Forum Moderators: open
Any ideas?
Also, does anyone think that we are becoming too obsessed with page rank? How is page rank defined exactly?
Is it possible that there is more than one Google algo? The secret to ranking well in Google is still a mystery - it appears just to be a mish mash of things that sometimes count and sometimes don't.
interesting find, never seen that page before.
How impatient do we as webmasters get for a G update? Then compare that to the impatience of link rot and worldwide combined spam :)
>>100 variables
Who wants to start listing them then ;)
Hypertext analysis is nothing but checking HTML tags. Additionally, it is "Then for each query". PR is calculated long before any query.
>>100 variablesWho wants to start listing them then :)
let's give it a try. i'd start with (unordered list!):
page rank
keyword in title
keyword in H1
keyword in URL
keyword density (total word count being considered)
keyword in title
keyword in links to page (anchor text)
keyword in bold / strong, etc.
keyword in other parts (full text, alt, title, meta description)
distance between keywords in all appearances (if search for 2+ keywords)
keyword position in all appearances (how far from top)
URL length
more guessing (that's how i'd do it):
======================================
absence of competitive keywords other than search term
all i can think of for now, feel free to complete.
muesli
Backing up to the original title a bit, you could only assume that a computer would be able to determine any sort of "theme" from both a pagerank score or the score of things on the page.
mitzyb....if you wanted to try and prove what you want to prove - why not get some webspace at geocities and use some googlewhacking words on the page and send a couple of unrelated links pointing towards it.
When (if) it gets in the index- it shouldnt rank well for those googlewhacking words you used. Otherwise it might be more likely that themes are not as important as you might have wanted.
/oops added
Make sure they are not literally "googlewhacks" otherwise they will be the only SERP returned and that wouldn't be very beneficial :)
In my opinion, Google ranking is primarily affected by keywords in Title Tag, keywords H1 or H2 Tag at top of page, and keywords in inbound links pointing at that page. Other onpage factors mentioned above are also definitely part of the algo.
I am an advocate of themes, and I think Google will eventually use them. But this is just a guess on anyone's part.
startup, I agree that there is more than one algo. I have been noticing during the last post-update month that my pages would cycle through 3 different ranking positions with three different total result counts for the search. This went on all month, and continues to go on today.
It's one seansible answer to google's page rank spam problems, and makes sense, plus it is already used in other engines.
Things move fast in the search engine world. The smart guys must be basing their design and optimization on what search engines will like in 3, 6, and 12 and maybe more months down the track than what they liked the month before last, which is the latest evidence you ever have.
* How often new inbound links to a site are established, or
* The rate of change of inbound links, or
* How many new links in a "class" (such as PR5) are added since in a period of time, like since the prior update ....
Things like that. It occurs to me that I have never seen a "rate of change" where linking is concerned in any discussion here, or anywhere, for that matter.
Non-really-related question:
How does Google know when a page "changed" since last time googlebot crawled it? Maybe there is a way to compute the "similarity" between two pages, and Google considers that 2 pages are different when the similarity changed more than a given threshold...
In unix hosts , an simple " touch *.* " will change the modification date of all files.
Google does use an algorithm similar to the one you describe. (i.e. "The #N add it should be getting X% click-thru rate for this keyword, if not move down.")
So the idea has certain bounced around the halls of the googleplex.
Though I wish it wasn't true, the ranking system does seem to factor in a pages history somehow. Quite accidentally, I made a new site and got it ranked #1 in a month for a $4 word at overture. I still can't explain what happened, other than there must be some kind of recency boost.
Brett has a page about file sizes [searchengineworld.com] also being taken into account. I guess that would make sense from a limited storage point of view.....while things you mention like keyword density being something that would be of dramatic difference in a 5k and 101k page. You could compile that into the list.
BUT. A rate of change can be equated to acceleration. Stomp on the gas in a powerful car, you nearly get tossed into the back seat. Is there a Google equivalent? Let's say 10 PR5 sites link to your site since the last update. We have the effect of those sites on your site as everyone here already knows, but do we also have an additional effect caused by the fact that 10 PR5 sites linked to you where as in the prior update period, only three did? If there is such an effect, I would think it would dissipate in terms of positioning and/or PR with subsequent updates when the rate of "acceleration" slows.
What I imagined was some kind of baseline for how many incoming links a site should have based on how old it is.
If a brand new (to Google) site has 50 incoming links, it's above what Google would expect. If a 3 year old site only has 50 incoming links, it's below average.
Again, no evidence for any of this aside from a new site that did better than I would've thought.
