I have to wonder: will google ever stop just counting links and start counting links that are from on-topic sites as 'better'? I'm sure they could do it, but so far, it seems they haven't. What I mean is, external links from a car site to another car site would be more valuable and count more in the algo than a recipe site linking to a car site. However, it seems that google doesn't give a rip as long as the recipe site is well established and constantly changing. And even then, if you get enough blog links, you can still help your cause. I have to wonder how much computing power it woudl take for google to acually look at a url and know that links from that site to other sites are on topic. Maybe there isn't a way to do it. Does anyone else every wonder about this?
I think that google already does give more weight to on topic, links than links from any other website. As far as I know, google has been weighting on-topic links higher for a while now.
I concur. I saw a site jump in rankings by virtue of about a half a dozen on topic links (competitive phrase), then when the links were removed, the site dropped again. I've also gotten a client ranking with <1000 strictly on topic links, where neighbours have hundreds of thousands. I tend to check the 'similiar sites' in Google when developing links. Trying to get links from sites that are similiar to those already ranking is good IMO. I also make sure that if a site is in industry X, that the similiar sites are also in industry X before I go looking for a link.
How would Google determine an on topic link vs an off topic link? Example: I have a site about snowmobiling. On my site I have links to bars & restaurants in the area I snowmobile. Very helpful for users who go to the same area...but how would G' determine that restaurant links on a snowmobile site are "on topic"? I have yet to see any solid, unbiased evidence that "on topic" links help any more than "off topic" links.
I think that could be argued Anyway, Adsense is somewhat decent at showing ads related to the page that the ads are being displayed on. Adsense doesn't display ads on sites based on what they link to.
No, but my point was that AdSense can determine page content and deliver ads relevant to that content, so there is a technology for determining relevance already in Google's arsenal. There's no reason why, thereetically at least, having made a determination of page content, it could not trace backlinks and compare them to the target page content in a similar fashion.
I definetly see an increase in importance of external on-topic links (including deeplinks) in the works, and perhaps a decrease in important of internal or off-topic links.
Using the example I provided above how could a coloration ever be drawn between restaurants and snowmobiles? I should add that I agree - at some point "on topic" links should and will be "worth" more in Google. I'm not that smart though and am just curious how, using only technology, it will be possible.
Possible, but I would think that there would need to be a more universal approach to scoring the relevancy of the links. i.e. geographic locations may very work in some cases but not in others. I would imagine that there would be some sort of scale used - it wouldn't be as simple as "on topic" or "off topic." There are degrees of relevancy I assume.
I expect you're right... in all likelihood, there would be some sort of relevancy rating rather than just 0 and 1. For example, you have a site about a local Chicago snowmobile club. Links to/from restaurants, gas stations, snowmobile clubs, parts supply depots, etc., in the Chicago vicinity would likely get higher relevancy ratings than the same services/business in Pittsburgh. On the other hand, snowmobile clubs in other cities would likely receive high relevancy ratings.
That's where it could get interesting. It doesn't snow enough in Chicago for a decent snowmobile ride. My theoretical snowmobile club travels to Michigan every other weekend for our trips, and all the links on our site are to restaurants, etc. in Michigan. Very "on topic" for potential members to our club since that's where we ride but I would think it would be hard for a search algorithm to figure that out.
The variations are intriguing, for sure. However, I do believe that if or when Google wants to do this (if they are not already doing it to some extent, which is my suspicion), they have (1) a sufficient available mass of intelligence, creativity, and ingenuity to pull it off, and (2) the technology and sheer computing power to make it work.
guys all this stuff might happen but it would be really hard to program first what are sites about? your snowmobile club site would countain hundreds of words, which ones really describe what your site is about they could possiblly use the higher frequency key words, or the page titles but how many subjects could one site be about? 1, 2, 10? There would have to be a comparsion to the site the link is pointing at, and hidden relationships, I guess relvancy could be made I guess it could be possible but the amount of extra resources involved would be large, if every site now has a subject its relevant to then thats another record added for every website, then also every link would need an extra record of some sort to store its "relevancy" score Not to mention it would be pretty inaccurate, there are tons of quirks in the english language.
ferret, you're confusing "site" with "page". A site may be about many things (Psychlinks has over 100 topics on separate pages) but a page is generally only about one or two. As for it being difficult to do programmatically, it's already being done. AdSense does this for millions of pages day in and day out. if you have a dynamic site like a forum, you can make a couple of new posts and watch the ads change almost in real time.