40 replies to this topic

[Jill]

My point stands that it is possible to be penalised WITHOUT spamming

It shouldn't be though. And that's something that the search engines (I would imagine) work very hard to make sure doesn't happen. Of course, we all know it does, but I'm sure the search engineers really hate when that happens.

This nofollow on paid links thing if you view it the way Alan is saying, i.e., that if you have paid links and don't know enough to put nofollow on them it could make your site more difficult (or impossible) to find in Google, could potentially end up throwing out more babies with the bathwater than less.

I personally think it's really just Matt Cutts blowing smoke and that Google would never be so stupid as to actually penalize sites indiscrimately if they have paid links on their site which do not have no follow on them. But he does seem to be saying that's exactly what they will do. (If they do, they're idiots, imo.)

[Alan Perkins]

Mike, whether or not it's spam is moot. It's ironic that you should introduce bad neighbourhoods into the debate, since THEY are the reason that rel=nofollow could be said to have been invented. It's the fact that Google is demanding/requesting that people use rel=nofollow to label paid links that's the issue. Try filling in this table; remember to take account of how Google, MSN and Yahoo treat paid links and rel=nofollow:

CODE

+-------+------------------+--------+----------+------------------+
| Link  |  Flows Pagerank  | Paid   |  Trusted | Use Rel=Nofollow |
+-------+------------------+--------+----------+------------------+
|  1    |         No       | No     |  Yes     | ?????            |
|  2    |         No       | Yes    |  No      | ?????            |
|  3    |         No       | Yes    |  Yes     | ?????            |
|  4    |         No       | No     |  No      | ?????            |
|  5    |         Yes      | No     |  Yes     | ?????            |
|  6    |         Yes      | Yes    |  No      | ?????            |
|  7    |         Yes      | Yes    |  Yes     | ?????            |
|  8    |         Yes      | No     |  No      | ?????            |
+-------+------------------+--------+----------+------------------+

Not easy even for a seasoned SEO! So what's the average Webmaster (assuming they have even heard of rel=nofollow) supposed to make of it?

[Alan Perkins]

Wow, that table was a real conversation killer!

[incrediblehelp]

From a Google perspective maybe:

CODE

+-------+------------------+--------+----------+------------------+
| Link  |  Flows Pagerank  | Paid   |  Trusted | Use Rel=Nofollow |
+-------+------------------+--------+----------+------------------+
|  1    |         No       | No     |  Yes     | Noooo            |
|  2    |         No       | Yes    |  No      | Yeeees           |
|  3    |         No       | Yes    |  Yes     | Yeeees           |
|  4    |         No       | No     |  No      | Noooo            |
|  5    |         Yes      | No     |  Yes     | Noooo            |
|  6    |         Yes      | Yes    |  No      | Yeeees           |
|  7    |         Yes      | Yes    |  Yes     | Yeeeees          |
|  8    |         Yes      | No     |  No      | Yeeees           |
+-------+------------------+--------+----------+------------------+

If your asking to fill it out just from my own opinion then:

CODE

+-------+------------------+--------+----------+------------------+
| Link  |  Flows Pagerank  | Paid   |  Trusted | Use Rel=Nofollow |
+-------+------------------+--------+----------+------------------+
|  1    |         No       | No     |  Yes     | Noooo            |
|  2    |         No       | Yes    |  No      | Noooo            |
|  3    |         No       | Yes    |  Yes     | Noooo            |
|  4    |         No       | No     |  No      | Noooo            |
|  5    |         Yes      | No     |  Yes     | Noooo            |
|  6    |         Yes      | Yes    |  No      | Noooo            |
|  7    |         Yes      | Yes    |  Yes     | Noooo            |
|  8    |         Yes      | No     |  No      | Noooo            |
+-------+------------------+--------+----------+------------------+

[projectphp]

Hehe, went away for two days, so I am still in the conversation

Offtopic

...bad neighbourhoods into the debate, since THEY are the reason that rel=nofollow could be said to have been invented

Wasn't it to stop auto-spam of blog comments?

