| 3:43 pm on Jun 30, 2005 (gmt 0)|
Yes they look at the age of the site and/or the age of the backlinks to the site - and whatever their age variables are, they seem to carry great weight. Older sites have less to prove when it comes to G's current algo; newer sites need to have quite a few "signs of quality" to do well in the SERP's.
| 8:57 pm on Jun 30, 2005 (gmt 0)|
You may wish to view this thread covering very interesting posts regarding "Information retrieval based on historical data"
New Google Patent Details Many Google Techniques
| 9:15 pm on Jun 30, 2005 (gmt 0)|
|new additions to the algorithms cannot always work on all sites.... so do google look at the age of the site (and perhaps other things like, owner, last content changes etc..) and apply a set of algorithms that will work for that site. |
Uh, what you are describing is one algorithm.
Those "other things like, owner, last content changes" are simply factors in the algorithm.
I think you are looking at "algorithm" from the mathematical standpoint, where it is a simple equation.
From a programming standpoint, "algorithm" also includes procedural constructs such as "if" statements, loops and branches.
| 10:07 pm on Jun 30, 2005 (gmt 0)|
|From a programming standpoint, "algorithm" also includes procedural constructs such as "if" statements, loops and branches. |
I'm well aware what an "algorithm" is BigDave - I've been programming them for many years... Sorry for adding the "s" in there at the end of the word... hopefully you'll still get a full nights sleep. Why do I even bother!
Thanks for that link reseller, I'll take a look at that.
| 10:30 pm on Jun 30, 2005 (gmt 0)|
Why don't your reread your entire first post then, instead of getting annoyed at me. It isn't my fault that you weren't able to clearly commuicate that you understand that there is only one algo.
Of course the algo is written to treat some factors differently depending on other factors. For example, .gov sites will certainly be given much more of a benefit of a doubt, even if it is brand new, than pages on angelfire.
Unless you were thinking that an algo is a simple matematical formula, then it should be obvious that is the way that algorithms work.
| 2:02 pm on Jul 1, 2005 (gmt 0)|
ok, i'll start by saying i really don't know anything about algorithms. :)
but i'm wonderin why it's assumed that there is only one? given that google does seem somewhat reluctant to discuss details, how do we know that many of the fluctuations that show up aren't the result of several algorithms making their way through the data?
perhaps there's something fundamental i'm missing here, but if anyone can fill me in i'd be interested in learning.
| 5:12 pm on Jul 1, 2005 (gmt 0)|
Google is changing their algo on a regular basis, and at any given time, you can have different algos on different datacenters.
But it is one single algo that produces one single set of results.
| 7:50 pm on Jul 1, 2005 (gmt 0)|
>Google is changing their algo on a regular basis, and at any given time, you can have different algos on different datacenters.
But it is one single algo that produces one single set of results.<
I subscribe to that. And thats exactly what I call The Rotating Algos.
The symptoms are described by fellow members on several threads especially after Allegra update (Feb. 2005). Today your site on the top of the serps. Maybe stay like that for few days or weeks. Then drop to #200+ or even further down. Then your are at the middle of the serps etc...
When I asked GoogleGuy a direct question about The Rotating Algos, he preferred to call it everflux.
Guess why Google implements The Rotating Algos?
| 7:57 pm on Jul 1, 2005 (gmt 0)|
reseller, everflux and multiple algos are two different things.
| 8:04 pm on Jul 1, 2005 (gmt 0)|
>reseller, everflux and multiple algos are two different things.<
I know ;-)
But our good friend at the plex wouldnīt admit to the existance of The Rotating Algos. I guess its Google policy not to discuss their algos, and I understand and appreciate GGīs position.
| 1:09 am on Jul 2, 2005 (gmt 0)|
"some new site that used the same seo techniques as old sites, will not acheive the same effects"
definitely, i am having first hand experience on this. my site is relatively new in its industry (<1 year old) and it had been doing extremely good in early stage of the Bourbon updates (first half of June) and sunk into page #8-9 after 16th June.
tried study my competition - check k/w density,whois info, site hosting ip, backlinks quantity and quality...all of these varies a lot EXCEPT the age of the site. all of them are domain registered since year 2002.
my site now is jumped back to the top of SERP by now early July but i am not that excited ...just waiting G to sink it back when they rotates their algo again.
do you folks here think that there might be some age info to be retrieved from the hosting as well? all other sites that host in the same server were dropped radically in the same time and they are back in the same time ...comments?
| 2:43 am on Jul 2, 2005 (gmt 0)|
Googleguy has certainly admitted to google changing and experimenting with new algos on a regular basis. I seem to recall about 18 months ago, back in the old monthly update days, that he was the one that suggested that they were trying about two minor algo per week in between updates.
