30 replies to this topic

[ghergich]

Patent Here

Ok its a little long but easy to read. Basicly its saying goolge wants check the age of a document and its freshnesh. I think these were already known but its good to see it from google themselves. They also mention detecting unnatural spikes in linking and what i feel is kinda scary looking at your bookmarks to determine if a document is important. There is plenty more so read on.

Edited by projectphp, 02 April 2005 - 09:14 AM.

[sweepthelegnate]

interesting...

[Grumpus]

Good spot. I hadn't seen this one before.

This is something that's been a part of the mix since the Florida update (or Fonzie or whatever the heck they called it here...) Based upon some of my older posts, it looks like they were collecting data to prepare for its usage around April 1st, 2004. You'll also note in that post I linked to that Google was already starting to establish the age of documents - or at least getting the mechanisms in place to do so. (See, it pays to listen to what the Grumpus says).

I noticed, too, after restarting my movie site after having it throw 404's for a year that it still had a lot of inbound links from places, but those links didn't really pack any punch - despite the fact that 90% of them were spot-on relevant to the page(s) they were linking to.

You'll also notice that once I figured out why Google was looking at the age of documents, I stopped talking about it altogether. There was a reason for that, but now that this is out in the open, I'll explain it.

This is one of those things that's going to get people all in an uproarious tizzy. The forums are going to be flooded with people asking whether writing a script to display some random text on a page will fool Google into thinking a page was updated. Everyone and their brother (or sister) will be suddenly scrambling to come up with a way to beat this one - if for no other reason than the fact that it seems like it should be pretty simple to do.

I'm not so certain that it is so simple, though - so, the frenzy will probably last a good long while.

Oh well. For those who want to remain sane - don't worry about it too much. It just means that you've got to constantly be looking for new links to your site - something that's no different than we've been telling you all along.

G.

[sweepthelegnate]

Oh well. For those who want to remain sane - don't worry about it too much. It just means that you've got to constantly be looking for new links to your site - something that's no different than we've been telling you all along.

so perhaps monthly link building budgets will become in vogue? good ol monthly link building budgets.

[Tom Philo]

Well, with 63 different items to code for, and try and defeat, that will keep the Page Rank fanatics going for a few years! LOL

I like the new term (to me) they put in "link churn".

(If you are in the British Isles, please mentally put the . inside the quotes.)

I cannot 100% accurately summarize this whole item, but on my quick read of it covers: age of link going to a page, how often that source page with the link is updated, how long it has the source page existed, revelancy of the whole source page to the target page; the target page how long it has existed, intital date found by Google (search engine), when it first found a link to the document (not from the same domain), how often the page changes, how MUCH it changes, how frequent it changes, past change rrequency, current change frequncy, delta changes, and if it changes on a regualar basis (stat analsyis of change to determin if normal for document, think sales page, vs. programmed change for change sake), how often new pages are associated with it comparted to total pages, weighting what is changed within the document (text vs headings, vs what changed in the "meat of the document," (, inside quotes this time to have equal opportunity of our British friends to unthink putting it inside quotes), how often the document is selected from the search results over time, scoring it based on search terms that is in the document over time (history data), if document is "stale" but people still go to it making it rank higher anyway, (so "old" revelant documents not updated often which are still used by people, no penalty), link pattern analysis of overall relationships to the document to detect changes (removed, added, source of links) rate at which they appear - link trend analysis, differences in anchor text to page meaning, changes in linked anchor text to document, how it differs over time to page (anchor text drifts off of page meaning lower value?), when anchor text link first appeared in relation to when the target page appeared, how long domain has existed, when domain expires, how long other documents on the domain have existed, how much the document moves in rankings (recursive analsysi in essence), how much it moves in the rankings over time, growth of links over time, peer links (likely in conjunction with domain ownership and registration timeframes to eliminate domain link farms, so 50 domains do not come up same day, week, month year, and then link all to the same target domain to boost rank), and using a set of forumlats to determin link churn based on quality, quanity, vebage, staleness, freshness, along with threshold values: all of the above is done to help rank the document.

Whew! Guess that is a summary!

[Hyperformance]

Hmmm, some very interesting stuff in there...

You're right, in that this will drive some to madness It is a good read though, (educational & informative) if your in this Industry.

I guess they have to give this kind of detail when attempting a Patent? Makes sense.

Nice find Ghergich! Thanks for sharing!

[ghergich]

http://www.socengine...ata-patent.html

here is a real good summary artilce - my take is they tried to write down all the ideas they could just to patent them even if they dont use them - something to keep in mind - like brainstorming patent session.

[projectphp]

This patent app is also great in that it says things like "A method to use.." and then never really says in which direction, e.g. is older better, or is newer better? Does a link's benefit fade over time, increase over time, or is it like a hump, with links getting more useful, before becoming "stale" and fading?