Spam is very much at least a part of the issue Alan. Your arguments have revolved around this quote (from the blog in your sig):

Since 2001, Matt and I have communicated regularly and have always seen eye-to-eye on issues of spam. That is, until now.

So the issue, as you have communicated it, is in your eyes about spam. Maybe I have that wrong, and/or maybe you want to change that opinion, but which ever it is, that was a big part of the argument to date.

I can accept, and have argued that, rel=nofollow is stupid, for a whole raft of reasons. I will not, though, submit that spam is the reason it is stupid, nor would I say that selling links makes one a spammer

Not easy even for a seasoned SEO!

And a good example of poorly thought out concepts and ideas, and a very good reason to avoid using the tag at all, especailly in unintended ways suggested by SE FUD campaigns. I say, in all instances, except for auto-generated content, and even then I waver, rel=nofollow is the wrong tool.

So is that the issue here, that clarity is lacking? If so, I think that is a fair criticism, but probably extends beyond just this specific incident.

My point stands that it is possible to be penalised WITHOUT spamming

It shouldn't be though.

I regret using the word "penalty", because I think it has the wrong conotations, and ads {typo joke for nerds} to the FUD. A better phrase would be "loss of trust". One term (penalty) implies damaging someone, the other implies you didn't get your full dose. I think it is fair to assume that you can lose trust without realising it.

Lets oversimplify and say that an SE wants to send users to the sites that it trusts to provide its users with the best, most reliable information. Trust in a source takes many forms and, ideally, a perfect SE would account for all of these. Some trust elements, like design, are nigh on impossible to quantify. Others, such as grammar and spellign, are hard to run algorithms for that provide meaningful, actionable feedback, especially at a page level and when there are many languages involved. Others, such as link analysis, are long winded to run, but possible and, currently, pretty useful.

When it comes to trust, the question arises: is a site selling links to unrelated PPC (Pr0n, pills, casino) sites likely to be more or less trustworthy? Can this one, isolated piece of data be used, as part of a larger set of data, to help define "trust"? I would argue that this data is useful, and such an action makes a site less trustworthy, beacuse sites that are trustworthy (The NYTimes, charities, government sites etc) are not likely to link in such a manner.

The fact you do not know that you are doing something to reduce trust is neither here nor there. You have done something that causes a loss of trust. And if you do know that certain actions reduce trust, to users as much as SEs, then you are making a conscious choice that you have accepted all consequences for.

rel=nofollow is an attempt (arguments about success aside) at disclosure. Disclosure, in most situations, is a substitute for blind trust, or lack thereof, i.e. the need to disclose who gives you money for a presidential campaign is an attempt to overcome teh efects of hidden bias. It could be argued that disclosing a link as paid negates the loss of trust that comes with selling links.

I would argue rel=nofollow is an aweful implementation, but that is separate to the concept of disclosing links, even with the concerns that the change from "do for people" to "do for SEs" creates.

[Alan Perkins]

From a Google perspective

Is it possible to produce a table that works for Google, Yahoo and MSN? Bear in mind that these engines actually treat the tag differently. For example, Google will not follow a link that is marked with rel=nofollow; Yahoo will follow it but not assign any link weight to it. If you used rel=nofollow on every link in your site, and the only outside link to your site was to the home page, then Google would only index your home page; whereas Yahoo could index your whole site. So you have to be very careful when using rel=nofollow that you get the desired effect.

BTW I don't agree with your Google perspective since it seems that, in this context, Google does not care about links that don't flow Pagerank; So when I say "paid links" it' pretty safe to add in your head "paid links that flow PageRank and attempt to game Google's rankings". - Matt Cutts. It's up to you to determine which links don't flow Pagerank (and which links don't attempt to game Google's rankings). Hint to SEOs: the links that don't flow Pagerank are those that can't be followed by search engines.

