If you talk to Frank tell him I’ve been dying to meet him! ill in charleston, sc for weddings and am getting back to LA next week…big things cooking, would love to meet up with you and him…

I’ve called it the ‘inverted snake oil’ approach – cashing in on LSI crap like many have over the years – simply taking the contrary approach. There are so many bad facts and poor representations with the Stomper vid one wouldn’t know where to start.

As far as (Dr. cough cough) Edel Garcia is concerned, not many in the IR world care for his theories and they’ve busted him more than a few times… here’s the problem in simplest terms;

A. The purchase of Applied Semantics (2003 not 06) was for Ad Serving… thus their flavor of LSA – (LSI) was for the AdSense/AdWords programs.. You know, matching the ads to the content… it was never for the reg index SERPs.

B. The so-called LSI patents mentioned by Leslie, are actually part of the series of 7 Patents on Phrase Based IR…. nothing to do with LSI actually. Strangely, Anna Patterson came on board in 2003 as well (just after the Applied Semantics purchase).

So in the end, is LSI actually being used? Unlikely, it would be more likely PLSI, HTMM, LDA or the Phrase Based approaches are at play.

I was also saddened to see the statements made by Leslie and found it a tad misleading to be certain (left my own lengthy comment there). At first I felt like tearing it apart, but in the end… bah… it wouldn’t be the first nor last bit of SEO crap out there from folks with a lack of understanding how IR works…

As for behavioral metrics, better known as implicit user feedback, that’s
an entirely different can o worms and my own rant on that BS is here;
http://www.huomah.com/Search-Engines/Algorithm-Matters/The-final-word-on-bounce-rates-as-a-ranking-signal.html

Anyway… its late… just wanted to throw some thoughts into the ring – I enjoyed the post and the comments… thanks for that!!

Ciao

A part of the response I am writing to this so-called “controversy” is based on the same things Leslie discusses there. Not surprising, since we’ve been working together for years.

When you “silo” a web site, some folks say that you are doing something that causes Google to view your “theme” differently. I was one of those who actually bought that stuff, back in 2000 or so when the idea was new. Then I learned better. It’s a little amusing to see the old ideas recycled as new inventions nearly a decade later.

What you are *definitely* doing when you “silo” a web site is radically changing the way PageRank is distributed throughout the site. You can actually do the math on that – I have written my own tools to do that analysis, others have as well.

So we *know* that PageRank flow is modified, but some folks just ignore that, and attribute the outcome to LSI or themes. That’s kind of like a baseball player taking steroids, but attributing their improved slugging percentage to the inspirational poetry they read in the morning.

One is based on a known and well understood principle, the other is based on speculation, and can’t be proven wrong because it is not even testable.

Clearly Google has some sort of topical/theme technology available to them – as evidenced by the “site flavored search” and personalized (topical) search betas they rolled out around 5 years ago. But the treatment of topic was very course – roughly 80 total topics if I recall correctly.

For those who choose to silo and would like to have a sense of what’s going on, I’d suggest examining things like index penetration *before* you begin. If the number of indexed pages increases, that is a far more likely explanation for ranking changes – and this is based on known facts, not speculation.

If anyone has actually come up with a real testable hypothesis here involving theme magic, I’d love to know about it. That means you actually have a description of the mechanism that does whatever you think is being done. That means you actually have a model that makes testable predictions about things that can be measured, and it’s possible to discuss the validity of the ideas without you falling back on belief or quoting each other for social proof.