Google's Rank Modifying Patent for Spam Detection

[tedster]

Patent hawk Bill Slawski has published an article [seobythesea.com] about a newly granted Google patent - one that seems to explain a lot of the odd ranking behavior that has been reported in recent times. The patent is named, quite simply, Ranking documents [patft.uspto.gov]

As I understand it, the idea here is to identify, by algorithm, what looks like an attempt to manipulate rankings in a spammy way. The patent details things like keyword stuffing, invisible text, link-based manipulations and so on.

Rather than allow the rankings to respond immediately and directly to those changes, the patent explains a system that would change rankings in unexpected, counter-intuitive ways - while the rankings change from a first position through transition positions and to the final "target rank" position. In other words, significant changes in position continue to happen even though there is no change in page's ranking factors!

Bill explains:

During the transition from the old rank to the target rank, the transition rank might cause:
a time-based delay response,
a negative response,
a random response, and/or
an unexpected response

If these transitional ranking shifts are followed by what looks like a responsive action from the website, the URL or website which was previously only SUSPECTED as spam, might now be positively tagged as spam.

Read Bill's full article [seobythesea.com] (or the patent itself [patft.uspto.gov]) and see if it doesn't explain some of the odd behavior people have been reporting in recent months:

[driller41]

This was the contention floated by Tom over at Zdnet. I can guarantee this is not the case. I talked to a Googler yesterday imploring them to chime in on the post, he said that they're tried to rationalize with him in the past, to no avail, so they didn't get involved.

I believe that fixing a Panda demotion is difficult as is a Penguin demotion - none of mine have gone yet.

So they now throw some randomisation into the mix for an as yet unknown period of time. Moving on and buidling afresh is a smart move - revisit in a years time.

Fighting these sort of variables is a timewaster - what is the rationalisation for sticking with a badly hit site.

[thegypsy]

[So they now throw some randomisation into the mix for an as yet unknown period of time. Moving on and buidling afresh is a smart move - revisit in a years time.]

Again, this isn't entirely new. "So they now" is a misperception.

Patent filed in Jan 2010. Which means likely working on it on 09. Furthermore, there are elements of a transition rank as far back as Dec 2003 with the historic ranking factors patents.

Main;

[patft.uspto.gov...]

Related;

[appft1.uspto.gov...]

[appft1.uspto.gov...]

[appft1.uspto.gov...]

Sister patent;

[appft1.uspto.gov...]

This is why there is context required for this story. Sure, the terminology has changed, (like the obvious KW stuffing of the term [spammer] in this one...LOL) but there is still an evolution and we shouldn't start looking for something new. More so, we should compare what we know and have learned over the years with this being a glimpse into causation in our daily activities.

Really want to get ahead? Stop reading patents and start reading papers. Research papers from Google or even those of major Uni (Stanford comes to mid). Why? Because patents are the past. Papers are the future. This is where the ideas and people begin the genesis for the future of search.

When one reads a patent, SOP is to research the people that worked on it. What department(s) have they worked in? Where did they go to Uni? What papers did they write at UNI? One seeks to get a deeper mindset into the patent itself. Those writing papers today, could be a Google employee tomorrow.

Anyway, rambling...

[tedster]

Speaking of papers... Google maintains a page for "Research Areas and Publications [research.google.com]" that can be intriguing - the link is always available at the bottom of our Hot Topics area [webmasterworld.com], which is always pinned to the top of this forum's index page.

[thegypsy]

Nice... a yummy looking page indeed.(the one here on WebmasterWorld that is...)

[RegDCP]

What I do take issue with them on though is the damage to SERP rankings of "innocent" sites inherent in the mysterious "shuffle" used to befuddle and identify those who zig zag in response to Google's moves

What "damage"?

Unless the transitional results places higher than the final result, there is no damage.

[aakk9999]

From what I can also gather:

a) During the "transition period" the ranking may fluctuate more than once (i.e. there could be more than one "transitional rank" employed)

b) Changes to the web page during the "transition period" in order to "fix" the transitional ranking must be positively identified as rank-manipulation spam by Google to suffer negative consequences

c) If the changes in reaction to transitional ranking cannot be positively identified as rank-manipulation spam, but are suspected to be rank-manipulation spam, the transitional phase may introduce even further fluctuation ("noise") in order to see if there are (and what kind of) further reactive changes and in that way positively conclude that the page changes are being done in order to manipulate rank

d) If, as a result of "transition period" the page / site is positively identified as employing "rank-manipulation spam", this will result in the negative ranking score that can be applied to a site, a domain or to a link

e) This "ranking transition add-on" (for the lack of the better name) may be applied to a page, the whole site or even to a number of sites with some common denominator (e.g. the same site owner, sites using the same template etc.)

