Forum Moderators: martinibuster
But I agree - if sites A and B get in trouble, then possibly C could be drawn in - how likely is that?
My intuitive assumption (no way I can prove this) is that we give G and others way more 'intellectual credit' than they deserve
"My intuitive assumption (no way I can prove this) is that we give G and others way more 'intellectual credit' than they deserve "
perhaps this is true, currently, but it all comes down to your risk vs benefit calculations, the SEs are getting and will continue to get more sophisticated over time, eventually these things will probably be incorporated.
The real challenge is the *rest* of the links going in and out of site A, B and C. For the links from site A to C to look natural (and/or be *worth* anything) A needs to have some "importance" and variance of it's own to pass on. so, you're not only spending efforts on SEO for site B, but also for site A, if you expect webmasters to see any "real" value in getting a link from site A. Site B needs links not connected to site A and site A needs links not connected to site B, etc etc for the pattern to truly look natural.
Can three way linking work? For sure. Can it be engineered to look natural? Of course - but as with any technique attempting to "game" the engines, do it knowing that *if* it gets detected, you may take a hit, or a devaluation as a result and that the "engineering" needs to be done by someone with experience and above novice to intermediate level.
My two cents anyway...
This would indicate some sort of automated inter-domain linking, hundreds of links between sites, will be detectable at some level. I think he's talking about link exchanges in a to b, b to c patterns on a smaller scale, as in replacing traditional a to b, and b to a reciprocal linking.
If a links to 100 sites and all those 100 sites link to b, that is pretty detectable, but if each webmaster has more than 2 or 3 sites and some are related this becomes more organic.
>>> but if each webmaster has more than 2 or 3 sites and some are related
The same could be said of recip linking - if people do it right. In my opinion, this type of triangular linking isn't much less detectable than recip linking - done well and with variance and a good set of other links, its great. Done solely for SEO benefit, lack of knowledge and with greed it can be an ugly thing for a site down the line.
I'll agree, I also think it is less important to work with what the SE's "can" do "now" in detecting "spam", but with what they would do if they could, because the technology will eventually catch up.
From practical point, if I were to implement this,I think I would be safe. For one, it wouldn't be just sites A and B rather A,B,C,D,E,...K, so I could achieve a greater degree of randomization. In addition each site would have a whole bunch of other kinds of links as well.
However, I must take issue with the statement that this is as easy (computationally non-intensive) as detecting reciprocals. For recip. algorithm is very straightforward:
for each link on site A, check if the the linked sits is linking back. Based on percentage of these -> devalue the links
For A-->C, C-->B, it is not that easy.
What you are really looking for is this pattern (if I understood your post correctly)
A-->C1, C1-->B
A-->C2, C2-->B
A-->C3, C3-->B
...
So, For each link comming from A
check all the links coming out of those linked sites
and THEN see if there is a large number of links pointing to the same site (B)
First algorithm is linear, second quadratic - a huge difference.
And nothing prevents you from putting another site in the chain...
Feel free to poke holes in this logic ;-)
It is the combination you put into place - the strategic plan - that combines all of these factors and makes it look as close to a natural and organic set of backlinks as possible that is the difference between being "detected" and then "devalued" or not.
Almost every method of gaming an engine is detectable when looked at as an individual factor - or when done "wrong" - without a plan. Types of links, anchor text, internal linking...
So, as I said before, triangular linking is certainly detectable in and of itself. It is when you add the knowledge to put the method to good and skillful use that it goes without detection.
I am sorry if I seem to be nit-picking over sematics here, but I would say that it is quite a bit less detectable, given the computational complexity of detecting it. Now, if my assumption on the complexity of the algorithm is wrong (which it very well may be), I would like a comment on that. Blanket statements like 'recips. can be detected' or 'triangular linking can be detected' aren't really that useful - while they sure are true. I definitely think that you are correct in stating that all of this has to be used in measure, not get link-greedy etc.
What I would like some comments on is basically a risk-assessmnet - side by side comparison. It seems that triangular linking is harder to detect, therefore it is less likely that SE will use computational power needed to detect it, therefore it is less risky than recip. linking and therefore it may be advisable to use it in larger percentage than recip. linking.
And, once again, sure you have to mix it up with other kinds of links.
As far as not being advisable for novice-intermediate linkers to use (myself included) -- well, discussions like this should help one learn more, shouldn't they?
given the computational complexity of detecting it
First, it's not computationally expensive. It's a straightforward SQL query on their existing data.
You assume that Google doesn't have the power to say "OK, show me all combinations of twho sites that have more than 100 links pointing to or pointed from a third site". The damned people can take satellite photos of my house; you'd better believe they can figure out how many degrees of seperation two sites sit from each other.
