homepage Welcome to WebmasterWorld Guest from 54.161.228.29
register, free tools, login, search, pro membership, help, library, announcements, recent posts, open posts,
Become a Pro Member

Home / Forums Index / Google / Google SEO News and Discussion
Forum Library, Charter, Moderators: Robert Charlton & aakk9999 & brotherhood of lan & goodroi

Google SEO News and Discussion Forum

    
Googlebomb Algo - new details from Matt Cutts
tedster




msg:3834460
 3:00 am on Jan 26, 2009 (gmt 0)

Here's a little extra peek behind the curtain from Matt Cutts:

we do two different things both of them algorithmic to handle Googlebombs: detect Googlebombs and then mitigate their impact. The second algorithm (mitigating the impact of Googlebombs) is always running in our productionized systems. The first algorithm (detecting Googlebombs) has to process our entire web index, so in most typical cases we tend not to run that algorithm every single time we crawl new web data. I think that during 2008 we re-ran the Googlebomb detection algorithm 5-6 times, for example.

[mattcutts.com...]

He goes on to say that the most recent running of the first algorithm identified both the "failure" and "cheerful achievement" bombs aimed at whitehouse.gov - but the last time that Google ran the first algo, it missed the "failure" bomb - something many people noticed, since this is arguable the most well known of all.

This makes me wonder if the first googlebomb algo sometimes catches too much identical anchor text that was intentionally placed by the site owner, too.

[edited by: tedster at 4:20 am (utc) on Jan. 26, 2009]

 

potentialgeek




msg:3834469
 3:39 am on Jan 26, 2009 (gmt 0)

Interesting... it seems he's more specific than usual. At least I'd never heard him previously mention Google uses more than one algo. It's not a major risk revealing this, however, because that type of bombing isn't very common. The two cases he cited, furthermore, were for entertainment, not commerce.

This makes me wonder if the first googlebomb algo sometimes catches too much identical anchor text that was intentionally placed by the site owner, too.

Not sure. That would be like Google giving out new info on the 950 penalty... which is one of Google's most closely guarded secrets. :/

p/g

whitenight




msg:3834540
 7:05 am on Jan 26, 2009 (gmt 0)

this is too funny.

The other day, I was just thinking about the 'disagreement' with tedster over the effectiveness of the original "anti"-Googlebomb thread as I had recently observed certain "discrepancies" between what Google SAYS and what the algo DOES over 'bombing'.

Nice timing MC. Beating me to the punch. ;)

As always, anyone who's truly interested in whether this is FUD or "truth" might want to TEST on whether:
the Googlebomb "diffusion algo" works...,
ever did work...,
or only applies to certain MAJOR well-known sites/terms

(much like the link-selling "threat" to MAJOR well-known sites)

whitenight




msg:3834541
 7:17 am on Jan 26, 2009 (gmt 0)

This makes me wonder if the first googlebomb algo sometimes catches too much identical anchor text that was intentionally placed by the site owner, too.

That would be like Google giving out new info on the 950 penalty...

You know...

The 950 started becoming really apparent around the same time as the anti-googlebomb algo was put into effect.

And as I postulated in the original thread on the topic, it seems "politically coincidental" that MC is discussing this issue AFTER the POTUS has been sworn in...

As it only became an "issue" of Goog's as the Primaries started to get focused on.

It would be truly sad, if the anti-bomb algo was what was indirectly/directly responsible for the 950 penalty...
which was caused by certain "non-discussed" issues regarding Goog's increased involvement in political matters.

callivert




msg:3834550
 8:04 am on Jan 26, 2009 (gmt 0)

The 950 started becoming really apparent around the same time as the anti-googlebomb algo was put into effect.

That would make sense. Googlebombing algos mean that the relationship between anchor text and SERPs is now really messed up. (ie non-linear, non-monotonic, etc...)