f) The history on how the site/page was changing over the time may be used to decide whether to apply "ranking transition add-on" to a page/site

I agree with thegipsy that it would be impractical to apply this across the board and that it is only done for a selected pages/sites that are already suspect to engage in what Google calls "rank manipulation spamming" (or perhaps where the jump in ranking is over x threshold)

[diberry]

What I do take issue with them on though is the damage to SERP rankings of "innocent" sites inherent in the mysterious "shuffle" used to befuddle and identify those who zig zag in response to Google's moves

I hadn't even thought of that. In order for a zigzag site to go UP, other sites must go down. There's less and less connection all the time between how you run your site and how Google ranks it.

[RegDCP]

I hadn't even thought of that. In order for a zigzag site to go UP, other sites must go down.

I really doubt that the zigzag will place the site above it's final resting place.

If all the zigs and zags are below the final position, no collateral damage is done as the sites below it's "in transit positions" will be going down anyway.

The sites above the in transit page, but below the final resting place gain a temporary reprieve.

[RegDCP]

@aakk9999

I think you hit it squarely on the head.

It looks like new sites *may* just be safe.

Perhaps, given the (unknown) criteria it will be better to drop the old page and make a completely new, or just make a new page with new content.

[netmeg]

(Thanks for taking the time to join the discussion here, Dave)

[Leosghost]

Unless the transitional results places higher than the final result, there is no damage.

If the transitional rank moves up just once during the "shuffle" and stay there for 30 days ..someone else has to have moved down..that is 30 days of damage right there..SERP positions of pages do not exist in a vacuum thats why they are called positions<= which means relative to each other..

I agree with thegipsy that it would be impractical to apply this across the board and that it is only done for a selected pages/sites that are already suspect to engage in what Google calls "rank manipulation spamming"

Yes ..but any "playing with the rank of one page" ( and the patent does not restrict movement to only downwards in transitory phases, nor to only one movement, there may be many , both up and down whilst G sees what the reaction is ) ) in order to better identify SEO must, of it's nature effect others..in a similar way to if a strong backlink goes away because the linking site falls or even goes off line, or closes, ..the receiving site can be affected adversely and slip down even a small slide can result in major traffic losses.

Now imagine that there are 5 or 10 SEOs playing in your "keyword space" ( or anyone else's ) that G is "shuffling" up and or down, the disruption to the other innocent sites in the "keyword space" can cause real damage, as they are moved around to accommodate G's "games"..

And G talks of the possibility of some of these moves lasting days, weeks and or months..

Even applied to 1% of SERPS..that is major disruption to sites that are doing nothing "wrong"..some of them may not survive being caught up in such manipulation..and meanwhile the SERPS are not the best quality that G can serve searchers..but are the best quality which serves Google and their anti SEO experiments..

Sure they are faced with a dichotomy ..they want to remove heavily manipulative SEO and spammy site construction..but it is kind of like "all the class has to stay in after school , because someone did wrong, and teacher isn't sure who"..even the kids whose last bus home will have left by the time they are all allowed out..

[tedster]

I first saw this kind of ranking behavior in 2008. It was for a household name website in a very crowded market with two synonymous keywords - and the rankings began to pop all over the place right after a big online marketing agency (not mine) began to work with them on-page, on-site and off-site.

They went from page three to #4 then #16 and so on. They ended up at #4. The ranking fluctuations only stabilized when the SEO agency STOPPED making changes to improve rankings, but that new stability took months. And we are talking about a very big international brand, here.

Some of the fluctuations were seen several times within the same day, but others only showed up as trends over an extent of time. Sometimes how you watch the data is as important as the fact that you watch it at all.

[edited by: tedster at 6:55 pm (utc) on Aug 31, 2012]

[aakk9999]

Yes ..but any "playing with the rank of one page" ( and the patent does not restrict movement to only downwards in transitory phases, nor to only one movement, there may be many , both up and down whilst G sees what the reaction is ) ) in order to better identify SEO must, of it's nature effect others..in a similar way to if a strong backlink goes away because the linking site falls or even goes off line, or closes, ..the receiving site can be affected adversely and slip down even a small slide can result in major traffic losses.

