While it’s never been a calm, cool, collected choir singing Kumbayah, since Google’s Panda hit, the Internet has become a maelstrom of boiling hot tempers and scammers claiming they can put out the fires.

To give you a bit of background, Google unleashed Panda (also known as Farmer) on February 24, 2011 in the US to try to eliminate or reduce from search results content farms, duplicate content, and other “bad influences.” Google’s stated goal was to keep low quality sites from ranking in their search results.

Rather than being a Google update like many others, Panda is software that Google runs approximately once a month, because to make it part of their real-time algorithm would not be feasible due to the intensive computer processing it consumes.

Panda was developed using a machine learning algorithm that took data entered by humans of their opinions of what a high or low quality site is, and then this data was input into a machine learning program that over time created profiles of good and poor websites, which were then compared against all of the websites on the Internet.

What resulted could be compared to an earthquake or hurricane on the Internet.

Many website businesses lost so much traffic during the Panda that they had to lay off people or even close their business. I read an interesting comment on one of the forums that stated “I was infected by Panda.” While this might seem funny, to those who lost their livelihoods due to Panda, it’s not humorous at all.

In later revisions to Panda, Google tried to scale it back to undo some damage, due to the tremendous uproar they encountered from webmasters, and subsequently, some webmasters did see a little return in traffic, but most who were hit by Panda have not recovered their previous traffic.

However, trying to undo a learning algorithm is a bit like trying to unlearn Algebra … do you forget the transitive law, for example, or make it only partially applicable?

I rather think Google in the more recent updates, or fixes, might have run a statistical curve on the input data and hacked out some of the data on the far ends, and then re-ran the machine learning algorithm. That’s what a mathematician would do. Whatever they did, Panda has taken Google search to a shadow of what it was

I’ve watched really high quality websites dive during Panda, and I’ve been dismayed at the poor quality of search results I get even today when I’m doing research about plants for my garden.

To be sure, indexing and ranking millions of websites is a formidable task. Doing it by hand when Yahoo! started would soon prove infeasible. Thus, software was employed and Google claims they use over 200 factors, from inbound links to keyword relevance, to order search results.

But even after all of this was done, at the point one enters keywords into Google and presses Enter, Google was not satisfied.

For many years, Google has fought “gaming,” that is, techniques webmasters use to gain a bit of advantage over their competitors to gain higher rankings. In addition, Google has exclaimed over and over they want a “better user experience” — they want the best websites to win the rankings. And so, it made sense that Google would try to find a method by which the best websites could win, based on user experience metrics, and thus Panda was born.

But the results of Panda beg the argument of can software really determine the best websites for user experience? Can human experience (note: I did not say, “intelligence”) be encoded into software to such a degree that it can discern a high quality website from a low quality website?

To some extent, I think yes. It’s the other extent that doesn’t work.

It’s ironic, really. Larry Page (co-founder of Google) recently became CEO of Google. While still a student at Stanford, Larry invented “Page rank” (named after him) or PR as it’s more commonly known (the little green bar), which attempted to order every website on the Internet on a exponential scale of 1 to 10, but as the Internet grew, this became an impossible task. The “orderer par excellence” helped usher in the most ordering filtering algorithm Google has ever deployed.

What happened? Well, Google did wipe out a lot of websites that are what we know as MFA (made for Adsense) sites — their only existence was to make money as an Adsense partner. By and large, they had poor content, they used every SEO trick in the book, and they made money from Google.

Since Panda was released, however, these websites took a dive (at least most of them). And then the websites who couldn’t get the organic traffic they had before, turned to Adwords, Google’s paid advertising. Paid Adword advertising revenues increased, while partner Adsense payout decreased, during first and second quarters this year, for a net gain to Google.

Was this chance or happenstance?

Was the increase in revenue the aim of Panda or simply a by-product? This is as much a deep, dark secret as Google’s algorithms.

{ 0 comments }