Then you are just guessing. Not necessarily: sometimes thousands of domains are hosted on the same C-class, and they could represent the pages of thousands of different webmasters. On the other hand, getting hosting on multiple IPs can be (relatively) inexpensive on some hosting packages, and the cost may well fall. Another way to get multiple-IP links or hosting is to sign up for lots of different free services (such as Geocities, Blogger, etc). If Google used IP as a factor in deciding the value of a link, it would be much easier for someone to buy their way into the SERPs, or to spam their way in. By analysing backlink patterns more closely, Google could avoid this problem. I'm not saying that Google don't consider the IP that a link is coming from: just that the situation is more complex than many seem to realise.
If a man walks down the street and drop his pants, people will look. Even the blind people will look. Some things are more obvious than others and I can gurantee that Google see all these things. What we don't know is how Google thinks about it. But if you put Google in that situation many times, you might just figure it out. 1. Yes, what are the chance that another web site related to yours are hosted on the same class c-network? It's not big but it's possible, especially if it is in a language for a small country. But even if some natural links would come this way, it's not realistic that all links would. Not even in the smallest country 2. Yes Google is an registrar and have whois-information but you can tweak this with protection on some domains and small changes in the whois-data. Additionally they don't have access to whois-data for country-specific domains like .se, .dk, .de, .fr and so on. 3. Yes Relevancy is of great importance and might just be the key to make a link on the same class-c network woth the same as on different nets. Edit: Guessing is maybe wrong word for what I'm doing. It's more like I'm speculating with my current knowledge and experience of the Search Engines. It's still in their interest to know it even if they will not do anything about it. They probably have algorithms for this and factors like this take place in it: - How many sites do Google know exist on a current class-c network - How many links comes from the same domain owner - How much perecentage of the links are coming from this class-c network - Are the sites realted to each other (more than having the same owner) And so on...
I used to link all my sites, then surprisingly all my rankings plumetted, I therefore went through all my sites and removed the links surprise surprise they recovered, this is just my experience. but i wouldn't do it, The server was my own running Windows on a fixed IP all sites using same range. Could have been coincidence but it worked taking them down, and i'm not about to try it again. PS they were relevant too one a loan site, one an insurance site, and one mobile phone insurance site.
If you are going to link, it shouldn't be across the board cross linking (such as all sites in the footer). Look at your sites as nodes, and all of your sites as a network. Connect them in a way that seems logical and organic.
Well after reading that article, I'd say that this issue is settled. Not sure why people continue to discuss it or argue about it.
Dont do it. I did it (innocent ) and got the site who linked to PR0 (from 4). Is still recovering (PR3) after 4 months. The linked sites doesnt got any penalty if they didnt linked on a site from the same ip. Also sitewides are a red flag, maybe 1 single link wont hurt. But anyway i dont want to test, i had enough testing
I do sitewides all the time, and my PR values are usually in the 4-6 range. There's no problem doing it at all. Google backs that up. If your rankings dropped, it was likely due to something else.
Those experiences by the individuals help us to understand what Google is doing, but I am still not convinced yet. Anyone has done more systematic experiments on this? I don't even want an educated guess any more. It's been too long
y'all keep saying sites linkin to ya oughta be "related" or "relevant". now, how on earth would google figure that one out? taint no quantifiable parameter. ya reckon google's algo thinks: oh, look here, site B links to site A, and site A is about shoes, because the word "shoes" is repeated 47 times, and it's in the "footwear" category at dmoz, but darn it, there's nary a "shoes" nor "pumps" nor "reebok" on site B, this link is bullshit, we ain't countin this link! and what about links from CNN or BBC? they are news organizations, aint relevant unless you make a special exemption and say, well, okay, CNN and BBC are relevant to EVERYTHING UNDER THE SUN but everything else has to be relevant, whatever that means. personally i think this relevancy stuff is another myth. if anyone has a matt cutts link on that, one way or the other, i'd be mighty grateful.
@jhmattern that doesn't prove anything. google are free to put anything on their guidelines, that doesn't mean they have the technology to check it. if you think about it, and how you might implement it, you realize that developing an algorithm to determine relevancy is a real bitch.
And they have the money and manpower to do just that, no matter how much of a "bitch" it might be. If it's in their business interests, they'd do it. And they have. If you want to naively believe it doesn't matter whether a link is relevant or not, no matter what they say, go ahead and live in your "cave." But then why waste your time asking for the info?
i'm not asking for info. i said that if there is a conclusive post by matt cutt's, lets end speculation and see it. you too are just as free as i am to believe whatever you want. obviously you don't care for a discussion about it, so why don't you go off and find some "relevant" sites to link hit for backlinks. meanwhile, i'd be interested to discuss whether it is actually possible to develop a relevancy checking system, and if so, whether google might actually have done so, with someone who has an interest in logical thinking and debates about technology and programming so forth. nick # did she say "manpower?" so you're going to fit every single one of the gazillion websites # out there into a category, using an army of, say, 500,000 "categorizer persons", and if # the backlinks spill out of the site's assigned category, there's a problem?
oh wait! there's no point in discussion and speculation, because the answer is right here: it's like dat cuz jhmattern says so, aight?
