New Google Patent Details Many Google Techniques

[msgraph]

Probably one of the best bits of information released by them in a patent.

Large number of inventors listed on here, even Matt Cutts that guy that attends those SE conferences. Explains a bit about what is already known through experience as well as comments made by search engine representatives.

Example:

[0039] Consider the example of a document with an inception date of yesterday that is referenced by 10 back links. This document may be scored higher by search engine 125 than a document with an inception date of 10 years ago that is referenced by 100 back links because the rate of link growth for the former is relatively higher than the latter. While a spiky rate of growth in the number of back links may be a factor used by search engine 125 to score documents, it may also signal an attempt to spam search engine 125. Accordingly, in this situation, search engine 125 may actually lower the score of a document(s) to reduce the effect of spamming.

USPTO version [appft1.uspto.gov]

< Note: the USPTO has at times either moved or removed this
patent. If that happens again, here's an online back-up copy:
Information retrieval based on historical data [webmasterwoman.com]>

[edited by: tedster at 3:04 am (utc) on April 10, 2008]

[is300]

Looks like we can finally replace Brett's article on how to have success with google in 12 months.

[webmasterworld.com...]

[Rollo]

Well, one thing is certain... everything you do will go down on your permanent Google record and it will be held against you sooner or later.

[Powdork]

- going to run off and renew my domains for ten years now; the weirdo seo tactic of 2005

So if I happen to use one of the largest hosting companies in the world and my domain registration is updated yearly because they only offer quartely payment options, am I screwed? Should I switch companies for G?
Does anyone else think this patent is Google's early April Fools Joke on SEO's?

ed. spelling

[landmark]

Forgive my ignorance of patent law, but I'm shocked that such a large and vague collection of ideas can be patented: ideas that almost anyone in the industry could have come up with; in fact many ideas that are already commonly accepted or discussed among the SEO community.

Does this mean that other search engines are not going to be allowed to take into account for example the age of a document when computing their results?

[GlynMusica]

Proof that PPC is the only method that gives you any certainty.

Also,
"[0039] Consider the example of a document with an inception date of yesterday that is referenced by 10 back links. This document may be scored higher by search engine 125 than a document with an inception date of 10 years ago that is referenced by 100 back links because the rate of link growth for the former is relatively higher than the latter. While a spiky rate of growth in the number of back links may be a factor used by search engine 125 to score documents, it may also signal an attempt to spam search engine 125. Accordingly, in this situation, search engine 125 may actually lower the score of a document(s) to reduce the effect of spamming. "

So if a page is new and goes from 1-10 backlinks overnight then it's popular, but it could be spam as well. Gee, that's really clever. Don't Patents make you laugh!

[claus]

Still haven't read the thing but jdMorgan is right here - it's good to cast a wide net. Apart from safeguarding you against competitors working around the patent (and getting to patent stuff you really haven't invented yet), it's also a smokescreen so that it gets harder to see what the patent is really about (as you have to publish the whole thing).

And, patent offices do grant patents to stuff that they really shouldn't even think about - just see this well known illustration [webshop.ffii.org] on European patents regarding internet shops. Or, remember the patents on one-click purchase, the hyperlink, and so forth.

I think i'll read it anyway - methinks it's just cover-up for the thing known as "the sandbox" with far too many specifics listed.

---

Random speculation:

The interesting thing seems to be that we now have five types of ranking, or "rating" as it might be:

• Standard (query independent) PageRank - for pages
  
• Some amount of "site wide" (query independent) rating, or "reputation"
  
• A history based (query independent) rating - either for pages, or for sites, or both
  
• Topical (query dependent or query-independent) rating/ranking, which may include LocalRank, Hilltop, LSI, "bad neighborhoods" etc.
  
• Query dependent ranking, which may include the above as well as the famous "99 other factors".
  

