« previous next »
Pages: [1]

Author Topic: 40 G00gle changes in February  (Read 3570 times)

hungrygoose

  • Inner Core
  • Sr. Member
  • *
  • Posts: 316
    • View Profile
40 G00gle changes in February
« on: February 28, 2012, 01:03:28 PM »
http://insidesearch.blogspot.com/2012/02/search-quality-highlights-40-changes.html

Interesting read, I had noticed some increase rankings in Google Places (annoyingly) but the interesting part is at the very begining and at the end:


Most of the updates rolled out earlier this month, and a handful are actually rolling out today and tomorrow

Link evaluation. We often use characteristics of links to help us figure out the topic of a linked page. We have changed the way in which we evaluate links; in particular, we are turning off a method of link analysis that we used for several years. We often rearchitect or turn off parts of our scoring in order to keep our system maintainable, clean and understandable.  <<  Some people are saying PR, some are saying Anchor Text... Any ideas?

Spam update. In the process of investigating some potential spam, we found and fixed some weaknesses in our spam protections. << Any bets on this... I'm thinking since HPBL and some blog networks seem to have been hit, it all about OBL, maybe relevance of them and quantity...


I haven't seen any ranking changes to be honest, I have noticed a longer than normal lag in adding new links and ranking changes.  Will be an interested few days.
Logged

TallTroll

  • Sr. Member
  • ****
  • Posts: 272
    • View Profile
    • Email
Re: 40 G00gle changes in February
« Reply #1 on: February 28, 2012, 01:10:42 PM »
>> Some people are saying PR, some are saying Anchor Text... Any ideas?

Definitely not anchor text. Maybe PR. It's been so thoroughly gamed, it might not actually provide a decent signal:noise any more. Lots of other metrics attach to links now, maybe PR just isn't worth it any more. It's a theoretically infinite computational problem, who wouldn't want rid of it?
Logged

hungrygoose

  • Inner Core
  • Sr. Member
  • *
  • Posts: 316
    • View Profile
Re: 40 G00gle changes in February
« Reply #2 on: February 28, 2012, 01:22:41 PM »
Is there a better way than plain old PR to judge the trust of a site and therefore it's links though?  And it still seems to have a strong correlation to the effect a link has on ranking, well up to today.

Link evaluation. We often use characteristics of links to help us figure out the topic of a linked page. We have changed the way in which we evaluate links; in particular, we are turning off a method of link analysis that we used for several years. We often rearchitect or turn off parts of our scoring in order to keep our system maintainable, clean and understandable.

I can't help but think there is a message in there somewhere...

"characteristics of links" what does that mean
Logged

TallTroll

  • Sr. Member
  • ****
  • Posts: 272
    • View Profile
    • Email
Re: 40 G00gle changes in February
« Reply #3 on: February 28, 2012, 02:18:17 PM »
>> "characteristics of links" what does that mean

Eh, go read everything Bill Slawski ever wrote :) A few I can think of :

TrustRank (TR) of the donor page, TR of the recipient page(s) (remember, co-citation data is your friend)
Difference in term vectors of donor and recipient page(s)
Difference in term vectors of donor page and anchor text
Text surrounding anchor element
Average lifespan of links from donor site
Average lifespan of links to recipient site
Position of link in code
Time since this link, specifically, was discovered
Was link present during the first fetch of this page?
Has link target ever changed?
If so, to a different URL on the same target domain, or to another domain?
Has anchor text ever changed?
Is donor site linked directly, or indirectly, with known link buying activity?
Is link reciprocated?
Are the ownership of recipient and donor site identical (may be determined heuristically, or by matching WHOIS records, for example)?

Some of these are more important, some less so, but the specifics probably vary quite a bit from vertical to vertical anyway. One of those things you pretty much have to work out on a case by case basis. I'm sure there's plenty not on that list too

>> I can't help but think there is a message in there somewhere...

"Don't buy expired domains for the PR any more"
Logged

Gurtie

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1628
    • View Profile
Re: 40 G00gle changes in February
« Reply #4 on: February 29, 2012, 08:23:24 AM »
I think it just means what a lot of people have been saying for a whie, that links are no longer the only means of citations.

The web talks about sites differently now, social mentions, indirect links which are genuinely placed, etc etc. If you go back to the principals of pagerank, which are perfectly valid, but need to apply it to the web now, given what we all know about how people buy links and how people use the web, then you might easily reach the conclusion that older methods of link evaluation don't add anything to the algo now, imho.

Also, I'm not personally sure you would see an obvious rankings difference. They're not saying they're switching off a method which has been 99% responsible for link evaluation up to this week and switching on another. They've been moving for years. Now they switched off something which was hanging on in there pretty much in name only in my guess. Perhaps this is just a headsup that toolbar PR will no longer update?