There's a mathematical formula for this, I just don't remember it at the moment.
Second, the question you're really asking is "is Google actively doing this, or only when they have a need?"
I don't think they're doing it on a regular basis (meaning on a scheduled process). I have seen evidence that they do, in fact, do it, and I'd bet they have a handy tool which lets them do it easily once they have a reason to do so.
I also think that when it's found, it's probably viewed as being much, much more suspicious than reciprocal linking, as everyone (even Joe's Bar) does reciprocal linking now. Triangular Linking = Intent
That you'll have to do on your own. Trial and error. You'll spend years doing it and as soon as you have the current recipe mastered, it will change again.
>>>well, discussions like this should help one learn more, shouldn't they?
Yes and no. There is no replacement for earning your own knowledge.
IMHO, these boards are a place to learn and a place to share, but there are limits on what people are going to share. Most advanced link knowledge is accquired through a lot of experience, a lot of testing and a lot of competitive analysis.
To expect someone to share hours, months or years of research and put it all into a message board post for all to see is not a reasonable request. Especially when any in depth knowledge is what makes the value of a particular SEO high (or not).
Take the "broad" opinions and statements you see here and test them - create ten sites using a variety of linking patterns - see what ranks, what doesn't, what has staying power, what dies off, etc - and form your own opinions and knowledge bank. SEO is largely opinion based and even "experts" disagree with each other on various tactics.
Then attend some conferences (or make some friends via stickies) and find people who know what you don't who could benefit from what you do and exchange to further line your arsenal. Just my opinion anyway.
But as you state further down - only when they have a reason to do this, which implies that they wouldn't do it automatically (because it is too expensive - what other reason?) , as they (presumably?) do recip. links.
"...that they do, in fact, do it, and I'd bet they have a handy tool which lets them do it easily once they have a reason to do so."
once they have a pool of suspect sites - then sure, they can do it really fast; I have no reason to doubt that
"I also think that when it's found, it's probably viewed as being much, much more suspicious than reciprocal linking, as everyone (even Joe's Bar) does reciprocal linking now. Triangular Linking = Intent"
Now, this is something I certainly can't disprove, and based on your experience I will just take your word on it. Personally, it seems to me that recip. linking shows intent just as much.
G. is just going to be more enraged because triangular links are a bit more hidden? (it may very well be the case ;-)
I understand what you are saying, and am not expecting this (sorry if it came across that way). But it's a sliding scale - there is something in between the most general statements that have been rehashed many times over and the advanced knowledge that nobody in their right mind would share.
Maybe I am still asking for too much...
I certainly will be doing my own research (just hired a full time linker), and will eventually find out what works and what doesn't.
Anyway, thanks for all the responses; I didn't mean this to deteriorate into one of those whiny nobody-wants-to-help-me threads.
And the full time link builder to test your theories is a smart move. Do more aggressive testing on sites you can expend and do the safest of the safe with the sites that you rely on to feed your family. ;-)
Has anyone had more success with triangular than recip?
Competitive intelligence wise, it would also make it harder for a competitor to track which of your links are reciprocal and which are one ways and which are actually triangular in nature.
They definitely work, as do reciprocals and one ways - we all know what the valid ingredients are - it is the recipe that makes them taste well when combined that is the aspect to focus on - again, in my opinion.
The key is to vary your linking. That is how you fly under the radar. Of course they can spot networks, but if it amounts to 5% of your links, it probably won't be seen as negatively if it was 99% of your links.
One way from directories
One way from resources pages from folks who don't really care about BL (i.e. edu sites)
one way from submitted articles, pr releases
one way - purchased
reciprocals and
triangular links (solicited or responding to solicitation)
decent networks such as dp, lv and adcaliber
Care to add a few?
(beside blog / guestbook /log /wiki spam)
Sincerely,
Stephen Betzen
<snip>
[edited by: martinibuster at 8:31 am (utc) on Oct. 16, 2005]
[edit reason] URL/self-promotion. [/edit]
Don't forget there are other things to be aware of besides just "type" - even the same types of links are not made equal. Lots of additional things to look at... IP ranges the links are coming from, inbound anchor text, placement of your link on the pages, what page on your site are they linking to (home page or deep links), how many outbound links the page you are being linked from has, how many times that domain links back to you...
I guess this is a good start of another 'list' - if we call the first one SOURCES, then this would be:
QUALITY OF SOURCES
==================
IP ranges the links are coming from
- use on-line C class IP checker?
inbound anchor text
- to this I would add theme of the linking page
- possibly theme of BLs (anchor?) to the backlinking page
(or is this going a bit too far?)
placement of your link on the pages
from best to worst (?)
