Slashdot Mirror
Google's Site Ranking Secrets
vivin writes "Ever wonder how Google's site ranking works? Wonder no more. Google recently filed United States Patent Application 20050071741 on March 31, 2005. This patent reveals a great deal of information about Google's site ranking algorithm and makes very good reading. For example, one of the criteria that they use is the number of years that your site has been registered. If your site has been registered for less than a year, then it counts against you. A site registered for a longer period of time means that the owner is probably serious about the site, and the site is probably legitimate. Google's Site Ranking algorithms reveal how hard they are making it for spam sites to get listed (on Google). This information will also make it easier for you to make sure that you get listed well in Google."
Note that there is no guarantee that Google uses everything in the patent or that they don't use other methods not described in any of their other patents.
or conversely how spam website can get higher :)
I prefer the official Google explanation:
http://www.google.com/technology/pigeonrank.html
Something is happening here but you don't know what it is, do you, Mr Jones.
'Google record the discovery of a link and link changes over time. The speed at which a site gains links and the link life span.' I fail to see how this would be helpful--if something's new and briefly popular, you only want to give it a high rank for a brief period and forget it once people stop linking. But if something's new and popular for a duration, you want to keep it well ranked.
Could someone explain how other crap search engines are getting high rankings in Google search?
Sometimes when I search for something specific, I get a bunch of useless links that have results of other "search engines" that invariably show something similar to "0 results for your search terms 'sheep+barn+slashbot+erotica'"
How do these sites get on the first page of Google results?
They've thrown every technique they could have thought of into the patent purely as a defensive mechanism to prevent other major engines from patenting them. Some of the techniques are thrown in as defensive FUD to prevent newbies from using them.
.. what do I know ..
Some of these techniques are just plain old bizzare and might be way too difficult to approach algorithmically.
Oh well
Argh... quit trying to game the system! If you read the article, it's entirely from the perspective of someone trying to corrupt the rankings for financial gain. Here's an idea: make good, useful web pages, rather then spending all your time an energy creating these BS link farms. The SEO world is the modern day equivilent of snake-oil salesmen.
I want to change. Please help me--I don't think I can do it on my own.
I always suspected this... When we've started our business, we used the domain www.interakt.ro (we're from Romania). However, because we sell software tools mostly to the USA and Western Europe, we've decided to go to www.interaktonline.com.
:D
Instantly, our ranking went from number one (for "Dreamweaver Php" for example, we were number one there instead of Macromedia itself a long time), to page 10.
Now, we're working hard to promote our site, we have links all over the place, but still our site don't get up again to page 1 (search for "dreamweaver extensions" - we have to pay to get our site in the first position). I even thought that they do this on purpose for us to continue to pay on Google Ads
Probably they say it too in the patent, but the best ranking tool is to use the right "title" tag in your pages. It's invaluable how well this scores as compared to the page content.
Alexandru
Nothing in the patent nullifies my pagerank defeating technique - put lots of links to my homepage in slashdot posts modded to +5 funny!
sheep.horse - does not contain information on sheep or horses.
The article dedicates only a couple of paragraphs to PageRank, the main algorithm that Google uses, and about 2.5 pages to the rest. If anyone wants to know more about PageRank, here's Page and Brin's original paper: http://www-db.stanford.edu/~backrub/google.html
I hate the one hundred and twenty character limit for signatures with an all-enveloping, all-destroying, incredible pass
So that explains a lot. What a crappy article, I wonder if the submitter is the same as the Author?
This space is intentionally staring blankly at you
Umm, you spelled 'genius' wrong, genius.
Each claim in the patent can be invalidated independantly. Most patents start off with an all encompasing claim 1 that would almost certainly get invalidated if it went to court, and define subsequent claims more narrowly, often in terms of the preceding claims.
one of the criteria that they use is the number of years that your site has been registered
is not the same thing as (from the article):
How many years did you register your domain name for?
Though the summary suggests that older sites do better, the article is stating that, in order to improve one's Google ranking, domain owners should purchase longer domain registrations.