I have seen some changes on local, unless I was seeing a test yesterday. Looks to be an improvement.
Logged

ukgimp

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 2241
    • View Profile
Re: 40 G00gle changes in February
« Reply #5 on: February 29, 2012, 10:56:22 AM »
>>Was link present during the first fetch of this page?

Loads of stuff like this you have to consider. In this case I always push for a new page to be created for each new link that I acquire. Otherwise it looks fishy. Whilst they probably cant compute this each time its the sort of thing they could run a job on to weed out link buyers/sellers.

Logged

TallTroll

  • Sr. Member
  • ****
  • Posts: 272
    • View Profile
    • Email
Re: 40 G00gle changes in February
« Reply #6 on: February 29, 2012, 11:34:27 AM »
>> I always push for a new page to be created for each new link that I acquire

Have you considered footnotes / updates? Flat out linking up an existing piece of content = spam, but adding an update to a news article = freshness
Logged

I, Brian

  • Inner Core
  • Sr. Member
  • *
  • Posts: 397
    • View Profile
Re: 40 G00gle changes in February
« Reply #7 on: March 01, 2012, 02:32:21 PM »
Quote from: ukgimp on February 29, 2012, 10:56:22 AM
>>Was link present during the first fetch of this page?

Loads of stuff like this you have to consider. In this case I always push for a new page to be created for each new link that I acquire. Otherwise it looks fishy. Whilst they probably cant compute this each time its the sort of thing they could run a job on to weed out link buyers/sellers.



Quite essential IMO - new links require new page created. Means more work with the link building, but very used to it now.

It was mentioned in a previous update that reliance on anchor text was being reduced, so I expect this happening further, and Google trying to use social signals more. However, with FB and Twitter limiting available data to Google (I think) I suspect this would be quite hard to do without an awful fudge.

Logged

Adam C

  • Inner Core
  • Hero Member
  • *
  • Posts: 626
    • View Profile
Re: 40 G00gle changes in February
« Reply #8 on: March 01, 2012, 03:41:19 PM »
Quote from: I, Brian on March 01, 2012, 02:32:21 PM
It was mentioned in a previous update that reliance on anchor text was being reduced, so I expect this happening further, and Google trying to use social signals more. However, with FB and Twitter limiting available data to Google (I think) I suspect this would be quite hard to do without an awful fudge.

Sorry to go OT here, but Brian's managed to hit on a pet peeve - the idea that Google's access to Twitter is restricted.

No, they don't have the pipe or whatever it was supplying them a real time feed, but they can certainly spider the f###er

https://www.google.com/search?q=site%3Atwitter.com

Quote
About 3,330,000,000 results (0.08 seconds)

...FWIW.  I think that number means 3.3 billion pages indexed which isn't bad going, is it?  I'd expect authoritative users will get spidered plenty freqently enough for Google to get all the signals they need.

FB's walled garden may be a different issue

https://www.google.com/search?q=site%3Afacebook.com

Quote
About 6,640,000,000 results (0.19 seconds)

or maybe not.

Obviously they're not getting the full picture from either site, but certainly a lot to chew on.
Logged

I, Brian

  • Inner Core
  • Sr. Member
  • *
  • Posts: 397
    • View Profile
Re: 40 G00gle changes in February
« Reply #9 on: March 01, 2012, 09:14:43 PM »
Quote from: Adam C on March 01, 2012, 03:41:19 PM
Quote from: I, Brian on March 01, 2012, 02:32:21 PM
It was mentioned in a previous update that reliance on anchor text was being reduced, so I expect this happening further, and Google trying to use social signals more. However, with FB and Twitter limiting available data to Google (I think) I suspect this would be quite hard to do without an awful fudge.

Sorry to go OT here, but Brian's managed to hit on a pet peeve - the idea that Google's access to Twitter is restricted.

Simply remembering Google's recent rant about how unfair it was that Google were allowing indexing of Google + profiles, but how competitors were not opening up their own social networks properly.

Certainly Twitter and Facebook are indexed to a degree, but not as open as Google want - or so Google suggest.

Either way, I'm minded of someone's recent comments about being surprised at how much social was going into Microsoft's Bing search, so am expecting this to be something Google are trying to turn up the knob on - though whether it's pushing on different ways of clustering sites and/or users or similar is another matter.

As for the impact of the new changes - not seeing anything yet but only just starting this month's client reports. Canary tests suggest no major impact.
« Last Edit: March 01, 2012, 09:16:57 PM by I, Brian »
Logged
Pages: [1]
« previous next »