-within the text
-few links at the footer
-just plain links page (top)
-just plain links page (bottom)
what page on your site are they linking to (home page or deep links),
- if you really want your visitors to land on the
home page, is there any advantage to deep links?
how many outbound links the page you are being linked from has
- I have heard different numbers being thrown around
but this seems hard to quantify. Intuitively, I
would rather be on a professor's resource edu page
with another 50 links, than on a typical recip
page with 30 links
how many times that domain links back to you...
- similar to IP issues
Also - post-linking quality check
- sneaky non-PR-passing links
- webmasters disconnecting the link page.
- domain changing owners, becoming part of bad neighbourhood...
A1-->C1, C1-->B
A2-->C2, C2-->B
A3-->C3, C3-->B
A4-->C4, C4-->B
A1-->C5, C5-->B
A2-->C6, C6-->B
A3-->C7, C7-->B
A4-->C8, C8-->B
where AX may be a range of 20 or so websites, in different categories, categories most close to the CX invited site.
or
A-->C1, C1-->B1
A-->C2, C2-->B2
A-->C3, C3-->B3
where we link out from a one site to get links to lots of different sites. we havent' tried this with multiple A's and B's in the same campaign, and I don't expect we will, just because of project size, and logistics.
Yes, reciprocal and 3-way linking is detectable. However, I would suggest that not only can the search engines detect them, but they do have this as part of their routine algorithmic calculation. If you rise above a threshhold, WHAM! I just think the dials are tweaked low to take off those sites that go overboard.
The fact is there are hundreds of thousands of sites (if not millions) that engage in reciprocal linking and they do help the search engines find and footprint sites. These sites help the engines more than they hurt it, that's why recips still work.
One thing about recirpocal linking is that once you have completed a decent campaign and your site is linked up good and you are showing decent PR, a lot of other sites conducting campaigns will link to you in the hopes that you will reciprocate. I find that the better place a site is in ranking, and the more backlinks it has, the higher the ratio of these one-way links to actual reciprocated links.
While it is true that most of my sites that rank do so almost entirely, initially, due to reciprocals, over time they also build up a large number of one-ways that were established in the hope that I would reciprocate. We do reciprocate to about 10% to 15% of these requests, to those sites that practice linking discretion. These one-ways dilute the ratio of reciprocals to overall inbound links. You might even say that one of the best ways to get one-way links is to have a high quality reciprocal link directory.
As for the computational expense of any of this, I think it is minor. Linear algebra and matrix calculatioins cover it and that's the way search engines are set up today, to do those type calcs. Each site serves as a node, that can easily stand out if certain parameters are met.
There is an intent with 3-way linking. It almost screams "we're trying to game the engines and are trying to keep from being detected at it". However, on the other hand, it can be used as a viable and perfectly legitimate means of promoting a site. For instance, if I have Joe's Tire's and Wheels, and later I develop Joe's BBQ, I might offer coupons at Joe's Tires for Betty's 5 & dime if Betty will offer coupons to Joe's BBQ, but from what I've seen there is little application of legitimate 3-way linking. Joe's BBQ doesn't have any traffic to send Betty because it is new, but I can send her plenty of traffice from Joe's Tires, to make it worth her while to send traffic to Joe's BBQ.
The prime reason 3-way linking is done is because the site owner feels it will degrade his site to put a reciprocal link directory on it. People seem to think of sites with link directories as trailer park sites, as somewhat trashy.
As for 3-ways being more difficult to do than direct recips, I have been of this school of thought and still am, except for the fact that we are doing a 3-way campaign for a site now that is getting a 35% link back rate from invited sites. This is a higher rate than we've ever gotten from a direct reciprocal campaign. I have my suspiciopns about why, but as sugarrae said, you can't expect me to post it all.
The key to making any of this work is balance and discretion. You really have to be careful who you link to.
The prime reason 3-way linking is done is because the site owner feels it will degrade his site to put a reciprocal link directory on it. People seem to think of sites with link directories as trailer park sites, as somewhat trashy.
That is the link dev statement of the week.
Many site operators dislike recip link directories for a variety of reasons.
Is it detectable? That's a great question. I agree with neuron that some people are getting away with it and others are not and the difference is probably on how high the knobs are tweaked to detect it. I suspect, as neuron implies, that there might be too much collateral damage if the knobs are tweaked to the level where you'll catch the most sophisticated triangulars.
Something to discuss with the search engineers from all the search engines at next week's Pubcon, eh?
hahahaahhaahhaah - dude, don't *do* that - soda all over my keyboard.
