View Full Version : Phi - "Google and the Golden Ratio"
rustybrick
02-14-2005, 11:07 AM
I wanted to open up this discussion to our PhDs and those who love to talk algorithms here.
Basically there is some new speculation, that started up mid last week, with an entry over at DG's Desk named Phi, Pi, Fo Fum (http://www.dgsdesk.com/index.php?p=29), which had a short message that read; "When writing link text, remember 1.618 Illuminating, really."
ThreadWatch then picked this up recently and a larger discussion started to brew, under the title of Google and the Golden Ratio (http://www.threadwatch.org/node/1452).
Basically, the speculation is that a ratio of your anchor text links needs to be 1.618 of the same anchor text and the remaining the similar words of anchor text (~).
Ok, that is all I have to say at this point. PhDs and others? :)
I'm not denying the elegance of phi, or it's mystical properties. But it's just sooooooo trendy right now. My guess is that it's partly due to the popularity of the DaVinci Code, in which phi plays an important role. Poor old pi can't even get a soundbite these days. It's all phi phi phi. Currently, phi is to math what Texas Hold Em is to poker and ponchos are to women's fashions. Just my 2 cents! :rolleyes:
randfish
02-14-2005, 01:35 PM
a ratio of your anchor text links needs to be 1.618 of the same anchor text and the remaining the similar words of anchor textI find it exceptionally unlikely that this is the case. What is possible is that Google has noticed 1.618 cropping up in their examinations of certain link structures, page structures, etc. - But, as a requirement for top rankings? That's too far-fetched for me.
Incubator
02-14-2005, 02:38 PM
I agree with Ranfish. That type of equation to Anchor text in links,seems to be pulling at it abit. In the case of the latest changes (IMHO) seems to be more the data center catching up with directions in placement through the (SEPRS)
It's still a bit early to gauge these changes as an ongoing inclusion (top ranking) criteria
2 cents
Cheers
WC
Pushe
02-16-2005, 11:43 AM
Hi there,
I can`t get a thing from the written above. Can you please explain it in more common english (it`s not my mother tongue), and if you can please give a theotetical example.
10x
Pushe
Phi. The hermetic rule of the golden mean. This goes back as far as the egyptians, the romans, all those dudes back then who used it supposedly for their architecture. I know many speculate that the pyraminds were built based on this. There's a very close relationship between Pi and Phi. Leonardo was a fan as well, and people think it can be found in the human body as well. Some say its devine and the work of god.
The coolest thing about Phi? When substracted from its square it results in 1. Beautiful.
The other thing? The Golden Number is an infinite series of 1.
I thought Google were into 0's.
Well anyway, sure you can find the equation in any maths book. Have a look if you are curious.
Google doing some special thing especially with this equation? I won't list them but there are plenty to choose from.
oilman
02-16-2005, 02:07 PM
I certainly can't see them using it to any real degree in anchor text. It would be way to easy to tank your competitors by buying a few links to throw the ratio out of balance. If they really do want to use a cool number to define things it's more likely we'll see a move back to on-page factors. In which case a good cloaking script and some basic math and we'll be cranking out thousands of perfect pages over night...
randfish
02-16-2005, 02:30 PM
Xan,
It's not nice to be so mysterious :)
How about a hint or a nudge in the direction of where we should go to make some educated guesses about this?
Incubator
02-16-2005, 05:36 PM
I certainly can't see them using it to any real degree in anchor text. It would be way to easy to tank your competitors by buying a few links to throw the ratio out of balance. If they really do want to use a cool number to define things it's more likely we'll see a move back to on-page factors. In which case a good cloaking script and some basic math and we'll be cranking out thousands of perfect pages over night...
Thanks Oilman,i think you nailed it
Cheers
WC
I can't tell you anything more, but there are a lot of resources around for you to look at.
The use of mathematical, logical, and formal methods, has been the basis of Information Retrieval research for long time
Vector, boolean and probabilistic models of information retrieval are used often enough. A mathematical exploration of content with Diophantine sets behind relevance feedback has been looked into (to show you how diverse the field is), as well a ton of other pure mathematical concepts applied to linguistics because it is seriously complex in nature.
I don't have a link but look at sandor dominich's "formal foundation of information retrieval"
This is simply to put things into perspective:
Complexity theory is the study of the resources required by algorithms, usually space and time, and is applied to computational linguistics.
Universal base hypothesis (Peters and Ritchie 1969) is used, according to which all natural languages can be generated from the same base grammar.
Zelig Harris 1968:
"Meanings of words are determined to a large extent by their distributional patterns."
Perron-Frobenius theorem used in Markov chain (a discrete-time stochastic process):
They are Square matrices whose entries are all nonnegative and have special properties.
Do a google ckeck for mathematical linguistic models or something like that.
Phi is just one amongst many. Why Phi was chosen? Its the golden equation. It is beautiful, but it is a bit useless in this case. Its the wrong tool for the job.
graywolf
02-17-2005, 12:48 PM
Irregardless if phi is the exact number used or not, having a greater link diversity is a good thing, and will go a long way toward making your site look more "natural" and less "manipulated".
Would you rather have 261 links with "blue widgets" as anchor text, or 100 "blue widgets" links and 161 links semantically similar "blue widgets" links?
From the SE's point of view which do you think looks like an SEO at work?
Marshall Clark
02-18-2005, 04:15 PM
I find it exceptionally unlikely that this is the case. What is possible is that Google has noticed 1.618 cropping up in their examinations of certain link structures, page structures, etc. - But, as a requirement for top rankings? That's too far-fetched for me.
I agree that Google intentionally requiring a 1.618 ratio in link text would be silly (though admirably geeky :)).
But if the 1.618 ratio is instead an immergent property of a naturally linked web population - the idea of emulating this ratio in link development becomes much more interesting.
inthedark
03-20-2005, 08:32 AM
It would seem fitting that as the web becomes more complex it would share traits commonly seen in other large dynamical systems, phi and fractals being the most common 2 (off the top of my head) seen in the large complex system I happen to focus on :-)
related links:
http://www.elliottwave.com/education/SciAmerican/Mandelbrot_Article2.htm
http://www.acadjournal.com/2002/v7/part3/p1/
(1+√n½) / 2
Phi. My favorite number, and that of many many others too.
Yes, you can find it everywhere, if you look.
"It would seem fitting that as the web becomes more complex it would share traits commonly seen in other large dynamical systems"
I would just replace "would" by "could".
Again, despite a great fascination for the number, I don't like to get carried away.
Here's the flip-saide:
George Markowsky's paper, "Misconceptions About the Golden Ratio,"
mathematics association of america (http://www.maa.org/devlin/devlin_06_04.html)