Paid Link Reporting and the Fall of the Machine

by Aaron Bradley on May 29, 2009

in SEO

When Google introduced paid link reporting via Webmaster Tools in 2007, in marked the second time human informants were solicited to help the search engine. The first – less contentious – invitation a year earlier was the spam report, by which conscientious website owners (or their minions) could report such things as sneaky JavaScript redirects, doorway pages and even "misleading or repeated words." These share with paid links the characteristic that they are potentially hard to detect, and indicate that the machines crunching through all those web pages are still occasionally outsmarted by some upstart human.

One should note that the first call for webmaster intervention concerning link value – but by judgment, not decree – was early in 2005, when Google introduced nofollow. It's interesting to note the evolution of nofollow. It was originally intended specifically to circumvent comment spam. There is not a single mention of paid links in the initial announcement by Google, and at that time even the full stated scope was limited to "software that allows others to add links to an author's site (including guestbooks, visitor stats, or referrer lists)." It was not until September that Matt Cutts suggested that sites who "buy links purely for visitor click traffic, to build buzz, or to support another site" should use nofollow.

While not first, however, the potential impact of paid link reporting is large. As links are a widely-acknowledged and widely-observed ranking factor in Google, it stands to reason that defusing a competitor's link could be a powerful tactic. Moreover, paid link reporting marks more definitely a break with a solely algorithmic method of indexing and ranking websites. Google probably didn't and doesn't need the spam report to do a pretty good job of ranking websites, as the nature of most problems specified in the spam report lend themselves to being solved technologically. However, Google felt paid links an important enough problem that they made much fanfare of their efforts to thwart the insidious monster. And that in turn suggests all the machines in all the data centers in the world can't unmask a paid link that really doesn't want to be discovered. Because, of course, they can't.

An algorithm or machine intervention can't fully keep all paid links out of the ranking algorithms as long as links are factored in – and there's no sign Google is ready to give up on links anytime soon. Google values human-created links as "votes" for websites, and since there's no logistical problem paying a human to create a link, this is a tricky potential exploit. If I meet "Bob" down at the pub and slip him a fiver with an understanding he'll link to me from his site, Google will never be the wiser. Ever.

(As as side note, there's ample justification for challenging the blanket proposition that purchased links are not reflective of value. A wealthier web enterprise can buy more links than their poorer competition, just as a large company has more money to spend on creating the much vaunted "content-rich" site smaller enterprises can't afford. And this isn't exclusively a philosophical point, as third parties are now organizing link networks involving complex partnerships to legitimize the network's linking environment – and, at the end of the day, providing a mechanism by which sites with more money can rank higher than those with less. Again, is the ability to participate in that marketplace an indicator for website value? This is a side note because currently Google explicitly rejects any such value in purchased links.)

Needless to say, such clandestine encounters as those with "Bob" in pursuit of PageRank are (still) probably quite rare, where as there's all manner of link-purchasing marketplaces available (open, semi-clandestine, semi-legitimate, etc.). So it may be that knocking down the networks and larger schemes are Google's intent in seeking webmaster (or SEO) support. Does that mean that for even those larger paid link entities Google has a difficult time detecting them without human intervention? Or is it that an anecdotal truism in the SEO world – that a single great link from a great site will out-perform a 1,000 lesser links – is the Achilles' heel of Google's algorithm, and geeky toilers have figured out to defeat the machine?

One way or another the route Google's has chosen is not to let things stand, not to more-or-less let the technology catch up to the problem. While midnight link deals might slip through, one could envision the evolution of detection mechanisms based on things like page placement, anchor repetition, link distribution patterns, and so on. Doubtlessly Google is pursuing these machine solutions, but they don't consider such measures sufficient to do without human paid link reporting.

For a search engine company that's always prided itself on not hand-ranking, that poses a bit of a problem. Not simply a philosophical problem (though it is certainly that), but a technology problem as well. Once you have granted that human intervention is necessary in order to rank web pages better in the case of a particular exploit, the system of machine-ranked pages becomes further and further diluted. You rely more on people to make your search engine work properly. If Google is miffed as to how to counter paid links and started soliciting informers, is it not conceivable they'll encounter more machine-insolvable problems down the road? Now that SEM has become such a large industry, they'll undoubtedly be more, and more sophisticated, exploits introduced in the future. It will be interesting to see how Google responds.

For what it's worth I don't think there's an army of forensic link police in the Googleplex, buried under thousands of paid link reports (though I have no idea, one way or the other). That is, just how much difference are paid link reports (as distinct from paid links themselves) making on the rankings of websites, versus the impact of "second generation" paid links that get through? Because I suspect "not that much," it makes me wonder about the wisdom abandoning at least the appearance of a hands-off (literally) attitude in regard to organic search.

This also formally makes the SEO industry necessary for Google – recognizing, of course, that it's the same industry gave rise to paid linking in the first place. Once you start putting rules of conduct out there for search engine optimization practices, you need specialists to interpret and apply the rules, just as you can't expect someone who doesn't know HTML to create a W3-compliant website. This is even more the case when there are only guidelines that require interpretation, rather than explicit rules. They may well have been a time when John and Mary could throw up their merry little website and Google would reward them for their content and following, but Johnny might now in ignorance buy a link without nofollow-ing, or Mary might employ a bad programmer that inadvertently ends up cloaking.

I haven't discussed the ethics of link reporting here (a necessary dialogue, but not in this context), in order to focus on the implications for Google's search methodology. Google is left with the much bigger quandary of how to fight SEO spam tactics, and to a certain degree its broader direction. Does Google put more emphasis on direct human evaluations for website ranking – even if only as inputs into the algorithm – or do they continue to work out web behavioral patterns and let their machines figure out what's real and what's Memorex?

Previous post:

Next post: