experimant page (link below) got a PR rating?