Generally when there is talk of things like "rotating algos" it is regerring to attempts to cause the "results" to rotate. The reason he would deny this is because they don't do this. It may seem like that is the result with sites that are not robust, but that is not their goal.
| 5:31 am on Jul 2, 2005 (gmt 0)|
And if one is trying to understand or infer the general intention of the SE's when observing the evolution of the SERP's (which occurs in anything but straight line fashion), then distinguishing between flux and testing of alternatives is critical.
| 6:21 am on Jul 2, 2005 (gmt 0)|
>>> "rotating algos" it is regerring to attempts to cause the "results" to rotate. The reason he would deny this is because they don't do this.
I think BigDave is right in this issue. If Google were to rotate its algo, then there is a higher tendency that we'll see full swing of sites and serp up and down at regular interval. But in reality, this is not the case. Those sites that are solid are less affected in each update, while those less solid are at higher risk of fluctuation.
| 6:49 am on Jul 2, 2005 (gmt 0)|
Guys please..you start this very interesting discussion while Iīm going on a short vacation with family on Monday.. Hopefully this discussion will also be here when I return back ;-)
>I think BigDave is right in this issue. If Google were to rotate its algo, then there is a higher tendency that we'll see full swing of sites and serp up and down at regular interval. But in reality, this is not the case.<
It doesnīt need to be on regular basis, but could be on random. If you view the posts of Allegra updates and Bourbon update you will read fellow members reporting to that effect.
>Those sites that are solid are less affected in each update, while those less solid are at higher risk of fluctuation.<
Exactly. At one of my posts in connection with Allegra I mentioned something about "essential sites" which you call "solid". Essential sites are sites which retain their positions and not affected by the rotating algos.
For examples.. sites which are created before 19xx or 200x or sites of PR5+ or .edu or .gov and you name it.
All depends on what Google aiming at within a specific algo rotation(s).
| 7:56 am on Jul 2, 2005 (gmt 0)|
Google's aim is to produce a set of results with the highest probability that the users query will be satisfed as early as possible.
If they have two algorithms that both yield, say, a 95% probability that the user's query will be satisfied on the first page then Google looses nothing in varying between the two sets of results.
Given that Google now has masses of statistical data regarding the success or otherwise of a variety of storage and retrieval techniques it isn't too wild a supposition that they have developed a number of algorithms that each yield an (almost) identical probabilty of satisfying the users query.
Each set may be subtley different and spammers trying to reverse engineer the results in this environment would find it almost impossible.
There may of course be just the one algorithm that has many thresholds (or filters) that adjust the way documents are retrieved and Google may now have a variety of settings for these thresholds that return equally valid SERPs.
| 9:25 am on Jul 2, 2005 (gmt 0)|
merlin31 - I totally agree with that.
From what I have seen it appears that google may use 2+ sets of results. The first set 'solid sites' use a less restrictive 'set of rules' with more weighting, and these usually appear higher in the placements. Then the other 'set of rules' kick in which have a more stricter set of rules with lower weighting.
Sorry BigDave - I know what you mean when you state only one alorithm. I know that an algorithm can be made up of infinite rules, but the basic purpose it to achieve the end result (in this case, position in the results). To just clear something up - I refer to multiple alorithms to try and make things clearer, perhaps wrongly. To rephrase it, I would not mention the word alorithm, I would have to say something like - google uses multiple sets of rules and applies a different set of rules to sites based on their age and history. Older sites which google does not hold the same 'type' of history for, seem to do much better that newer sites. From what I have personaly seen, sites that are newer that March 2004 have much more to prove and cannot be established as quick as sites established before this date.
From what I've seen, my guess is google will try and slowly merge the rule sets (which probably has already started), this should probably balance things out over a period of time. Meaning that older solid sites may drop and newer ones can have a real chance to do better. But who know when the balancing will be over? BTW, this has nothing to do with the everflux!
On to a very similair note, regarding pagerank - Im seing sites with very high pagerank doing very poor in obtaining high placements in results (even in low competitive topics). This is clearly a factor and effect of the above mentioned 'Alorithm' - but I still beleive that the value of pagrank from these sites is as important (in that they can pass it to other sites which benifit, and may pay off in the long run). To give an example, I started a site on March 2004, the domain has a 2 keywords (not very competitive), it reached number one in msn in about 1 month and remains. It has never seen the light of day in google, but has pr of 6. It did take almost 8 months to get any pr, but even after it was received, it did no better in google.