Offtopic

Offtopic

...bad neighbourhoods into the debate, since THEY are the reason that rel=nofollow could be said to have been invented

Wasn't it to stop auto-spam of blog comments?

No, it doesn't stop autospam. It stops the links in the autospam lending weight to the recipient or harming the reputation of the source.

Spam is very much at least a part of the issue Alan.

Fully agreed! But I said it was moot, i.e. debatable, but further debate would be pointless.

Suppose somebody offers to pay me to link to their site. I check their site and think it would be OK for people who visit my site to visit their site too. I place the link on my site, fully labelled with anchor text that describes to my visitors exactly what they will see if they follow the link; no deception involved. The link is fully visible to all visitors to my site; no invisible pixels or any shenanigans. I label the link using the words "Sponsored Link" on the page. What have I done wrong, exactly?

Ah, silly me. I should have known that Google existed, that something called Googlebot would follow the link, that the link would "flow" something called "Pagerank", and that, as a result, the site I linked to might appear higher in Google's results for some phrase than a different site that I didn't link to, and that would be a "bad thing" even if the site I linked to was better than the other site which I didn't link to. Whew. I've got it now.

[Lakshmi Narsimhan]

...I personally think it's really just Matt Cutts blowing smoke and that Google would never be so stupid as to actually penalize sites indiscrimately if they have paid links on their site which do not have no follow on them. But he does seem to be saying that's exactly what they will do. (If they do, they're idiots, imo.)

I accept your statement Jill. If Google happens to actually penalize sites indiscrimately if they have paid links on their sites, then there would'nt be any informative sites such as Webopedia.com etc., listed on Google SERPs.

[projectphp]

Offtopic

Offtopic

Offtopic

...bad neighbourhoods into the debate, since THEY are the reason that rel=nofollow could be said to have been invented

Wasn't it to stop auto-spam of blog comments?

No, it doesn't stop autospam. It stops the links in the autospam lending weight to the recipient or harming the reputation of the source.

Really? I thought it was to stop the link pasing PageRank, thus stopping the benefit autospammers received. this was a (failed) attempt to stop them spamming, as the incentive to do so was, suppssedly, taken away. Can you point me to something that says it was to protect the reputation of the source, because I can't find such information.

[quote]What have I done wrong, exactly?[./quote]
And who says they filter those links, and who says that anyone views them as wrong? If there are two camps, black and white, right and wrong, spammer or not spammer, then maybe sucha scenario fits Google's definition of wrong more than right. But who ever said there were two camps?

The expression "the straw that broke the camels back" is a good one for this situation. No single straw is ever going to break a camel's back, but the 1,000,000th might.

If you sell one ontopic link, I doubt that is a major concern. If you sell two, three, four or five, ditto. But at some point, I think we can all agree you become a link farm, and if you don't agree with that, lets make the number of links sold an extreme 1,000,000 per page. Anyone see a legitimate reason for that? So working backwards from 1,000,000 to 1, there is either a distinct boundary, like the age of consent, or a wihy washy blurry line, like any laws that require interpretation. I have no idea where the line is but, like pr0n, I know it when I see it, even if I can't define it for you

Secondly, is the issue that Google's penalty, which until this point is a FUD inducing bogey man without form, is too harsh? Or is it that using links as a means to downgrade is wrong fullstop? Or that providing people with tools they ask for (a way to disclose links) is wrong? Or is that an unknown person without knowledge of Google should never be able to make an unintentional mistake?

IMHO, from the second link analysis was used, the game changed greatly, and suddenly external influences started to play a role. To assume that the links you provide to other sites wouldn't be a factor in your own rankings is unrealistic, and if both sides agree they want dislosure for such things (rughtly or wrongly), if the best response isn't to provide it, why?

[Alan Perkins]

And who says they filter those links, and who says that anyone views them as wrong?

