Forum Moderators: Robert Charlton & goodroi
I've checked other places and they say the old PageRank but maybe we're on the verge of another PageRank change?
Anyone notice this?
Not sure I agree with all of his views but the concept is at: (no www.).... [clsc.net...]
[edited by: tedster at 5:18 am (utc) on Jan. 13, 2007]
[edit reason] make link clickable [/edit]
The local rank or score is based on the top 1000 or so based on PR.
Are they talking about the top 1000 on the Internet or top 1000 in a given search or do they have some other way to dividing up sites by topic and selecting the top 1000?
[edited by: encyclo at 11:31 pm (utc) on Jan. 13, 2007]
[edit reason] fixed link [/edit]
One of my sites has a PR5 on its non-www url, and a PR4 on the www version. I have it set in WMT to show everything on the www, so I'm not even sure why the non-www would be still be in the index.
Subsia, this is a topic that has been thoroughly discussed here on WebmasterWorld.
Setting preference in Webmaster Tools is not enough. You will need to decide which version of the url you like and apply it sitewide.
Most people link to me using <http://www.example.com/>, so I chose to redirect every request to the www version of my homepage. Visit the Apache forum to find the code needed to do this redirect in htaccess.
[edited by: tedster at 5:15 am (utc) on Jan. 13, 2007]
[edit reason] use example.com [/edit]
My understanding is that it is for a the first 1000 of a search query. Joe public does a search for "widgets with blue stripes" and a result is based on standard pr calculated before any search is done. The local rank pr is then calculated as if these were the only sites on the web. The clever bit is a theme has been established between all these sites via the initial search, so now the local rank pr is a totally theme based pr value, in this case for "widgets with blue stripes". This is all done on the fly.... how clever is that!
What the new patent implies is a ranking, then a reranking, then a weighting, and then a display. It goes something like this:The usual pagerank algo (or another suitable method) finds the top ranking (eg.) 1000 pages. The term for this is: the OldScore
...
traditional pagerank evaluation, as described in the early papers by Brin & Page, is still basic to that newer form of evaluation.
This is all done on the fly.... how clever is that!
For technical reasons it is impossible to calculate pagerank on the fly, even with the earth-simulator. All you can do on the fly, is query that value from a database, which contains the pagerank-values precalculated off the fly in a fairly long and cpu-consuming process.
My understanding of the new infrastucture is this: It very flexibly enables google to let any search query flow through a cascade of von-Neumann-Environments, the output of which is the actual result-page, which may change from day to day. One of these von-Neumann-Environments is presumably a (set of) database(s) containing the pagerank-values (not necessarily the first in this cascade). Another one may be an algo calculating 'local-rank.' Another one is probably a database containing values for LSI-vector-spaces precalculated from the cached pages. Another one might be the "similarity engine" as described in this patent [patft.uspto.gov]. I have no idea about the true structure of this cascade (or is it a network?), and I use "von-Neumann-environment" as a working title for pieces much more complicated than what we usually call "filters". I'd speculate that my search query is passed through quite a number of single machines at the plex, before the result-page is finally sent to my browser.
And it is also quite clear, that the database containing the PR-values isn't only a single machine in each datacenter, but comprises a respectable cluster with redundant copies, in order to cope with that massive bandwidth. What we experience as a pagerank-update is the process of shovelling the values from the "off the fly-" calculation-machine(s) to the many PCs, which answer the "on-the-fly-" search-queries. Datacenter by Datacenter all over the earth.
Until google disproves me with CONSISTENT values in my toolbar and shows a much smaller "page-rank-not-yet-assigned-" bar in my webmaster central console, I continue to believe that there is something either utterly wrong or completely new with the pagerank-values shown.
But I don't want to drive this thread OT. Tedster plz shut me up, open another one, or point me towards an earlier discussion, where my naive speculations have been falsified.
On another quite new site (September 2006), backlinks are updated but I do not see a PR change, although it has been predicted to become a PR4-5.
But the output of this algo is not that tiny little bar then, right? As you, tedster confirmed, it refers to a search-query, not the overall measurement of importance of an url, which is condensed in the toolbar.
One childish question regarding the emperor's clothes and the exact role of trademarks and patents: pagerank is a registered trademark. if I remember correctly, it is also refered to as such in some of the patents, which I tried to understand. Does this imply that the way google calculates pagerank-value will always remain the same? Is the trademark "pagerank" so closely connected to the "method" of its calculation as described in the patents? Is the word "pagerank" always used with the tm-sign in all the later patents?