I should probably stop going on now :), hopefully some of you have noticed similiar effects!
| 9:55 am on Jul 2, 2005 (gmt 0)|
can i be a booze hound
| 9:55 am on Jul 2, 2005 (gmt 0)|
not till your sixteen
| 10:34 am on Jul 2, 2005 (gmt 0)|
>On to a very similair note, regarding pagerank - Im seing sites with very high pagerank doing very poor in obtaining high placements in results (even in low competitive topics). This is clearly a factor and effect of the above mentioned 'Alorithm' - but I still beleive that the value of pagrank from these sites is as important (in that they can pass it to other sites which benifit, and may pay off in the long run). <
I donīt know howmuch importance we should assign to the current PageRank (PR) that we see on the toolbar/ directory. We know that it hasnīt been updated for ages. And I really doubt that what we see is exactly what we get.
My theory is the PR Google use "internally" isnīt anymore the same as we see on toolbar / directory.
| 12:17 pm on Jul 2, 2005 (gmt 0)|
|My theory is the PR Google use "internally" isnīt anymore the same as we see on toolbar / directory. |
I think internally PR is live and dynamic but the toolbar just shows the last 'download' of PR ratings.
You may be looking at a PR 5 site but the toolbar shows PR 3 because that's what the site was last time toolbar PR database was updated.
They wouldn't want spammers to see the 'live' data because then they would start testing it.
| 12:23 pm on Jul 2, 2005 (gmt 0)|
so if you compare toolbar PR to the current results you may be comparing 2 algo's
| 3:28 pm on Jul 2, 2005 (gmt 0)|
>so if you compare toolbar PR to the current results you may be comparing 2 algo's <
I also assume that the factors involved in the "internal PR" calculation are different now than those the last time the PR on toolbar was updated. Maybe more TrustRank and United States Patent Application, 20050071741 things are involved now. Who knows ;-)
| 6:37 am on Jul 5, 2005 (gmt 0)|
the truth is out there
| 10:07 pm on Jul 5, 2005 (gmt 0)|
I am not real experience in this as I feel most of you are by reading your posts. I have been doing a lot of reading in the last month on this site and I noticed one curious thing. A lot of talk about dropping down in the serps like back when their site first started. Multiple algos and so forth. Sites that seem to have popped up from the back of the pack that used to do well before certain spam filters were in place. And one very interesting post I noticed from one person saying that googlebot was visiting his site and asking for pages that were long deleted. Old content. Now my question is: does the algorythm in place affect googlebot's behaviour? Is it possible that during bourbon and even off and on now an "old" algo was being implemented? One with old filters.
Like I said...I am running full speed to catch up to you guys and feel like I am just falling farther behind so if I am way off base don't hammer me, Okay?
| 5:05 am on Jul 6, 2005 (gmt 0)|
>>> And one very interesting post I noticed from one person saying that googlebot was visiting his site and asking for pages that were long deleted. Old content.
One possiblity is that there is/are old rotten link(s) somewhere pointing to that deleted page and Googlebot tried to follow that link to fetch the page.
| 8:02 am on Jul 6, 2005 (gmt 0)|
there is some utterly wild speculation in this thread, so I'll throw in my own fiction here: I imagine google to be like a 3-bladed propeller, with a 3-bladed propeller on the end of each of the larger 3 blades, all spinning away as it were. In this, there are actually four incarnations of the algorithm being processed, the central hub, which is never exposed to anything but the three peripheral algorithms. The three propeller/algorithms on the outside get exposed to a period of about two weeks of spidering, caching and recording new document and link data, then a period of about 2 weeks crunching this new information and sychronizing with the central hub, then about two weeks serving SERPs up live at google.com, and then to repeat through the cycle again.
In this way, you still only have one algorithm which is a conglomeration of other running algorithms and processes, but you can still see different faces of that one algorithm. Each of the different incarnations could have their dials tweaked to certain settings to test certain ideas and then it would appear a different set of SERPs was running.
| 8:03 am on Jul 6, 2005 (gmt 0)|
Very interesting thought about the algo effecting googlebot.
Perhaps google is just getting lazy and is using old cached links (from old indexes) to try and increase the size of the index. Or it may be trying to roll back time to apply new filters to old sites!
Google algo is the 8th wonder of the world, lol.
| 6:26 am on Jul 7, 2005 (gmt 0)|
I suspect that google did revert back to some old data a few times during the "302 crisis".
texasville - this topic calls for nothing but speculation since there is no way to verify any hypothesis.
I think 404's are a real problem in google, since the meaning of the code is unclear at best in the HTTP 1.1 documentation.
404's will keep coming back like skeleton's in the closet every now and then.
if you have this problem you want to force a 410 instead - that is very clear and specifically targeted at removing pages from search engines.