I studied engineering. I chose bioengineering, because it was the hardest most interesting. I got a master’s degree and then went for the Ph.D. For that, I chose electrical & computer engineering because it was the hardest most challenging interesting. My dissertation topic was in brain imaging. I chose brain imaging because it was so interesting, and brain imaging of the motor impaired because, well, their brains didn’t work right. More fun than looking at regular brains, eh? Well, obviously since I completed all-but-dissertation and never finished that dissertation, I achieved my goal of finding something that was too hard for me. Eventually I found SEO. It’s harder more fun than that engineering stuff.

Over at DaveN’s blog today I see a response from a Google employee that is telling. I don’t really know DaveN. Having only met him once (briefly) and never having been drunk with him, I’m sure he’d say he doesn’t know me either. But despite his bull dog behavior on the SEO forums, he does have connections at Google and gets Google people to reply to his questions. He asked why his page would appear to rank for a phrase that does not exist on his site, and which seems completely unrelated. Those are the good questions. Here’s DaveN’s question:

also where the fuck did …christine dolce naked….that search term come from …???

and here is Vanessa’s response from Google’s Sitemaps team:

My early morning, under-caffeinated guess is that you linked to this threadwatch story (http://www.threadwatch.org/node/7076) in your “industry news” section and at some point, that may have been on the same page as links pointing to this post: http://www.davidnaylor.co.uk/archives/2006/03/21/naked-truth-about-shoemony/ and possibly some anchor text pointing to your site includes the word “nude” (the cached page info seems to indicate so). And when searching for christine dolce naked became a popular thing to do, your site may have been an early one to have all the keywords.

Sure enough, Dave then discovered an instance of such a link to a story on ThreadWatch about Christine Dolce’s MySpace page, and can explain the nude thing as well:

Yep I agree.. yahoo shows the outbound link in the cache on one of the pages, nice spot vanessa … the nide link will have been when i did the Naked Bulling thing

Wow. Just when you thought it was safe to go into the water (queue the theme from Jaws). Google is saying that a link on his page to a page about Chrstine Dolce (even without the words Christine Dolce on his site at all), and a link on his same page to a page ON HIS SITE about Naked Bull Riding (perhaps influenced or at least re-inforced by the use of the word “nude” in the anchor text used by other’s linking to Dave as the Naked Bull Riding SEO), led Google to assign relevance to his site for the virtual semantic “Christine Dolce Nude”.

Since any celebrity+nude search is at least mildly competitive, it should take more than this to rank (Dave’s blog was reportedly #5 and #8 in the US for Christine Dolce naked and nude). Vanessa suggests that AGE of content and/or links played a role at a time when POPULARITY of the search “Christine Dolce Nude” was rising rapidly on Google. In her own words,

when searching for christine dolce naked became a popular thing to do, your site may have been an early one to have all the keywords.

Personally, I love it when Google insiders think out loud after just waking up, before coffee. Maybe it’s best not to compare this to pillow talk and competitive intelligence gathering - LOL, but nonetheless it does show Vanessa’s stream-of-consciousness thought line. Priceless.

Every SEO should now be tuning their thought process in accordance. Your page links outwards to other sites, thus voting for their relevance to your topic. They are popular/keyword rich, and carry a theme that is less site-specific than page-specific (ThreadWatch is not a celebrity blog, but an SEO and marketing blog). Your SITE contains pages that other’s link to from their own keyword-rich content, using anchor text. Google puts the words together and subsets them looking for overlaps. Absent any pre-determined ranking factors (in the case of a sudden rise in popularity of a search) the presence of subsets across that keyword conglomerate gives rank. We knew most of that.

but fine points separate the practical from the theoretical. Subsets matching and overlapping is something Google can do fast… some parts of LSI we know from linear algebra are able to be computed fast, and we see AdSense doing well with that all the time - it enables contextual advertising. Other parts are too difficult to do in real time, and sure enough Google falls back on links and anchor text, apparently with a heavy page emphasis. No time for imposing restrictions or a-priori knowledge about a site, so the traffic flows to DaveN and the Chrsting Dolce fans are disappointed until enough of them query Google the same way (and click through???).

I would have expected Google to already have a pretty good handle on Christine Dolce…. it’s not like she got her 1 million MySpace friends and a PlayBoy shoot by staying under the radar. But Google likes a query to be unique, and a website to be very, very specific. Christine Dolce Nude is aparently not as similar to Christine Dolce as many would have expected in SEO world. And I guess Christine Dolce naked and Christine Dolce Nude, once analyzed and stabilized by the G machine, don’t stay that way (stable) for long (?). Interesting.

So what triggers Google to re-visit the query and impose restrictions or weighting factors? That might lead to “when”, which reveals whether or not we can expect to practically turn this kind of opportunity traffic into a reliable stream (via redirects, links, etc). We all know how to get it. Shoemoney can tell you what a few days of #1 for “free (insert hot new cellphone model here) ringtones” is worth. Google could really use those analytics, eh?

I also think the day of a SEO cooperative is closer than ever.

As I waited the two or three minutes Wordpress2 needs to post a small edit to this blog, I wondered why I was so casual about ripping backlinks out of the WordPress templates I downloaded yesterday. That issue is blog-worthy, I think. So this time, I smartly opened a second tab before hitting “save”. So while WordPress takes another 2-3 minutes to update the post slug, I can blog about stealth links in open source software.

I’ll go back and flesh out the issue later, but let’s just say there are plenty of direct backlinks hidden inside these “free” downloads. Some time ago I helped expose a case of user agent cloaking hidden within a front end re-write ruleset for the Invision Power Board forum. In that case, the author had inserted a cloaking script into the front end of a mod designed to make Invision’s forum “search engine friendly”. It quietly inserted 5 or 6 backlinks to his own pop culture websites, so only the search engines would see them. Nasty. We got him to fix it, though.

Now WordPress2 comes with a ton of themes. Each one is a set of code files, and each enjoys ample opportunity to insert backlinks. I always go and remove sitewide footer links because they are clearly not justified (except perhaps with a nofollow…haha) but this time I found myself stripping out several aditional links buried in the code. Some were in sections marked “do not edit anything here”. Some threatened “if you touch anything here, don’t even think of asking for support”. That’s fair enough, but disclosure would be much more…. ethical?

Yawn. Maybe I will start digging and see just how many free hidden backlinks are working for these people. And how many disclose, how many seem to hide the links, or gasp… maybe some or encoded? A task for a rainy day?

Alex King has promoted WordPress themes on his site for years, and gets many submissions. From this post I see some have computer virus/worms embedded, and others have hidden links. I’m not sure what the review process is today.