I have read the patent and from what I understand, whilst the transitionary movement can also be "UP" , this transitory "UP" is never higher than the target ranking (if I misunderstood this, could someone please quote the patent sentence.)

Therefore, when the transitionary ranking for the site is UP, it means that other sites that have dropped -1 position would have dropped anyway when the target ranking is reached*. In fact, this will mean that some of sites that will eventually drop when the target ranking is reached* may not drop straight away and are (because of the patent) given some grace period.

(*) Of course, providing that during the transition phase the page is not marked as using spam methods

However, if the page was (during the transition phase) marked as having used spam methods, it will be penalised and the target ranking will not apply any more. The page will be penalised. In which cases the sites that during the transitionary period were -1 will come back for 1 spot. In fact, such sites will just gain because of the page being marked as spam since if there was no patent applied, the other site would just overtake them since it would come to its target position straight away.

I am not sure if I explained well. To try an example, lets say that my site is at #7 and another site is currently #12, and then they do some changes that reflect in ranking:

1) With no patent applied:
- my site is #7
- another site gets its target rank calculated as #2
- I drop straight away -1 position to #8

2) With patent applied, but manipulation was NOT seen as spammy
- my site is #7
- another site gets its target rank calculated as #2
- the transitional rank comes into play and moves the site to #5 first
- I drop straight away -1 position to #8
- the transitional rank now moves the other site to #9
- I climb back to #7
- the transitional rank patent now moves the other site to target rank #2
- I fall back to #8

3) With patent applied, but manipulation WAS seen as spammy
- my site is #7
- another site gets its target rank calculated as #2
- the transitional rank comes into play and moves the site to #5 first
- I drop straight away -1 position to #8
- the transitional rank now moves the other site to #7
- I remain at #8
- the transitional rank patent now decides the other site was using spam techniques, and penalises the other site. Target rank does not apply now and the site is moved to #50
- I climb back to #7

As you can see from the above examples, if there was no transitional rank patent applied, my site would straight away dropped as a consequence of another site coming up above.

And even during the transitional phase, my drop can either be delayed or I will drop anyway for this -1 position.

And only if Google decides the other site was using manipulative tactics, then I may climb back - to a position where I would not be should the patent not have been applied.

If Google can determine on whether the site is using spammy methods well enough, then I would not have problem at all with this patent. The problem I am seeing really is:

- Google guidelines on what is acceptable and what is spam change over the time
- It stops the webmaster testing what works and what not
- For each keyword, there are many good sites but there are only 10 places on the first page of the SERPs. If you think you have a good and a relevant site, it is only natural you want it to be ranked higher. Many times these changes are not done using so called spam-manipulatinon, instead they are done by using what you have in the best way. And testing this may now be problematic.

[RegDCP]

If the transitional rank moves up just once during the "shuffle" and stay there for 30 days ..someone else has to have moved down..that is 30 days of damage right there..SERP positions of pages do not exist in a vacuum thats why they are called positions<= which means relative to each other..

Right, but the site would have moved down anyway when the page assumed the proper level.

[RegDCP]

And only if Google decides the other site was using manipulative tactics, then I may climb back - to a position where I would not be should the patent not have been applied.

But you would not have climbed higher than your original position.

[diberry]

Can anyone answer me this?

I have several friends who, thanks to unemployment, have gotten interested in building websites to make a living. They come to me for advice. On quite a few occasions they've told me they've read from several sources that you should [black hat technique] and I have to warn them, no, that could toast you with Google.

The problem is, not all black hat advice is stamped as such, or presented in a way the typical newbie can identify as untrustworthy. In fact, I've seen some white hats sharing advice over the years that I'd consider black or at least gray.

If they didn't have me, and they genuinely didn't know better than to try the black/gray techniques, and then they got hit by a penalty and realized the error of their ways, how do they recover with this patent? Does anyone know? Or can they not? Is Google just determined to destroy any site that's ever engaged in manipulation?

And what if Google bans them, but their site appeals to visitors so they get a decent traffic stream from other sources? Is Google going to jump on board and "forgive" them, or... I'm just not quite understanding what Google hopes to get out of this. It seems like they've given up on creating a good SERP and are all about keeping anyone from figuring out the algo. Which I think would be a really bad move for them as well as us. I just don't get it.

[Leosghost]

On the internight all hats are grey..

