Forum Moderators: open
[webmasterworld.com...]
ciml's quote stands out the most for me:
I don't know anyone who's found link-theming in Google using controlled-conditions test.
I guess what I'm trying to say is that I haven't seen any evidence of "link-theming", hubs, authorities, or anything else other than your standard, 2002, "it's got more links, so it's ranking well, and links from it mean more" type results.
I'm asking someone to prove to me that Hilltop/link-theming is being used. Let's revisit the concept discussed in that thread: Is hilltop in play?
I, for one, have not swallowed the "Hilltop" magic pill yet.
We are very strong in one sector, although we play in many. Evidence suggests that we have maintained our rankings because of hilltop effect, whereas in other sectors where we used to do well, we have dropped.
>I haven't seen any evidence of "link-theming"
My understanding of hilltop is that the 'link theming' is done at run time for a search. A search is done for 'widgets' and 1000 sites/pages collected. They now by default have a common theme... the search term. Hubs and authority links are counted within the search results and ranking established.
We think we can see this effect for searches related to Scotland, where we are based. Over the years we have naturally attained links in for Scottish themes, having done many sites for clients. If we put up a new site, the Scotland related searches ( like 'widgets in Scotland') rank well from day one (even in competitive areas), whereas other searches like London Widgets just don't seem to happen. I think this is a direct result of hilltop. For "london widgets' we just do not have the support of sites linking to us that appear in the results, but for Scotland we do.
I figure that if I focus on cultivating links for the sake of relevant traffic, that will provide enough theming for Hilltop to take care of itself when/if the day comes.
I, for one, have not swallowed the "Hilltop" magic pill yet.
So even if there is a hilltop algo it would not show up in competitive serps.
So whether Gogole uses hilltop or not, links from links pages will count fine.
If Hilltop were in play then I wouldn't have expected something like this to happen.
So far I can only guess through limited observation that:
* maybe the age of links matters;
* duplicate content doesn't show up on the first page if you're second to have it indexed, and possibly links from duplicate content pages do little for ranking.
On one other site, duplicate content and links pages haven't been crawled in ages and are now listed in Google as a URL only.
Affiliation
Evidence of affiliation assessment is apparent to me, but with a caveat. We saw after Florida (when I believe Hilltop kicked in) that - in several commercial categories where we had multiple site entries with insufficient identity safeguards - we lost a bunch of those sites...meaning that their positions in the SERP's dropped dramatically. That was interesting. What was also interesting was that in most of those categories, our lead site maintained its overall position or even benefitted.
Of note, after taking care of items related to methods for determiniming affiliation, we saw most of the sites come back. And those that did not come back happened to use common templates to others that were again thriving. This seemed to reveal one of the roles that filters were playing in the weeding out process.
The caveat It has always struck me as possible that what many including me view as Hilltop-style assessment could instead be little more than a series of well constructed filters. At which point, are we looking at Hilltop, or a system influenced by Hilltop.
Theming
Here too I believe we have evidence, but no strictly contolled tests.
Since we operate a large number of sites, we did take the opportunity post Florida to evaluate different hypotheses in our quest to regain some lost traffic (commercial categories).
One thing we did was to identify a pair of very similar sites in different categories. The sites were deemed similar by virtue of size, construction, PR, linking patterns, and performance in the SERP's. Call them site A and site B.
For site A we went and got 20 good backlinks (PR 6-7) from non-affiliated sites, in categories unrelated to site A's category. No help; the site stayed buried.
For site B we went and got 8 good backlinks (PR 5-7) from closely related sites (two hubs, six authority). Within four weeks site B had popped back to its former glory while most webmasters in the immediate post Florida environment were still bemoaning the disappearance of their sites.
During that time we noticed another interesting change. In some SE's, the site listings in the SERP's were occasionally showing URL's taken from the backlinks (nothing new there). What was intersting, and still sometimes is, is that post Florida the URL's were typically associated with authority sites. Before Florida, when we saw that, the URL's more typically reflected high PR pages. The assumption here is that a really important backlink is displayed, but that seems a good assumption to me.
On a related note, though I can't call this technically Hilltop, we have virtual certainty that links from unaffiliated, relevant pages that are tightly connected to our own topics perform better than identical links from unrelated pages, for certain kw searches.
All circumstantial I know, but with things like this I have my own 80/20 rule: As a manager, I try to understand 80% of the technical side of things that I'm no expert in. Trying to understand the last 20% is counterproductive. In the case of Hilltop, I'm 80% convinced that it's there. But more importantly, if it's not, I'm 100% sure that operating on the assumption that Hilltop is there is beneficial to the long term health of my sites. ;-)
So I don't see how Hilltop could be playing a role in these competitive commercial categories.
I see the same domain listed twice in the first page or two all the time and most of the sites are doorway style sites to whoever they are affiliated with.
So again, as I have posted in other threads, is Google broken?
Personally, affiliation detection wouldn't surprise me at all. Monika Henzinger co-wrote an interesting paper on the subject (which I can't lay my hands on right now).
On link theming, I'm not so convinced. Keep in mind that a theming engine can be much easier to manipulate, and trying not to be manipulated has been a major pillar of Google's changes over the last year or so.
People go on about content being king, hilltop, semantics etc but I think the effect they have seems minimal.
I am sure it can tip the balance in competetive areas when sites are 'neck and neck' though
But from all the serps in my area the allinanchor results are all but identical to a standard search
At the same time, let's not confuse two issues. I've noted steveb pointing out any number of times that bulk inbounds from low quality irrelevant pages can help a page high in the SERP's. True enough, unfortunately. Quantity has always been one way to overcome lack of quality. But the fact that high rankings can sometimes be achieved with mass low quality backlinks does not negate the possibilty of Hilltop. More than one way to skin a cat.
