Forum Moderators: open
The question is how Google does that and how long does it take until an off-topic link gets removed?
I saw a competitors linkcount go to 300% with lots of guestbook entries. I do hope that in the next update those are gone!
Id imagine it will open up a new can of worms though, with regards of determining what links are on theme / relevant.
All the "my backlinks arent showing" posts will turn into "Google has banned my backlinks"! ;)
JOAT
So the links "disappeared" somehow from Google!
may be it is possible for industries and profit sites.
but, how to calculate the relevance of a link at a site of a political party?
will G analyze all manifestos?
Or inbound links to someone's favorite political party? Such links don't count unless your site contains the lingo of their manifesto?
I have been making a lot of use of Google news and it seems that it is getting the hang of aggregating news from multiple sites on a single topic. Therefore it would seem not unreasonable to assume that Google is getting better at identifying links around common themes.
In any case, I guess we will see a lot more on this subject in the coming months, with the result that linking strategies will have to change significantly.
Last month I noticed 2 PR3 sites with few links listed as backlinks. I was also missing several PR4 pages with significantly more links. Could they be showing links according to passed PR instead of the PR of the linking page?
Good points... makes sense to me.
But I suspect theme relavancy is kicking in.
It's more than likely that the missing links being off-topic is purely coincidental.
The pieces for easy themeing have been there for years. How well does the anchor text match up with the content on your page? How about word word frequency?
If you are linking to the Widgetarian Party website from your cooking site, it will be less on theme. If you link to the Widgetarian Party site, and you link directly to the page on Widget Rights, from your page ranting in favor of Widget Rights, it is VERY on theme and will likely use very similar language.
Themeing IS easy. And it can be done with at least 90% accuracy without adding too much processing time. The question is, is 90% enough? It probably is. There are enough good sites out there that will pass the themeing filter that they do not have to worry about those that slip through the cracks.
It is just hard to accept when your site is the one that slips through the cracks.
I don't think this one flies.
PR for sale and off-theme links AND advertising may all be related to the drive to combat PR for sale.
Site A (PR9) is about widgets it has a link to Site B (a PR3) website. Now we don't know what site B is about...if it is RELATED or SIMILAR to Site A at all.
From this unknown we can assume that either Site A is selling PR, advertising for Site B or perhaps is off-theme...; OR it could be a similar website and a valuable resource.
Solve the unknown and kill 3 birds with one stone, or promote a worthwhile resource of information.
It doesn't have to be a little 'off-theme', just past a certain point/threshold.
Find the off-theme link:
color>rainbow>sky>hockey sticks>
How did we know which link was off theme?
Another:
color>rainbow>sky>rain>trees>wood>hockey sticks
Hmmm...on theme to me!...well sorta. I'd let this one pass muster.
Solve these questions and get on on-theme set of results.
I've been watching this for awhile and I'm convinced that if a SE could do this...they'd have an algo that combats a lot of 'problems'. (as in THEIR problems...not ours)
I would think the 'similar pages' portion of the current algo would go a long way to helping this along. If the similar pages of Site B are NOT similar to Site B...it doesn't quite pass the test.
Or...I could be wrong:)
Solve these questions and get on on-theme set of results.
This is exactly what is being worked on with "Google Sets" in the Google labs section.
A few minutes spent working with the tool though, and it becomes quite clear that this technology has not yet "arrived".
You can bet that if they can get it refined, it will indeed be factored into their algorithm. If successful, it is likely to change the landscape of search forever. (much the way PageRank already has)
If Google wanted to somehow come up with an algo which would determine "on theme" links, they only need to determine keyword density for both sites to figure it out. That in combination with each site's category listing should do the trick.
A site which is all about travel in India but which links to a car manufacturing site should soon be flushed out of a site's links index. Of course, this is an over simplification of the problem ... (otherwise, I would be working for Google and making millions) but you get my meaning. :)
Results 1 - 3 of about 6.
Results 11 - 13 of about 28.
Results 61 - 63 of about 128.
While these 3 show a lower ratio
Results 11 - 18 of about 48.
Results 11 - 15 of about 118.
Results 11 - 16 of about 278.
Results 91 - 92 of about 276. This site dropped around 60 links at the last update, according to my theory, it'll lose more at the next one.
The site that used to rank no.1 in the category has gone below PR4, and so is not showing any links, I put this down to it being the only one without a DMOZ listing.
I had a look around it, it's basically un-optimized, no tricks, no dodgy links, no reason for a penalty, IMO.
The one thing common to all the lost links, was they were non-theme, with little or no relevance.
Let's use a mom and pop liquor store in Idaho linking to a construction company site in Alberta.
OK, yeah ... they grow the wheat for the booze in Alberta right? :)
Perhaps a top score of 10 to a link that came from a site with a higher PR with no link going back to it. A link in it's purest form, unsoloicited from a more authoritative site.
A lesser score for the above if there were a reciprocal link back. That link could be simply the result of a "deal" between two parties and have nothing to do with any site value.
And so forth, the same logic applied down the line. The bottom being one from a link farm. One of that genre being given a minus "value score".
If I have an entry at DMOZ I have to avoid
- to link DMOZ or
- to link to somebody who has a link to DMOZ or
- to somebody who has a link to somebody who has a link to DMOZ
- ...
?
Perhaps a top score of 10 to a link that came from a site with a higher PR with no link going back to it. A link in it's purest form, unsoloicited from a more authoritative site...A lesser score for the above if there were a reciprocal link back. That link could be simply the result of a "deal" between two parties and have nothing to do with any site value.
Interesting idea, but wouldn't it require a lot of computer resources, if only because most sites that reciprocate inbound links are doing so from other pages? On a site with several thousand pages, Google might have to do a lot of digging to identify the reciprocal link.
Also, the system would be open to abuse. A Webmaster could tell the person who's providing an inbound link, "You link to my PR5 bagels site, and I'll link back from my PR5 doughnuts site." That way, Google would think both the original bound link to the bagels site and the reciprocal outbound link from the doughnuts site were "purest form" links.
ie; If I have a web design site and yahoo links to me, even though they have no content on web design, it still may mean a lot. Not a great example but there are an infinite number of possibilites.
Also, how can you determine the theme of a site like Google itself?
If I have a web design site and yahoo links to me, even though they have no content on web design, it still may mean a lot. Not a great example but there are an infinite number of possibilites.
Actually, DMOZ, Yahoo, and other directories make it even easier for Google to draw conclusions about a site's theme, because their links are neatly organized into categories. Yahoo's home page may not be on your theme, but >companies>internet>web design (or whatever the category might be named) certainly is.
Things get trickier on a site that isn't structured like a library catalog. Let's say that my Europe travel-planning site had an article about Bruges and I linked to a story on the annual Holy Blood procession from a "sightseeing" or "miscellaneous links" page. My article and the Holy Blood story would be much more closely related, in terms of theme, than one might think if the linked story had only one or two instances of the keyword "Bruges" in its text. And if my page were titled "Bruges - Sightseeing" and the linked story were titled "Brugge - Holy Blood Procession," Google might become even more confused, since Google's algorithm would be unlikely to know that "Brugge" is the Flemish/Dutch equivalent of the French name "Bruges."