I guess it depends how you interpret Matt's blog post:

Q: Now when you say "paid links", what exactly do you mean by that? Do you view all paid links as potential violations of Google's quality guidelines?
A: Good question. As someone working on quality and relevance at Google, my bottom-line concern is clean and relevant search results on Google. As such, I care about paid links that flow PageRank and attempt to game Google's rankings. I'm not worried about links that are paid but don't affect search engines. So when I say "paid links" it's pretty safe to add in your head "paid links that flow PageRank and attempt to game Google's rankings."

Now there are a lot of subjective (i.e. contentious) words in there, like "quality", "relevance", "clean", "attempt", "game" and "affect", but I interpret Matt's comment as saying that the example I gave above would be an example of wrongdoing, even though there is no deception whatsoever, because the link would be paid for; would not be labelled with rel=nofollow; would flow PageRank; could therefore "affect" Google's rankings; and, therefore, could be viewed as an "attempt" to "game" those rankings.

The only imagined deception is that the link was paid for and not labelled as paid for, but (in danger of repeating myself here) rel=nofollow is not designed to label paid links; nor is it appropriate; nor is it necessary. I see things exactly as Matt himself commented last year:

In an ideal world, nofollow would only be for untrusted links. Let's take the example of a forum that wants to avoid linking to spam, but the same advice applies to wikis or any other web software. If an off-domain link is made by an anonymous or unauthenticated user, I'd use nofollow on that link. Once a user has done a certain number of posts/edits, or has been around for long enough to build up trust, then those nofollows could be removed and the links could be trusted. Anytime you have a user that you'd trust, there's no need to use nofollow links.

[projectphp]

I guess it depends how you interpret Matt's blog post:

LOL. I would rather interpret Shakespeare myself

But again, at the point of repeating myself, I don't think just right and wrong are the only options, and a view that is binary makes decision making harder. PageRank is a positive form of building an analog ranking symbol. I wonder if

I also agree entirely that rel=nofollow is not the correct strategy for disclosure, but then the question is do you have an issue with machoine readable disclosure fullstop, or do you think there is a form of machine readable disclosure that would be acceptable? Seems to me IF there is some selling or buying and/or selling penalty and IF people know about it and don't want to suffer, wouldn't there be the base demand that deserves some solution?

[Alan Perkins]

Your questions are very like Dan's earlier. My answers:

I sell advertising without disclosing the relationship, so they can't trust me.

No I don't agree with this at all. The fact is that you have no means to disclose the relationship other than to write something like "Sponsored Links" in the content itself.

rel=nofollow is not a means to disclose paid links. Google is trying to subvert it to that, but it wasn't ever designed for that, and it doesn't do an adequate job of identifying a paid relationship, especially when its proper role (identifying untrusted or unverified links) is still in use by both Google and other search engines.

As I recall, you were all for the FTC telling the search engines to label the paid links in their editorial content. I think we both still think that was a good idea.

Yes, definitely. And I think all publishers should label paid links clearly - to humans. I just don't think that rel=nofollow is an appropriate way to label paid links. It's a very blunt instrument that was designed to say "I 'm not sure that I trust this link", not "I have been paid (in some undisclosed way) for this link".

I'm not sure what the difference is here - is it the idea that someone should label paid links, or is it that Google shouldn't be saying to do it? Or maybe, you just don't like the specific implementation*? Don't you think that paid links should be disclosed? In an electronic media, isn't a machine-readable disclosure more useful?

Agreed, but rel=nofollow is not it.

I'd argue for machine readable paid link disclosure, but I could understand somebody arguing against it. It's fraught with problems, not least its reverse application to the billions of Web pages already out there. If the W3C, the FTC, and other authorities got together and invented a standard for labelling paid links, then the search engines could choose to weight labelled paid links accordingly. Maybe such labelling could include the relationship between the source and target of the link (supplier/customer, parent/child/sibling, partner, etc.), the amount that was being paid and the basis of payment (CPM, CPC, tenancy, etc.), none of which are encapsulated by rel=nofollow. AFAIK there has been no call whatsoever for such a labelling scheme to be developed, and it would be years off and very difficult to gain widespread (e.g. global) acceptance.