Google Patent

What is that confidence based on? And what new algorithm?

 

On order to be competitive in the search engine war I cannot see them ignoring the necessity to provide some better results than there are right now. Maybe it won't be major changes but some seem necessary. I could be wrong but I don't think so.

 

That old result I mentioned also ranks prominently in Yahoo and MSN.

 

hmmm. ... some old sites are really good. I just don't think Google will give as much preference to age in the future as they are doing now.

 

If so, that's not good news for DMOZ. Or the Resourceless Zone. Although in the case of the RZ, I'm also surprised they've not been hit with a duplicate content filter - pretty much every thread is the same over there.

Please excuse this shamless example of crossthreading and threadjacking. We now return you to your regularly scheduled program...

 

This sounds very interesting!

It's too bad the Patent Office's web site is undergoing some maintenance right now.

 

http://www.socengine.com/seo/guide/google-historical-data-patent.html there is a copy of most of it.

 

Clarification:

I shouldn't post anything after 3 am

That story is more than 5 years old -- true. And I have seen it sitting in the top 10 and usually the top 5 for that search term on Google and other SEs for at least 3-4 years.

However, the story was picked up and "reprinted" by several other websites (the one in the URL supplied is NOT the newspaper that originally printed the story) and I cannot state that it has been the same web page(s) containing the story over that time period. This, with respect to the arguments about "staleness", it may be an old story but it is not necessarily old for the page currently occupying that #3 spot. For all I know, it may have been added there last week.

 

I don't think that G are saying that staleness is *always* bad and freshness is *always* good.

I believe the point is in reference to the search terms. For some terms (e.g., habits of cave men) stale sites may be preferred and therefore would rank higher, whereas a search for 'space station inhabitants" would prefer fresher sites in rankings.

G goes on to say that for some search terms, staleness vs. freshness may not be much of a factor at all since the 'plurality of documents' within the search results seems to remain somewhat unchanged over time.

I think that means you need to know your industry and terms. If it's a hot industry, you'll need to keep up. If it's a cold industry, and if you constantly update you'll probably stand out and could be penalized at some point for doing so.

 

I suspect that two different matters are getting conflated here.

Yes, in principle there might possibly be some queries for which older pages might be preferable to fresher ones--and indeed, the patent application does say that, and does not say that freshness must always win.

But that is in principle (and in a patent application, which makes all its terms and claims as expansive and universal as possible).

In practical reality, why would any search engine ever prefer older material to newer? Even if the topic is one with, presumably, well-settled knowledge, the betting (especially by an algorithmic machine) would always be that the newer material includes more up-to-date analysis, or data, or interpretations. Why should an article on--to use your example--cave men written in 1997 be superior, on the date alone, to one written last week, instead of vice-versa?

I truly believe that there are no pages whatever that any search engine algo is going to give positive weight to, as opposed to negative, owing to their being old, or otherwise "stale" as the patent application variously defines indicators of staleness.

(The occasional site whose pages haven't changed in three years, seven months, and twelve days but is in the top 20 will very likely be one with a spectrum of other things--probably high-count backlinks--that outweigh sheer age.)

That is not particularly bad news for, say, dmoz, since their listings should be expanding or changing regularly. More worrisome, I think, was this:

For example, if a particular document appears as a hit for a discordant set of queries, this may (though not necessarily) be considered a signal that the document is spam, in which case searchengine 125 may score the document relatively lower.​

True, they cover their ass with a "not necessarily", but one wonders exactly how that would be applied in practice. Many reference pages will at least mention a broad range of topics, the totality of which might well be considered "discordant" to a blind algo.

Well, we don't have the term "Googlery" for nothing.

 

The more complicated Google makes their algo, the easier it is to inadvertantly trip it up and get delisted, or ranked so low your site is dead.

I still have a site where there is a cache on someones 302 url showing my clients home page - supplimental result from Sept04 - my clients site top of Yahoo and MSN for many terms, dead as a duck on Google for every page on site.

There will be a way to sort it out, but havn't figured it out yet.

The above merely as an example. The same applies to like the "same site" issue where too many links between sites causes only one to show for common search terms.

It becomes very hard to isolate exactly why your site has disappeared, and VERy hard to get the site back

 

Yes, I think they are. Sorry for the confusion.

I do agree that in practice, fresher documents would typically be preferred. My ‘cave men’ example though is a poor one. Instead, how about a search on ‘Old Testament’ and let’s assume I were searching for a source document. I’m not at all interested in new interpretations, or new analysis. Rather, a nine year old site that contains verse-by-verse ancient writings might very well be preferred.

One could expand this thinking to cover historical documents in general, archived documents in general, original case law documents, etc. I assume these types of searches represent a relatively small percent of total searches, but they may in fact exhibit a preference for older, original documents or ‘stale’ sites.

Google says that freshness may play a role in a document’s score. Taking that statement within the context of a patent application, the rational for making the statement is clear.

Who knows if it will ever be acted upon.

 

Excellent post as usual Owlcroft. The above snippet is uhm Google quote of the year.

 

here's a quick summary of the patent for all those who don't feel like reading it all www.tg9.net

 

So who wants to test this by searching for their own keywords and clicking their own links?

Hey, you'll only need to do it a few thousand times to make it statistically relevant.

 

So... my next script should be one to bookmark my own web pages thousands upon thousands of times...

How to do this so Google will notice? How do they track this? The Google Toolbar?

 

i would imagine this type of tracking would need to either come from the google toolbar, or from a google browser (which they are working on)

 

This is what Google appears to be telling us to do:

Domains
1. Register new domains as early as possible.
2. Register your domain for as long as possible (10 years).

Web Sites
1. Put content on the new site as soon as possible and get inbound links to that content to register an inception date.
2. Add several pages to the new site to meet Googles "predetermined number of pages" criteria.
3. Add new pages to the site frequently, and regularly.

Web Pages
1. Create new pages as early as possible.
2. Put content on the new pages as soon as possible and get inbound links to that content to register an inception date.
3. Update the content of each page frequently, while staying in the same theme.

Linking
1. Get more links, slowly but steadily.
2. Get links from pages with content similar to your own.
3. Get links from pages that are frequently and regularly updated.
4. Get links that are stable over long periods of time.

User Behavior
1. Use Google to search for your own keywords and click only your links.
2. After selecting your own link, stay on the page for awhile.
3. Make sure your site is listed in Alexa keep your Alexa score healthy.
4. Bookmark your own pages and encourage users to bookmark them also.

 

I was about to mention that section. It was mentioned in other white papers and patent filings. What I wonder is whether they include CTR data from AdWords campaigns landing at those pages. It might not sound fair for them to use it by why would they ignore a good source of such 'quality' indicators?

In the filing they don't distinguish between natural listings and paid results. If I were them I would use ad CTR as a measure of popularity and quality of the target site.

 

They can only collect such data if you have the toolbar or use their browser. Or if they sneak code into other toolbars and browsers. Just goes to show the big brother type of world we live in.