I mean: If I order a glass of Coca-Cola, I want a Coca Cola, and even if its ingredients are still kept secret, I can rely on getting something identical no matter whether I'm in Tokio or New York (And I will recognize someone serving me a Pepsi or Club-Cola!). This effect of re-cognition is the essence of trademarks and branding, isn't it? So, no matter what the toolbar temporarily shows, as a stock-owner, searcher or adwords-customer I would expect a minimum of continuity regarding the trademarks I'm planning to invest in?
Without google violating this principle, it would still be google's choice to decide, if some links and urls are deleted from the index BEFORE that iterative core-algo starts. That would explain all surprises sufficiently. But the pagerank-algo istself, taken literally as a trademark and patented method:
Is pagerank still pagerank?
A trademark and a patent are two entirely different animals.
A trademark can evolve in Design, is the logo design, the word, the label and device of Coca-Cola the same it was way back when it started. Probably not. That's a trademark.
Is the Coca-Cola we drink today the same as it was in the beginning? That would be more patent related.
Is the taste of Coca-Cola the same everywhere, not sure it is identical everywhere but very similar indeed. - That would be more patent related.
A Trademark does evolve to meet market conditions, a Trademark is simply a representation or words to identify a particular product and associate it with a company and a level of quality be it good or bad.
So in essence, Page Rank can change behind the scenes as much as Google likes, it's their patent. The trademark, the physical representation of Pagerank, be it the green bar or the words will evolve when it become outdated or stale as do all Trademarks to appeal to people's eyes!
[edited by: Pico_Train at 5:31 pm (utc) on Jan. 14, 2007]
Anybody have an idea how old this PR is? And could the PR have been increased since then?
Added: The PR doesn't seem to be constant; sometimes it's there, sometimes there's only a whitespace.
[edited by: OutdoorMan at 6:00 pm (utc) on Jan. 14, 2007]
I mean: If I order a glass of Coca-Cola, I want a Coca Cola, and even if its ingredients are still kept secret, I can rely on getting something identical no matter whether I'm in Tokio or New York (And I will recognize someone serving me a Pepsi or Club-Cola!).
Actually, that's really not true. There are even regional differences in the formulas used for these brand within the same country, and in particular two things that vary a lot are sweetness and degree of carbonation. Bottling plants are regional for these products, so it makes it easy to provide products that meet the preferences of the local populace.
I'm afraid your ability to discriminate these differences is perhaps a lot less effective than you think.
Anyway, the patent issue you bring up is an interesting one. I don't know the answer, and what happens if a company has a patent, but over a period of years modifies it's product.
Which dc's are you checking? I see my three months old subdomains, still at PR 0...
I'm not actually checking DC's, I use the Google TB.
But if I check my Home Page PR in a PR check tool on the web, I see these results:
DC: 64.233.167.133 Toolbar PageRank: (1/10) Cached: Yes
DC: 64.233.187.115 Toolbar PageRank: (1/10) Cached: Yes
DC: 66.102.1.80 Toolbar PageRank: (1/10) Cached: Yes
DC: 66.102.7.107 Toolbar PageRank: (1/10) Cached: Yes
DC: 66.102.9.99 Toolbar PageRank: (1/10) Cached: Yes
DC: 66.102.11.189 Toolbar PageRank: (1/10) Cached: Yes
DC: 66.249.81.133 Toolbar PageRank: (1/10) Cached: Yes
DC: 66.249.83.84 Toolbar PageRank: (1/10) Cached: Yes
DC: 66.249.89.19 Toolbar PageRank: (1/10) Cached: Yes
DC: 66.249.89.104 Toolbar PageRank: (1/10) Cached: Yes
DC: 66.249.91.18 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.203.115 Toolbar PageRank: (0/10) Cached: Yes
DC: 72.14.205.81 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.207.100 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.211.44 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.215.81 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.215.83 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.217.101 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.219.104 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.221.44 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.223.80 Toolbar PageRank: (0/10) Cached: Yes
DC: 72.14.247.99 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.247.184 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.253.83 Toolbar PageRank: (1/10) Cached: Yes
DC: 72.14.255.84 Toolbar PageRank: (1/10) Cached: Yes
DC: 209.85.129.44 Toolbar PageRank: (1/10) Cached: Yes
DC: 209.85.143.100 Toolbar PageRank: (1/10) Cached: Yes
DC: 216.239.51.84 Toolbar PageRank: (1/10) Cached: Yes
DC: 216.239.53.147 Toolbar PageRank: (1/10) Cached: Yes
DC: 216.239.59.99 Toolbar PageRank: (1/10) Cached: Yes