Shaddows




msg:3834575
 9:18 am on Jan 26, 2009 (gmt 0)

Interesting observation. So the OOP may or may not be intentional. That would support observations that you get -950'd by having too much anchor text that does not match your web pages. You're not getting OOP'd, you're getting anti-Googlebombed.

I see potential for abuse. Althogh there is a certain risk in getting a competitor links using a keyword phrase they might target (otherwise why bother getting them banned for it) but haven't optimised for onsite.

OTOH, dialling down onsite keyword stuffing has reportedly got people released from the OOP, suggesting that it has an onsite element.

tedster




msg:3834893
 5:53 pm on Jan 26, 2009 (gmt 0)

One thing seems clear, the first anti-googlebomb algo seems to be pretty light handed, rather than causing lots of collateral damage. I say that since there are certainly bombs that get missed - the "failure" bomb itself was missed at the end of last year, by Matt's own admission.

I'm assuming that blog links are a big part of that first algo - they certainly have been a big part of the history of the googlebomb.

JS_Harris




msg:3835103
 9:21 pm on Jan 26, 2009 (gmt 0)

There can be no single algorithm to rule the net. The moment one is released data begins collecting to ensure it's failure.

I think Google is lowering the value of links by location as much as by filters. Example: any links in footers, external links in sidebars, internal links found in "similar page" sections.

It also looks like Google gives only full value to links within content so long as the link was there the first time the page was crawled. I think these factors play into the Googlebomb scenario as well which leads to confusion when people try to figure them out.

[edited by: JS_Harris at 9:22 pm (utc) on Jan. 26, 2009]

nealrodriguez




msg:3835141
 10:02 pm on Jan 26, 2009 (gmt 0)

did bush ever request to remove listings for 'miserable failure' and get ignored? did google just remove these bombs when king obama requested it?

whitenight




msg:3835154
 10:25 pm on Jan 26, 2009 (gmt 0)

did bush ever request to remove listings for 'miserable failure' and get ignored? did google just remove these bombs when king obama requested it?

The whole thing STINKS to the heavens.

It originally smelled fishy when Goog first gave their reasons why - which i went on a massive "political" rant on - which now is becoming more obvious to others.

Let's call it "Separation of Search and State"

When I saw Eric Schmidt (not-so) subtly campaigning for a candidate on Meet the Press, I immediately got nauseous.

Again, not because of WHO he was in support for...

Because Goog, as the de facto supplier of 80% of the world's information SHOULD NOT BE INVOLVED IN POLITICS...period

"Separation of Search and State"

Google SHOULD be the neutral "news and information" giver.
Regardless of who's offended, not offended, for Goog's policies, against Goog's policies.
(I won't even start on the possible big brother implications that are obvious from Goog being buddy buddy with policy makers)

"Separation of Search and State"

And NOW, its looking like, once again, "politics" have corrupted THEIR OWN ALGO.

One would be hard-pressed to prove the -950 and the anti-WHITEHOUSE-bomb are NOT related.

How very sad. Truly sad.
Strike that.

How very infuriating. Truly outrageous!

tedster




msg:3835173
 11:11 pm on Jan 26, 2009 (gmt 0)

I don't know of any information resource that is without political agenda - none whatsoever, not in any medium. That's not the way things work, anywhere anytime, on planet earth.

The anti-whitehouse bomb was just one very visible example of a more widespread problem with the existing algo - and Google took some steps to try to put a finger in the dike. This thread is about some new comments that give a bit of extra insight (one part runs only a couple times a year).

I'm not saying that Google is totally "pure" or anything like that. I am saying that this is not a cause for anyone to elevate their blood pressure over. If I were in charge of Google, I would probably be more heavy handed than they have been when someone tries to manipulate the service.

