Sunday, December 02, 2007

Are Paid Links 100% Untrustworthy?

In reply to Dave

> from reading your post you appear to be suggesting that Google
> allow pages to buy their way up the organic SERPs?

Not really. Unless you count as "buying their way up" something like hiring Scoble to write for you. Then people link to his stuff and the stuff he writes about and as part of that your content rises.

What I think is that it is not the most effective model to treat certain paid for content as completely untrustworthy. It seems to me the proper model to factor in the bias.

When Fox has news stories about their TV shows my personal view is that is a significant indication of bias their "news" (and of course, as I think any logical person would do it brings into question the merit of any other "news" presented). But I don't think they are 100% untrustworthy. When Some prestigious academic is funded by a drug company I believe they are biased but I do not believe the results are 100% untrustworthy.

If Matt writes about some neat new Google product I know there are biases - even if the only bias is he is much more likely to hear about a cool Google product than one from some other company because he works for them. But I figure if he says he likes some new Google that is worth factoring into my decision whether to look at it. And in fact I give him something like 97% credibility on doing so. If he thinks some new Google offering is junk I don't think he will say it is great. I just don't, I could be wrong but I am confident of that. Would the Fox site links to some new TV show that is horrible? Yes I do believe it would.

If some other blogger I trust highly writes on say Ruby on Rails and then says they are sponsored by the annual Ruby on Rails conference which they attended the last 2 years and it is great and they highly recommend it I find that valuable. And yes I do believe the best search result would be enhanced by giving value to this recommendation of this blogger even though it is likely they will provide more links to the conference due to the conference than they would have if they did not receive pay to do so. This certainly could be seen as buying your way up the search results.

I don't automatically ignore all advice where the author has some personal financial incentive or stake. That is one factor, but it is not a binary operation (0% trustworthy or 100% trustworthy). It seems to me to select the best search results you would need to apply the same logic. Granted I have no real idea how to do that. But I do have the opinion Google can figure out good ways to do that. And I think that would be the correct strategy.

I also think this is an interesting area to think about. How to find the best results for a search. How to reward behavior Google would like (say good title text) and how to punish behavior Google doesn't. But really those are just methods to the true aim which is the best search results. Google does it pretty well I think but I think they can do much much better. Which I think is pretty much what the people working at Google think. Users of the search engine don't care what the title text is (basically) they want the best pages for them. They don't care if they comply to Google's recommendations. It does make it easier for Google to find the best results if the title text is relevant but that is an issue for Google. So Google tries to help sites make it easier for Google to return results by providing guidelines - that is very sensible. And in the same way I have no problem with Google having guidelines on certain types of paid links.

I happen to think on this matter - that certain paid links are to be 100% ignored is not the correct strategy (to create the best search results). But I may certainly be wrong, maybe it is. I think the best strategy would be to rank the content and links on other factors. The pay is not the problem. Some bad practice that are encouraged by pay are the problem. So weigh those. If it is links to low quality sites then factor in any link to "low quality" sites. If it is unrelated links in a footer across a whole site, factor that in. If link keywords are the problem then attach that - whether they are paid for and how they are paid for does not invalidate them (as a source of useful data to rank search results), in my opinion.

Certain authors are much more swayed by payment. That again is a factor as a human we consider. If a great Ruby on Rails blog promotes certain gambling, mesothelioma and mortgage loans with no insight I lose respect for their opinions. Wether they were paid or not (but obviously pay is a likely reason for such writing. On the other hand, if they include a few off topic posts that are sensible I have not problem accepting that as more input to consider (probably I give a bit less weight than I would to posts on Ruby on Rails) but I still value it.

I don't have any problem with Google campaigning against certain paid links (I believe Matt has said several times the current campaign is against specific types of paid links other links that are more indirect [and which as a matter of course most often involve much more money] are not being campaigned against. I just find it interesting to examine the assumptions behind the current strategy.

Related: Google's Search Results - Should Factors Other Than User Value be Used - Google's Displayed PageRank - Improvement Ideas for Google (2006) - Web Search Improvements (2005) - posts on Google management practices

No comments: