| 5:41 pm on Oct 8, 2002 (gmt 0)|
anchor text is much lower now though in the ranking scores. In highly competitive searches this has the impact of making the results less relevant
- Very high PR sites trump, so sites like MS can get to be #1 even though the page has 0 relevancy
- SPAM can find it's way in
Perhaps a reduction in the relevance for anchor text was to combat googlebombing like "go to hell", but simply checking that the anchor text was in the destination page would have achieved the same thing without reducing the quality of the index - which has certainly occured.
| 5:51 pm on Oct 8, 2002 (gmt 0)|
What is the PR of the sites that you are claiming that anchor text doesn't count? There are a lot of conflicting blanket statements about this.
| 6:38 pm on Oct 8, 2002 (gmt 0)|
It's a fair assumption that every factor counts - it just depends how it is weighted in the algo. Presumably each factor creates a score that then determines the ranking.
So if you have a very high PR (8+) then it may become the dominant factor - there is a thread about PR being dead but I don't see that at all - I see anchor text reduced in importance in the algo, but it is still a factor & if you have a lot of such relevant anchor text then it can still win through. I see for highly competitive search a site #2 PR8 - no on page keywords (well except for some hidden text) & some anchor text. Previous update #2 was PR7 - lots of on page relevancy & masses of relevant anchor text - dropped to #13. My site was PR7, on page factors & moderate level of anchor text - #6 down to #26
Problem with PR on it's own is that it is not keyword specific it is just a popularity contest - so if PR has a higher weight then it can create a lower relevancy.
It's all just guesses based of what i have seen.
| 6:53 pm on Oct 8, 2002 (gmt 0)|
|keyword proximity seems to be a casulty of this update |
I couldn't agree more. We find high PR sites with one, or two of the phrases but not together. I am not affiliated in any way with word1 word2 word3, it just helps illustrate the piont.
Search "word1 word2 word3" (no quotes)
1 = word1 otherword on the otherword word2
2 = a word3 otherword about the word2 otherword
4 = word1 day otherword otherword ... word3 and word2 are nowhere on the page
9 = (word2 in domain): word1 otherword ... word1 otherword ... little otherword word3 ... otherword word2 otherword
10 = "word3" on the "word1 word2"
12 = word1 otherword word2 of otherword ... alt.word3
In all of these examples "word1 word2 word3" appears nowhere together on the page. What do we see here?
|Very high PR sites trump ... even though the page has 0 relevancy |
[edited by: ciml at 7:42 pm (utc) on Oct. 8, 2002]
[edit reason] Examples generalised. [/edit]
| 7:02 pm on Oct 8, 2002 (gmt 0)|
What I have stated in a few threads is that you can have a factor's importance specifically weighted by another factor. This is most likely the reason for conflicting theories as to what is important.
People seem to think that the algo is something as simple as (all factors 0-10 and grossly over simplified)
PR + link_text + title + domain ...
Then to get a little more complicated some will consider weighting of each value instead of just tossing things out (each "weight" would be between 0 and 2)
(PR * PR_weight) + (link_text * link_text_weight) + (title * title_weight) ...
But it would be very easy to decide the importance of a factor given it's other factors.
(PR * PR_weight) + (link text * (link_site_PR / 5)) + (title * (percent _matching_link_text * 2))
And this would not even begin to do all the different calculations justice.
The point I am trying to make is that link text does/doesn't matter arguments are just about pointless. If you only have PR2-PR4 sites linking to you and the link text doesn't seem to make any difference, it doesn't mean that you shouldn't worry about it when that PR8, authoritative site in your field offers you a link.
This is why there are probably so many conflicting opinions of what factors are important.
| 7:04 pm on Oct 8, 2002 (gmt 0)|
and the adwords are alll "word1 word2 word3"
- I am not saying the 2 are related - not at all, just making an observation.
[edited by: ciml at 7:45 pm (utc) on Oct. 8, 2002]
[edit reason] Words generalised. [/edit]
| 8:14 pm on Oct 8, 2002 (gmt 0)|
My post was similar to austtr's before it was edited. By substituting "word1 word2 word3" for the actual phrases searched we lose site of the irrelvancy involved. I respect the moderator, but by 'generalizing' the phrases, at least in this case, we lose the point of the argument. I made no reference to a commercial site, but was illustrating that keyword proximity is very important when searching, and the new algo seems, for the most part, to have devalued it's importance.
At the risk of being edited again,
big screen tv is very different than tv on the big screen
6 of the top 12 results for big screen tv (no quotes) have nothing to do with "big screen tvs"
[edited by: kris at 9:57 pm (utc) on Oct. 8, 2002]
| 9:23 pm on Oct 8, 2002 (gmt 0)|
such editing is a bit over zealous - but I guess it's the rules
| 2:33 am on Oct 9, 2002 (gmt 0)|
I have read a lot of what has been summized, in both this and the other long thread on the subject. So forgive me if I missed any reference to this possibility (of which I have seen none).
Going back to the "florida surfing" results, and others I have looked at. Can anyone else see a return by Google to giving (limited) outbound link sites an advantage. By limited I mean maybe links to 2 other closely related / linked (interlinked) sites, possibly / probably of the same owner (but where the subjects are complimentary, not copies), plus links to purely reference (.edu / .org) sites? Of course, if there are a lot of links out as with one of the florida surfing examples, maybe this is % based, not number of outbound links.
| 4:09 am on Oct 9, 2002 (gmt 0)|
|The diminishment of pagerank this time around means perhaps some of them aren't very good sites, doesn't matter, they have on topic content and are trying to be relevant... there are only so many words you can put in a title and these sites choose the two words. |
Except that we're told to think like Joe Surfer. Joe Surfer doesn't care that the title is relevant. He wants to know about the waves in Florida (thats easy, IMHO they s*&k, except for the Sebastian area). Many of these sites don't fit the bill and therefore , to Joe Surfer, are poor results. She may figure its not the SE's fault, or she may go try somewhere else. IF Jane Surfer went to alltheweb and did the same search she would find far less satisfactory results. The aforementioned #10 is up to #5. MSN comes in ahead of atw , slightly behind Google.
| 10:21 am on Oct 10, 2002 (gmt 0)|
I am new to SEO and this goes against the general consensus of:
|Very high PR sites trump ... even though the page has 0 relevance |
I have a #1 position (low traffic phrase) with a PR0 above a PR4 and a PR7 with relevance - in that order. Optimization and good code still have some value.
As an aside, I learned that I was hit with a Penalty and dropped to PR2 on index and PR0 on internals. No matter what you hear you cannot use WPG for any reason, not late at night, not on the weekend, not with 90-second timeouts, not with 1 simultaneous search, and not for "just" 12 phrases, etc. Thank goodness for my three #1 positions for key phrases in Y! Wait a minute, that's gone too.
There's no place like home. There's no place....