This is a new way to think about it for me - normally when i've considered eg. LocalRank i would have considered that as the end ranking (which also confuses me all the time, as sometimes i think it's there, sometimes it's definitely not), but now it's just an ingredient or a knob that can be turned. The same goes for the rest.

Of course it's all pure speculation. So is this: The better your overall "rating" (top points) the less it's weighed into the "ranking" (bottom point).

Whatcha think? Is it too far-fetched?

[trillianjedi]

Er, excuse my possible lack of understanding of US patent filing, but where does it say that this is Googles patent, or anything to do with Google Inc?

I see it's been filed by a group of individuals some of who are (or are ex-) Google employees, but that's the only link I can actually make.

TJ

[shri]

>> I see it's been filed by a group of individuals some of who are (or are ex-) Google employees, but that's the only link I can actually make.

Corporations usually own the patents of the employees.

This is a very broad statement and there are exclusions. If you thought about it at work and its related to your employers business, the chances are you have assigned the rights to it, to your employer.

[sem4u]

Yes a lot to digest but nothing really new for readers of this forum. It does help to tie up a lot of speculation though.

One thing I did notice. Take a look at the date the patent was filed: December 31, 2003

[trillianjedi]

Corporations usually own the patents of the employees.

The law is essentially the same in the UK, if on the employers time etc.

But the corporation would file the patent or trademark in this situation.

It's not clear here as it's been filed in their personal names. There is no reference to Google at all.

This thread seems to be discussing this as some new fangled algo weighting and SERPS ranking created by google, owned by google and used by google.

While there are elements of the patent application which would fit in with the way Google Search behaves, I'm not actually convinced it's anything to do with Google Inc.

There also seem to be references in the thread to this being another "April 1st" joke, but the document I'm looking at was filed over a year ago, in December 2003.

Are we all looking at the same thing? What am I missing?

TJ

[Import Export]

Run some cross searches using the individuals names and the company filling for them. There are other similar documents that show with older dates just like this one... I don't know what data people are using to support the April 1 theory, but doesn't seem like thats the case to me... Albeit this one gets pretty broad, but the others I've seen in the past weren't too far from this. I dunno, then again, maybe they were?

[Import Export]

Ok.. I have printouts of all previous patents G has filed in the past sitting in a file in front of me.

Every single one of them assigns the patents to:
Google, Inc. (Mountain View, Ca)

[trillianjedi]

Every single one of them assigns the patents to:
Google, Inc. (Mountain View, Ca)

Assigned?

So are they filed by individuals (as per the original Brin/Page patent) and subsequently assigned to Google Inc?

Are you looking at granted patents or just filed patents?

[elguiri]

I think the date is irrelevant. These seem to be lots of things G has been working on, or, as often as not, pondering, for some time. They've almost all been discussed here. The curious thing is that someone penned them and published them.

Looking at the collection as a whole, it seems that the guy with the keys the sweet cupboard is a bit of loon.

In conclusion, the usual stuff of this industry - information, disinformation, smoke and mirrrors.

Nothing else for it. Back to testing testing testing...

(PS. Wouldn't it be "funny" if the mystery element to Florida - why some went up and others disappeared - was the domain renewal date? What a strike for relevancy!)

[zgb999]

"it's also a smokescreen so that it gets harder to see what the patent is really about"
IMHO Claus is right again in saying so.

In a way it also is a nice way to make a wish list public and see what people all over the world think about the different ideas. Like that you get input for those things you really want to implement.

As the patent was filed on December 31, 2003 we certainly see a lot of things Google has been working on and is still working on.

As not all country domains can be paid for more than 1 year I would also asume this would be one of many signs for potential spammers (if implemented at all) and not a penalty as such.

[elguiri]

Here's another thought. Could it be that the patent is in the name of employees rather than G so that no one would notice? If the patent date is Dec 2003 it seems this has worked up to a point.

That it's been found just shows the power of "Information retrieval based on historical data"!