Yes, play watchdog if that's your choice. Every dog deserves a good watcher.

grasshopper




msg:3835191
 11:37 pm on Jan 26, 2009 (gmt 0)

well said, tedster.

if you're feeling thoughtful, or have insomnia, there's an interesting PhD thesis called "Through the Google Goggles: Sociopolitical Bias in Search Engine Design" here:

[web.archive.org...]
Note: link downloads pdf

[edited by: tedster at 11:45 pm (utc) on Jan. 26, 2009]

[edited by: Robert_Charlton at 12:05 am (utc) on Jan. 27, 2009]
[edit reason] make link live [/edit]

whitenight




msg:3835193
 11:40 pm on Jan 26, 2009 (gmt 0)

When you start with the WRONG premise, you get faulty reasoning and logic, tedster.

So, FIRST we have to RE-visit history with the MORE INFORMED VIEWPOINT so that at least we can find better FUTURE solutions, yes?

The anti-whitehouse bomb was just one very visible example of a more widespread problem with the existing algo

Wrong premise, wrong assumption.

Please name the thousands of sites that had Google bombed their sites to the top and weren't on-topic?

Or are you claiming the THOUSANDS of posters in the -950 threads are there due to "bombing" their way to the top?

IN FACT the whitehouse bomb was horribly obvious for years before Goog did a whiff of anything about it

AND,

here in the states, we STILL have this little thing called FREEDOM OF SPEECH in this other little thing called the CONSTITUTION

The "bomb" was only embarrassing to those with a STRONG political agenda. And certainly not "corrupting the results"

(Again, I ask you, as I asked you 2 years ago, who EXACTLY was searching for "miserable failure" looking for "miserable failure products, widgets, info" and getting whitehouse.gov "corrupting" their search?)

This thread is about some new comments that give a bit of extra insight (one part runs only a couple times a year)

Perhaps you missed my above argument about the Google bomb algo and the -950 algo?

What more "insight" do you want?

If people want solutions to the -950, I suggest they go back and re-read potentialgeek's thorough posts of his recoveries, experiments, and tests.

The title of this thread is "Googlebomb Algo" and until one REALLY understands that the algo was only SERIOUSLY implemented from pressure of unknown POLITICAL sources, in regards to "miserable failure" then we are talking AROUND the issue, so as not to cause a "ruckus"

This is a disservice to the thread topic, webmasters, AND Google.

So again, i postulate as in my FIRST Post, that the algo is STILL FAULTY, both in it's effectiveness for "bombing" ON-TOPIC anchor texts and OFF-TOPIC (according to who again?) anchor texts.

It doesn't filter EITHER very well and the CONSEQUENCES are what affects WEBMASTERS on this board.

I don't know of any information resource that is without political agenda - none whatsoever, not in any medium. That's not the way things work, anywhere anytime, on planet earth.

Thanks for the insight Tedster.

Since their political agenda is AFFECTING the SERPS, seems like one might want to be a "watchdog" about such thing, yes?

Unless you don't CARE about your SERPs?

Then of course, it's better to roll over, play dead, and find a LESS politically motivated business, where no one has to worry about having the blood pressure raised by politics OR unexplainable "heavy handed" penalties?

tedster




msg:3835199
 12:08 am on Jan 27, 2009 (gmt 0)

here's in the states, we STILL have this little thing called FREEDOM OF SPEECH in this other little thing called the CONSTITUTION

Which protects Google's algo as well as our right to publish as we see fit - but always within certain legal limits.

How could define what an "algo without faults" would look like? Thinking that such a thing could exist is an unrealistic premise.

Every algo including Google's is, and will be, "flawed" - if even makes sense to use such a word. And even further, every entity on earth must cope with political pressure, even as it attempts to exert political pressure. To think otherwise is to hold onto an idealism such as I have not entertained since I was 12 or so.

The paper that grasshopper linked to above (welcome to the forums, grasshopper) is a prize-winning thesis and quite balanced in its presentation. This is not an area where wild-eyed, knee-jerk ranting will bring about much of a valuable result.