I really don't know, and think that it will inevitably vary by search term, and for good reason. I have been sayiong for a while that different searches suit different ranking methodology, because they are very different.

Two contrasting examples of this would be the Michael Jackson molestation claims and OJ's Trial. Jackson's case is actually two distinct cases, one years ago and one now, so some pages on the old case will be stale, where as OJ had only one trial, so old links are still very valid, and the topic can not go stale.

I like this new patent a lot, and think a lot of the ideas make good, solid sense. Not sure why Google need a patent for this, but they seem to feel they do, for whatever reason.

I also think that a lot of this has to do with the Google Search Appliance, as these sort of things would be most useful in a company setting, as a search for "Company Leave policy" would benefit most from this sort of tech.

IMHO, these developements are good for the internet, because the more complicated Google and Search Engines get at ranking sites, the less spam we will see. Many will say these moves invalidate SEO, but I don't believe so. In fact, I think the opposite. The more complicated it all gets, the more important having a sound SEM strategy becomes. Not a algo busting stratehy, but a long term, well rounded strategy.

Interesting times are ahead, and we will perhaps see a return to the Good ole days of forums, in which "innovative strategies" become the norm again, as people try all sorts of rotating scripts, old / new content as they struggle with the new playing field. Facinating times ahead!!!

<edit>ehicha?? What was I thinking???</edit>

Edited by projectphp, 02 April 2005 - 01:44 AM.

[Randy]

I couldn't see a reason for requesting a patent either pPHP. In fact, some of it seems to be a tad over-broad to the point of being difficult at best to obtain a legal patent for. I mean, can only one company say that they're going to analyze links in order to factor that into their algorithm?

Beside, MSN has had their little sliding bar thingee to return newer or older pages since they went into Beta.

I like the general concept laid out in the patent. And like that they were very vague in what's going to be best to rank well on the aging side of things.

Oddly, the first thought that crossed my mind when scanning the patent was that maybe they were looking for something that was sort of like PageRank, which is of course held by Stanford, but not. Still dealing with links, but extended enough that it's not the same.

[Grumpus]

It's similar to a pagerank score, really. But, it also takes into consideration the age of the link (combined with some other factors - such as how frequently the linking page is updated). It's not so much the age of the page that's critical here, it's the age of the link.

G.

[Randy]

Right G. That's the way I read it too when I finally took some time to really read through the document this morning.

Does this perhaps mean that we're going to see Google drop PR for something else? FTR, if they do, I don't think it would have anything at all to do with Stanford holding that patent. My understanding is that they have a pretty much perpetual license to use that algorithm and name.

I'm thinking more along the lines of them changing the name to give all of the PR chasers/abusers a massive headache.

Personally, I think that would be a very good idea. They could certainly generate a huge amount of press. I can see the headlines already... The New Google ! (When really it's the same old Google since Nov 2004, and as Grumpus already pointed it it goes back a lot farther than that.)

[tempy]

And I suppose it all depends on where they are going to use this. Are they going to use it on regular results, or just on Google News results? This would make more sense and it is also an area that they have been 'questioned' over recently.

When I do some searches, I don't want the latest results, often. I also don't want websites 'updated' just for the sake of keeping a top position in Google's regular search - and I'm sure Google understand that also. Didn't someone on here (Grumpus?) recently point out that a lot of news and blog sites appeared to have been 'moved' across to the News section?

[smakyyy]

Very interesting - am a new bie here and i found this site to contain so much good stuff that i have an information overload

[robbinsr]

With regards to examining what users bookmark/favorite, which I agree would be a good measure for a S.E. It seems to me there would be two ways for Google to do this, one being with some form of spyware as was mentioned; call me optimistic but I don't see Google going down this road as this is something they get accused of often enough with Gmail etcetera even when they aren't doing it. Or two, if they were to create a free web space for their users a la Yahoo, and then have a spot for users to manually enter their favorite sites to be bookmarked, again a la Yahoo, then that would be a good source for those types of links, and would be above board since you would have to manually type them in yourself and give them to Google.

Besides which, they are already increasing the amount of space you have available in Gmail, someone posted a rumour earlier about Google offering free web space in return for having a Google S.E. on your free space (sorry I can't remember who that was or which posting I read it in) and they already have some similar involvement/familiarity with the Blogging community through Blogger. So who knows? Remember you read it here first if I'm right, and you've never heard of me if I am way off base, deal?

On the other hand, they could simply be trying to bury a good idea they came up with but don't agree with before another S.E. company thinks of it and tries to use it by patenting it. This also would fit in with the company's ethos so who knows?

[thx1138]

Excellent find Ghergich, thanks for that, what a great forum this is.