[SevenCubed]

I don't think black-hatters have anything to worry about. Spam techniques that this patent is chasing is totally different from black hat tactics. I've watched a local company here in my field climb to the #1 spot among 300 others vying for that same spot. They have been building a pyramid of sites for at least 4 years since I've had my eye on them. Their fully transparent public facing site is at the pinnacle. There's no way of picking them apart algorithmically so unfortunately they are untouchable unless someone reports them.

[Elsmarc]

@diberry - A bit off your topic, but...

I would tell them the Gold Rush is over. I have had the same situation, relatives and friends, asking me to tell them how to do it. They think it's easy and anyone can do it.

I tell people this: It's work and it's something you have to learn. I can't teach you. I can discuss things with you, but YOU have to get one the internet and LEARN by reading, reading and more reading.

I have a cousin who came to me 3 years ago wanting to make money off the internet. He had a small site set up on Godaddy and wanted to sell jewelery (essentially as a "representative" for a bigger company sorta like an Amway thing). I went to Google, put in the key word and showed him how many tens (hundreds?) of thousands of hits there were. He then told me Godaddy "promised" him that for US$40 they would give his site "visibility" and was upset he wasn't on page 1 in a Google search. And this guy didn't even know html (not was he interested in learning any thing about it).

Google isn't "out to get" anyone. The biggest problem is there are what - something like 300,000,000 websites on world wide web. Something like 20% to 30% are "active". That, my friend, is competition.

Just something to think about when a friend or relative who knows nothing about the internet, and making money from it, asks for your help.

[diberry]

They understand it's hard work - they've watched me put in 80+ hours a week for 10 years. The problem is, for example, when I started out in this white hats were advising you submit sites to directories. Now that's considered entirely manipulative - back then, they considered it good marketing. (See also: buying links. It was done online, BEFORE Google existed, to build traffic, not pagerank. Now it's considered a bad thing according to Google, but big media sites get away with it because it has ALWAYS been done in print media - people would buy citations or mentions in news reports.)

There isn't an amount of reading you can actually do to avoid making honest mistakes. The line between SEO and good marketing is fuzzy, and it's not like Google has a specific good/bad list of marketing techniques. You can easily get bad advice that READS like really good quality advice online and make mistakes.

I never said Google is out to get anyone. I said it SEEMS like they're more focused on keeping the algo beyond anyone's understanding than in providing good SERPs, and yet that would be a stupid business move, so I can't figure out what they're playing at.

[TypicalSurfer]

so I can't figure out what they're playing at.

They are an advertising company. What they deem "spam" is competition to their ad model, it has little to do with "science".

[RegDCP]

buying links. It was done online, BEFORE Google existed, to build traffic, not pagerank. Now it's considered a bad thing according to Google,

It is only a bad thing if it is a hidden purchase.
It the site is openly selling links (advertising space) it is not.

[TypicalSurfer]

In a science environment, accurate categorization of documents is helpful, in google-advertising-engine-land it's BAD.

[indyank]

It is only a bad thing if it is a hidden purchase.
It the site is openly selling links (advertising space) it is not.

Google would still want people to add "nofollow" to those links and I feel it is reasonable because they are making it very clear that their algos cannot make the distinction otherwise.

[Martin Ice Web]

The very big questions is, what is considered as SEO reaction?
If I update my pages every day, because to update my stock info and my site was lowered simultaneously because a other site was moved above mine by transitional movements.
Is the algo now regarding my site because it was moved down?

How big must the changes be onpage to trigger this algo?
Is it a complete new page redesign. Is it new links added, title change, small word changes?

This algo needs a lot of machine power because every single transitional change would trigger the algo to look at the "moved" sites. This is a circle.

[RegDCP]

Google would still want people to add "nofollow" to those links and I feel it is reasonable because they are making it very clear that their algos cannot make the distinction otherwise.

One would think so but a look at Yellow Pages and Kelly Search source code does not show the nofollow anywhere.

[indyank]

One would think so but a look at Yellow Pages and Kelly Search source code does not show the nofollow anywhere.

surely you can find a lot of exceptions in every niche. ;)

[RegDCP]

surely you can find a lot of exceptions in every niche. ;)

Those are just 2 of the major directory sites.

Matt Cutts told us a short while ago that nofollow links are but a very low single digit percentage of all links.
This would seem to indicate that the nofollow parameter is not a huge success in Google's eyes.

I'm going to go out on a limb and say that Google has other ways to determine if a link should or should not get "link love" for PR.

I think that relevance would be the metric.
If a page is relevant and adds to the information silo, a nofollow markup would take away form the total.

If a page is not relevant and has a dofollow tag, it is discounted.