I'm all for watch dogging - but in this case, most definitely NOT a governmental watchdog. That is the future to be feared, IMO - and much more than any Google bias. The dark possibililty of governmental requlation in the search industry is one reason why I prefer a more even-handed and nuanced look at the events we see unfolding from Google.

whitenight




msg:3835212
 12:27 am on Jan 27, 2009 (gmt 0)

Free Speech

Which protects Google's algo as well as our right to publish as we see fit - but always within certain legal limits.

Sigh...who or what EXACTLY do you think the free speech clause applies to?

Hint.

ESPECIALLY for negative commentary on the GOVERNMENT.

-- should i post a "non-ranting" link to the "prize-winning" Constitution and the "prize-winning" theories of the founding fathers here? to prove my point?

(Btw the founding fathers ranted...ALOT...) --

Again, Goog had NOTHING to worry about with webmasters placing "miserable failure" links to the whitehouse.gov

and

had NO REASON to change the results/algo.

(so let's do some BASIC guessing as to why they WOULD CENSOR the results or EVEN CARE)

idealism such as I have not entertained since I was 12

Then why are arguing AGAINST my point so adamantly?

Have you replaced your political idealism with "Google idealism"?

Seems alot of people have done that.

I'm all for watch dogging - but in this case, most definitely NOT a governmental watchdog. That is the future to be feared, IMO

Sigh, i said three times
Separation of Search and State.

NOT Government oversight
NOR Google CEO being a key advisor to the POTUS.

again reference the "prize-winning" theories by US founding fathers about what EXACTLY the people should be afraid of...
and who ULTIMATELY should be "watch dogging" the government AND huge government-like corporations like Google

And if THIS doesn't get your blood boiling then please step to the side while those of us, who DO CARE, get a little peeved.
(is that allowed?)

What exactly is WORTH watch-dogging? if NOT this? Praytell?

---------

Now, if you want to counter-argue that the -950 penalty and the anti-WHITEHOUSE-bomb algo are NOT related
(which i believe you were even pointing to in the first post), then i'm all ears.

But lets start on the CORRECT assumptions before moving on.

Miamacs




msg:3835281
 1:32 am on Jan 27, 2009 (gmt 0)

...

... and in other news of planet Search, Mr MC of G commented on the agenda of his country in a rare behind-the-scenes public PR post:

I think that during 2008 we re-ran the G Image Search algorithm 1-2 times, for example. AT MOST!

( 1-2... I'd say that's about accurate ) [j/k]

...

CainIV




msg:3835298
 2:15 am on Jan 27, 2009 (gmt 0)

It seems pretty evident to me that the issue is inherent in the algorithm itself, and still is to this day.

Some modifiers and filters have been put in place it seems to dampen the effects of this phenomenon, however, it is pretty apparent that it is live and well.

For what its worth - it's unfortunate that specific instances were addressed while others remained unchecked, and is rather convenient that it was this particular political problem was addressed.

ogletree




msg:3835360
 6:08 am on Jan 27, 2009 (gmt 0)

I think that during 2008 we re-ran the G Image Search algorithm 1-2 times, for example. AT MOST!"]I think that during 2008 we re-ran the G Image Search algorithm 1-2 times, for example. AT MOST!

That so explains a few things.

[edited by: tedster at 6:21 am (utc) on Jan. 27, 2009]
[edit reason] fix quote box [/edit]

Shaddows




msg:3835460
 10:07 am on Jan 27, 2009 (gmt 0)

I'm less concerned about successful people advising government than I am about undeclared political influence over private companies (or private influence over government). The fact is, if you disallowed successful individuals from advising the State, then government would necessarily be made up of (and be advised by), er, miserable failures.

Anyway, on topic. Does anyone know how the "failure" failure was officially rectified? Did they tweak the algo so it got detected and filtered, or just put a filter on the URL for that phrase as they would with a manually penalised site?

And if you were designing this filter, how would you go about it?

If I were Google, I would look for unvaried anchor text that did not fit my co-occurence profile for the destination URL. At a threshold, I would make a post-ranking minus filter apply to that URL for that phrase.