[msgraph]

Every single one of them assigns the patents to:
Google, Inc. (Mountain View, Ca)

Many patents that are in the application process don't carry the assigned name until the patent is issued. Yahoo usually has their name when they are in application form however Google usually doesn't.

[Import Export]

I have $10 that says this was done by godaddy..

[shri]

>> I have $10 that says this was done by godaddy..

Err .. remember which company became a registrar recently. ;)

[Kirby]

I dont have patents, but I have trademarks. The application process is similar with regard to casting a wide net. Examining attorneys dont tend to approve applications as written. You always throw stuff into the app just so the attorney has something to reject. It then gets whittled down to the finished product. Sort of like ice sculpting, the final version that is approved will be different than the original application, but pretty close to what you intended when you started.

This thread seems to be discussing this as some new fangled algo weighting and SERPS ranking created by google, owned by google and used by google.

Its not new, but merely just a list of potential ingredients - the infamous (or perhaps mythical) 100 or so factors reported to make up the algo. Google always adds smoke for flavor.

[zomega42]

I'm no expert on patent law, but I thought you were not allowed to patent "wishful thinking". You can only patent things that you have actually developed and actually implemented and made available to the public, or will do so soon.

Am I wrong about this? If not, it means ever one of these factors is actually in use in the ranking algo, not just a flavor of what's to come.

[Kirby]

It means that they reserve the right to use it in their algo.

[blaze]

For about 80$ you can apply to patent making planets out of cheese.

They don't get reviewed.

I think the date is very significant.

This is very very typical of Google's sense of humor and strategy.

I think it's probable that it was meant to be released and meant to be released in this way. Released in a way that creates doubt in the mind of its readers.

However, Gmail was also released on April Fools, so I wouldn't be quick to discount it entirely.

Also, for anyone who has worked with neural nets / pattern recognition, you would realise that the various datapoints which are enumerated in this patent are necessary for creating signal strength on neurons and genetic algorithms.

The length of a domain subscription is a signal strength. Which used in conjunction with a number of other neurons it can result in a meaningful description of the appropiate way to rank the domain..

As a standalone factor, obviously it's completely useless. When used in cooperation with 25 other bullet points it could probably be quite effective.

I know I'm going to renew my domain for 5 years next time, that's for sure.

[walkman]

"I know I'm going to renew my domain for 5 years next time, that's for sure"

as all spammers started renewing the domain name for 10 year terms, Google quickly noted, and held that as a negative. ;)

[walkman]

"It means that they reserve the right to use it in their algo."

ummm...you can't reserve things. You either have "invented" them or haven't.

[mayor]

A patent gives them ownership of the invention. Nothing says they have to market the invention.

Equally important, having ownership of the invention prevents the owner's competitors from owning it, and therefore prevents their competitors from marketing it without being granted the right by the owner.

By the way, keep an eye open for international patents by Google. I'm not up to date on international patent laws but I suspect there are many countries in the world that don't recognize US Patents and the Web is indeed worldwide. Complications like this often led companies to protect their technology with trade secrets rather than patents.

[mfishy]

Any legal experts have any idea how any of this could be enforced? It's not like the way other SE's score sites is public info.

FWIW - My bro is a patent attorney that works with fortune 500 tech companies and he seems to think this is pretty much a joke...

[trillianjedi]

I think the date is very significant.

How is December 2003 significant? What happend at that time?

TJ

[Rollo]

I've read a few posts suggesting that this might be some disinformation campaign a la black propaganda aimed at webmater/spammers. I seriously doubt this. Wouldn't a single post from Google Guy have been enough in this respect?

I think they're trying to get out in front of their competators so when MSN tries to incorporate such gems as domian registry length into their algo, Google can cry fowl. I think they are totally serious about experimenting with these ideas and incorporating ones that work (though many don't seem especially well reasoned).

[Watcher of the Skies]

What "boredguru" said in Message #68.....