You asked for info decisively from Matt Cutts. The only reason that would matter is because he would "speak for Google." So go right to the source like someone with half a brain, and you'll see that Google already laid out your answer "decisively." If you don't like it, that's fine... but it doesn't make it not true. And for the record, I don't "hit" relevant sites up for links. I focus my link strategy on building quality content, and they come naturally... and frankly, it works. I don't have to worry about living and dying by Google and their rules. Try it sometime.
Well, personally I am part of "y'all", and I never said anything of the sort. However, I have in the past noticed patterns that do point to relevancy playing a role... just not as big as most people would think. Tell you what mono, if interested this is actually something you could test for yourself, and regardless of whether or not it turns out to be myth the test results themselves would make decent link bait. Set up a test scenario. Create 2 brand new sites, and link to each of them from about 10-12 pages, using a particular anchor text. Choose a low competition phrase, but not too low. Something with enough results where a decent sized differential in rankings would be noticeable. On Site A, link using focused anchor text, but from randomly themed pages. On Site B, use the exact same anchor text, but have whatever is in the anchor text also in the titles of the pages. If relevancy does play a role, you should be able to notice it. -Michael
ahhh.... the empirical approach! yes, michael, you do have a point, i suppose it is testable, though perhaps it would be better to do a clinical trial involving 30 sites + 30 sites, to weed out the statistical probability margin of error thingummy. of course, i would also need to be the sort to roll up my sleeves, or in this case put on my white lab coat, trim my moustache, and carefully mark the results on a clipboard, using a very sharp pencil. regrettably, though, i am not that type of person - i do hope a person who is indeed that type of person will make use of your very excellent suggestion, and come up with a set of conclusive results, measured, and graphed, and thereupon fashioned into a powerpoint presentation. now, i know that no amount of intellectual reasoning from a drunken lout like me can compensate for good old fashioned hands-on science, but my thing is to sit down at a bar with a tall drink and a pack of smokes, and expound ideas, using soggy paper napkins, and borrowed pens, to develop and demonstrate theories, for the benefit of, or perhaps more accurately detriment of, the hapless soul unlucky enough to be sitting next to me. what i would tell him is something like this: let's say you have a site about your cat, maybe a blog. every day you post about your cat, the pet food it eats, the kitty litter it uses, the hair balls it coughs up, the way it meiows, and the generally cat-like behavior it indulges in. your blog will also have lots and lots of photos of your cat, all of them identical to outsiders. i am sure we can call this a "cat site", yes? ok. so if you link from this site, to a company that makes kitty litter, this would be a "relevant" link, right? ok. the kitty litter company, possibly aided by the expert SEO skills of jhmattern, will be overjoyed to get this very relevant link. but what if, in addition to blogging about my cat, i also blog about my dog, a cowardly mongrel prone to fleeing at the slightest hint of danger? and my pet mouse, a bloodthirsty little rodent who loves sinking his fangs into my finger? then my blog would be a "pet site". ok, so the link to the kitty litter company would still be relevant, kinda. but what if, in addition to blogging about my pets, i also blog about my ex-wife and the insane things she's done? (we are, incidentally, not using a fictitious example here.) and what if i add a couple of posts about iraq, and china, and so on? ok, so this would be a "personal blog with a pet flavor" right? ok. but what if i also blog about stocks i like? the state of the world economy? (which incidentally will collapse soon, but never mind.) what you get is a hodge-podge site, just like digg, or stumbleupon, or, as i mentioned before, the most eminent BBC. consider also this, what if an analyst at a wall street firm writes in an article, at a widely followed site with PR8, that he likes a chemical company, an manufacturer of advanced weaponry, and a producer of kitty litter? the link from this article would not appear to be relevant AT ALL if you look at the keywords or the history of the site, which never ever in the past wrote about furballs, but of course the link is relevant, and probably carries more weight than any other link. thus, we can see, from a purely logical standpoint, it is just about impossible to identify or even demarcate this thing called relevancy. therefore, it is impossible to practically implement relevancy filtering. in fact, relevancy does not even exist as a useful concept, except in Google's Guidelines For (Hopefully) Keeping Webmasters In Check, where it means "a probable lack of total irrelevance." therefore, i postulate to those hapless souls unfortunate enough to still be reading this, it is quite impossible to develop a computer program which establishes relevancy to any measurable degree, and therefore google cannot check relevancy. if it did, it might as well go for the thuring test. nick
Nope, not even close. From your hypothetical scenario, we can conclude that in theory there are unfocused drunken bloggers out there who don't stay on topic. This does nothing whatsoever to prove the point that there is no such thing as a topic, or that Google couldn't detect what one was, or that if there were and Google did detect it as such, that similarly PageRanked links from same themed topics wouldn't carry more weight than unthemed ones. My point when disputing the "you must have similarly themed links" nazis has never been that Google doesn't count theme, but more to the point, the effort gained by obsessing over it does not equal the value returned. Turing test. Alan Turing. -Michael
chile, chile... what they say is, stay relevant. it doesn't say that the relevancy of links is a factor in the way sites are ranked in the serps. the wording is very vague, and that's intentional. a guideline is like a commandment. it is not a clarifying statement regarding the workings of an instrument. remember how your mommy used to tell you, "noooo jenny, don't stick your finger in nose!" right, that's like a guideline. but if she would have said, "jenny, every time you stick your finger in your nose, i am deducting 5 - 50 cents from your allowance, depending on how deep it goes," that would not be a guideline, but a clarification of the inner workings of something you are interested in. nick