Now, either I would predict or I would subseqently notice that a bunch of sites had "accidentally" been affected. What would I do then?

Well, I would have 2 choices. Dial up the threshold (widening the window of abuse), or modify the filter so it had a dual function to deliberately weed out repetitive, manipulative link acquisition.

Now, imagine I found that a filter (possibly written for political reasons) intended to have a relatively minor role was now having a proper, major and widespread effect. What would I do now?

I'd start tinkering at the edges, trying to get consistant results. Now, given that I'm playing with an unintended consequence, the methodical fiddling is likely to have quite choatic consequences, with people AT THE MARGINS being pretty much clueless as to why they are penalised, released, and penalised again.

This is all just a thought experiment. As stated by others, faultless reasoning (not that I'm claiming such) is still belied by false premises, and I've only just begun to consider the OOP and Google-bombing within the same concept (although a moment's dispassionate thought shows both rely on the same trends, at least for off-site elements).

I guess some key question for Google include:
- At what point do healthy optimised inbounds become unhealthy repetitive bombing?
- Can cooccurance and phrase-based indexing be the whole answer to this (it can't because that would be easy to diagnose, and frankly no-one builds a link profile with indentical anchor text that is irrelevant to the destination)?
- Can inbounds from off-topic sites be evidence of bombing, as well as paid link aquisition (and if such links are toxic, how does G guard against darkside manipulation)?
- On-page factors (existance of keyword) HURT the site in OOP, but may validate the site as not a G-bomb target and thus AVOID the filter. How does this tension resolve?

econman




msg:3835626
 1:47 pm on Jan 27, 2009 (gmt 0)

There were two aspects of the original "failure."

1. It gave the appearance that Google was promoting a particular political viewpoint. This was an uncomfortable situation for a publicly held firm, particularly since the viewpoint in question happened to be one that was likely shared by all 3 members ofthe triumvirate that controls said publicly traded company.

2. It gave the appearance that Google was doing a lousy job of quality control and/or that its algorithms don't work very well. This was an uncomfortable situation for a profit making firm, particularly since the firm's reputation for high quality search results is fundamental to its success.

It seems very likely, however, that the attempted solution to this problem led to a new interest in focusing on certain flaws in Google's approach to measuring relevance based on links and anchor text. The subsequent attempts to overcome these flaws probably caused some collateral damage -- particularly if they experimented with different solutions and observed results which, at least in some situations they viewed as improvements. Bear in mind that Google's view of an improvement might not be the same as yours or mine (push highly optimized or SEO'd sites down 950 notches, thereby encouraging webmasters to invest in Adwords, rather than SEO).

ogletree




msg:3836205
 4:24 am on Jan 28, 2009 (gmt 0)

high quality search results is fundamental to its success.

That has nothing to do with their success. There were plenty of search engines that were doing fine before they came around. Google is a success because they made a search engine and only a search engine. They then figured out how to make profit with that. People keep saying that quality of search is what makes or breaks a search engine. That is only true if every search brings up spam. All the major search engines do a good job at bringing up results that are helpful. Very few people do complicated searches. Most people use google as an address bar to get to their fav websites. Did any of you study the aol data that was released. If you have a popular website you know that your top keywords are always your domain root. People type in company names. By letting wikipedia float to the top of every search google knows that will answer most questions typed in. Google does not care about advanced users all they care about is huddled masses that don't have much time for the Internet and just type in a few searches here and there. Those are the people that click on adwords and make them money. People in the search world think that everybody else searches the way they do. Google is at the top right now and will stay there as long as the vast majority of people keep "googaling" things. You can't watch tv or listen to a random conversation without somebody saying that.

Global Options:
 top home search open messages active posts  
 

Home / Forums Index / Google / Google SEO News and Discussion
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Home ¦ Free Tools ¦ Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About ¦ Library ¦ Newsletter
WebmasterWorld is a Developer Shed Community owned by Jim Boykin.
© Webmaster World 1996-2014 all rights reserved