Also, there's nothing that says that Hilltop had to be implemented precisely according to the original papers. :-)
The way I read it, the Hilltop/LocalRank "affiliate" filter is quite subtle, and it would need a pretty heavily cross/interlinked domain farm targeting a single category with relatively few "outside" links for a dramatic drop in the SERPs.
There again, irregardless of Hilltop/LocalRank, such a set-up could well suffer a penalty anyway ;-)
I see tons of evidence that links from thousands of different IPs is valued in terms of not facing lag time, but otherwise they are algorithmically recognized as algo trash and the target sites only manage to rank at a middling level like in the teens (which of course is better than in the hundreds if you are lagged).
I see clear evidence of recognizing affiliation (not by IP). Use the allin: commands and see affiliated and pseudo-affiliated sites often grouped in a bunch. By pseudo-affiliated sites I mean something as simple as being on the same dmoz page, but it is more obvious when multiple links are involved.
The way I read it, the Hilltop/LocalRank "affiliate" filter is quite subtle...
...would need a pretty heavily cross/interlinked domain farm targeting a single category with relatively few "outside" links for a dramatic drop in the SERPs.
All I know is that where we had mutiple sites in a category, most got knocked out by Florida, and the combination of changing similar site traits and acquiring certain kinds of backlinks helped us recover in short order, when others were still unhappy. FWIW.
Personally, the reason I think G may be using Hilltop with affiliations is because I believe Hilltop is also at work in theming, so once you buy into the theming thing, assuming that Hilltop is also at work with affiliations is no big leap.
Still, I would not bet much on this as I'm not technically savvy enough to figure it out. All I know is that operating as though Hilltop is at work seems to help us. That's all I need to know. :-)
Is it the case that you feel that as long as your anchor text is handled appropriately, it does not matter whether the external page/site from which it comes makes a difference, other than PR? (I know that's not what you said, but is that what you think?)
Also, on a related front, do you feel that outbound links to relevant (tightly themed) sites help both pages rank well for kw's related to those two pages? That is, more than would be the case if the two pages were unrelated thematically. ;-)
In both Paper and Patent, does the affiliation test not come into play only on the inter connectivity between the first set of results?
So no links, no affiliation test, no drop in ranks?
A links to B. B links to C. A and C are affiliated.
A and C were our sites. B is the auth site with which we have a good relationship. In this case B linked not only to C but also back to A. But C and A did not link to each other.
C was toast after Florida.
However, as I noted, it could well have been dup filters of some sort that killed C, temporarily. I do not know.
Also, there's nothing that says that Hilltop had to be implemented precisely according to the original papers. :-)
This thought has occurred to me a number of times. Could it be that Google is applying some sort of affiliation filter, but it's using a variety of affiliation tests... not just an A to B, B to C thing?
The Google "Similar pages" results, which I believe classifies pages with common inbounds, could be further cross referenced and/or also correlated with hosting criteria, cross-linking, etc, and perhaps also weighted relative to the number of independent links ("clean" inbounds not traceable to a common source), and you'd have a measure of affiliation that certainly applies to some sites that I've seen drop like rocks.
I have a site/page that is #1 for widget insurance.
On the home page of that site I have a link using the term "widget insurance" as link text which links to a Google Directory which has 52 listings on the broad subject "broadwidget insurance". Within those are 5 that are on widget insurance and which use stems of the word widget in their listing text. One site listed (not mine) has the exact words "widget insurance" in the link text which appears in Google Directory.
Search for widget insurance (prefs set to 10) and:
My page is at #1, the competitors page that has the term widget insurance in the GD link text is at #2 and interestingly the Google Directory category is at #9.
Before I put the link to the Google directory (as part of my strategy to overcome the florida effect) in February that directory category did not appear in the SERPs for widget insurance and to be honest you wouldn't have expected it to be.
Best wishes
Sid
Also, on a related front, do you feel that outbound links to relevant (tightly themed) sites help both pages rank well for kw's related to those two pages?
A few weeks after Florida and loosing all rankings, as an experiment, I decided to link out only to broadly related super-authority sites from my index page. A week later, I was ranked with or above the super authority sites for the kw combo "business 2 business". My site is a broad match to this kw (never targeted and not normally ranked).
In addition, "business 2 business" was nowhere to be found in my site (not just the index page).
This proved that G was ranking my site based on a broad-theme-match from the outgoing links on my index page.
As soon as the links were removed, my ranking was removed from this keyword combo. I've not been able to duplicate this since update Austin. Now, it seems the more I link out to important and useful information, the less it affects my ranking status.
But alas, it helps my visitors and that's all that matters... right? :)
So, let's assume it's something else. Let's say you had, well, let's just say 1000 students of any discipline, and you had been given a task to pick the ten best ones, aka the top 1%. What would you do? ...okay, but then what about if they suddently more or less all had figured out how to get those A grades, how would you do then?
You know that some of the good guys are plain lazy, and you know that all of the bad guys are cheating, and you know that if you want to be best at picking that 1% then you just can't afford to rely solely on those grades anymore (although you still need them to some extent), so... what's next?
Oh, and it has to be a really simple solution too, btw.
Edit: Reference to "Generic head ache pill name" removed
So if Site A always lies, and Site B always tells the truth, Google need just ask one site what the other would say is the best... and then rank the opposite first.
Pardon me, I'm going to go link to my competitors.