And another small note... Initially, we have used an HTTP 403 (Permanently moved) from interakt.ro to interaktonline.com. This caused us a MASSIVE degradation of our position, so right now we just do a transparent redirect from interakt.ro to interaktonline.com, without the Permanently moved headers (and this is how we've reached page 2...)
Alexandru
Some of the tatics detailed in the article require a spyware (google toolbar?). It is not possible for google to know when you came back to the search engine from your site, or another one (unless you have a link in your site to google). It also impossible for google to know if you have a bookmark.
Google does have a click-through engine attached to the results, but many people find this in adition to the single identifier cookie that googles push into you abusive already.
We all thing google is doing a good job, and it did managed to incorporate adds and an add service that is well accepted by the people. (I wonder why people still think it is a good idea to make blinking and noisy flash adds?) The point is how much we trust google? I personaly don't mind very much the click through, but do not accept the cookie and will not install a toolbar.
[]'s Victor Bogado da Silva Lins
^[:wq
The story is so old I can't believe it made it to slashdot.
Some more on info the subject:
1. U.S. Patent Application - it's best to read what's exactly been patented.
2. interesting discussion on webmasterworld
Personally I think that while some of the stuff is interesting, most of it is made up rather to confuse SEOs (google doesn't quite like them, you know that, right?). Before that, they had couple factors to think about and work on. Now, there's a shitload of stuff that just makes their work harder. Also, more factors influencing SERPS means it's much, much harder to make a trial-an-error research on what works well and what doesn't.
This type of spam (showing a page to the crawler and another to the user) is called cloaking. Cloakers have anticipated this sort of move and can detect a search engine's crawler by not just the user agent but also the IP address range it comes from and other heuristics. In order to beat them, search engines would have to crawl from unpredictable IP addresses and behave like regular users.
A while back I proposed a distributed approach like this in the Nutch mailing list. The problem is that it would be hard to implement and it may not be worth the effort, since there are cheaper ways to fight spam.
See charts for twitter trends on Trendistic
Just look at the patent application yourself.
I haven't read the whole thing, but just having taken a quick look at it, I have to agree with the posters who said that Google purposefully tried to cover any conceivable technique to index and rank pages. The application discusses multiple implementations of the various techniques that could be used to rank a page. Therefore analysis of the patent application is probably of limited utility for those trying to game PageRank (which was certainly a factor that Google's very competent IP lawyers considered before prosecuting the patent).
For those who are worried that Google is doing evil with this patent application, given the breadth of the patent and the fact that it discusses a plethora of techniques which Google may or may not be using, I will be surprised to see Google try to use this patent (or be able to use this patent) to push another search engine out of the market. More likely, I think, is that this will constitute prior art to enable Google to withstand challenges from other patent applicants for infringement. Of course, if you know anything about PageRank, you know that it was getting published in Scientific American long before Google was the dominant search engine. So this patent application is probably more to prevent allegations that Google infringed by adding on all the other checks and balances to the original PageRank technology to discourage spam sites.
Moiche
No, its not that simple. Lets say I have a small business, I sell garden tools, lawnmowers,etc, in a certain region. And yet I do a search on google for garden tools + region, I am nowhere to be found. What do I do? I optimise the hell out of my site, caking it with region name + garden tools information, and I set up a links exchange program, getting in links left right and centre from related sites. This is SEO, and it will only affect people that enter a search for "garden" "tools" "my region". In other words, those that actually want to find my site.
Theres a distinction between SEO and spamming; if I was to optimise for a garden tools site and set up a poker site there, that would be spamming.
What he can't kill, he has sex on. Trent.
If any of you have worked in a small online shops you know what a fucking holy war this is between marketing and pretty much everyone else. I specifically remember saying at one point, "Do we have to make ALL of the money RIGHT NOW?"
Good for Google for coming forward and telling peole they won't be a part of that slimy shit.
Bad for Google for saying all of this to drive up prices on their AdWord sales.
s'wut i sed.
Remember, this is a patent which requires no working model. In other words, this could be how Google *envisions* their search working as much as it indentifies any of the things it does do.