LSI Doesn't Exist. So Stop Talking About It.

[Cataclysm1987]

For years, SEO's and internet marketers have been buzzing about this term and all its glorious uses to improve SEO rankings.

Yet, for all it's built up to be, very few people have ever heard from anyone other than themselves or each other that this concept even exists in any way.

In addition to that, Google has never publicly stated that they use Latent Semantic Indexing in their index. They've never denied it either, but let's look at the facts.

1. Latent Semantic Indexing translates into a search engine algorithm determining patterns of use and associations in order to find a common meaning among words.

This means that words like cat and feline would be considered essentially the same thing.

If this were true, one could manipulate LSI keywords in order to keyword stuff articles and blog posts beyond normal capacity. This is what marketers and SEO's swear by when they do articles or build links. LSI works.

2. Google has made no claims to confirm or deny the existence of LSI indexing in their algorithm. LSI deals with one of the most complex and open ended concepts a computer can comprehend: meaning.

Meaning is something taken for granted. The computational energy required to determine meaning is so overwhelmingly huge that it seems highly unlikely Google would integrate it into their index considering they have to reference billions of pages of content. They need the most efficient way of doing it.

3. The Google search function itself completely reverses the supposed claims of those favoring LSI indexing. If LSI indexing were really active in the search engine algorithms, you could type in a synonym and end up with almost the exact same search results.

Go to Google and type in the words cats and felines. Notice how other than Wikipedia which leads to different pages on the same sites, the sites listed are entirely different.

This doesn't explain how Google could possibly understand that the two are the same thing. If Google were smart enough to understand meaning, it would understand that these two terms are identical and that a person who types in either one is looking for the same thing.

But that's not the case, is it?

Yet the LSI facade continues...

Why LSI Came About...
LSI is merely the result of SEO's who have spent years working on a solution to create keyword relevant content by providing a ton of keywords into one article, as much as they could possibly handle.

People use LSI because of two reasons:

1. They believe it works. The placebo effect is so powerful that when two groups of people who both had minor knee problems went in for knee surgery, a control group wasn't given any kind of operation while the other group was.

The result?

Both groups got better.

People being convinced of something working is that strong. Because people feel like LSI works without any controlled, tangible way of measuring it, it causes them to work harder and feel like they've accomplished more. They haven't.

2. It forces them to use related keywords, which are a powerful part of the Google index. People are always crossing these two terms over each other almost interchangeably. LSI and related keywords are pretty different, but using LSI makes a person use related keywords almost by default.

This makes people keyword stuff their articles in a way that actually is productive for SEO purposes.

And, in fact, LSI keywords are often times considered related keywords because of their tendency to be used in the same pieces of content.

This doesn't make them LSI, nor does it prove that LSI even exists. It simply proves what Google already knows and leverages in the most efficient possible way.

That sometimes people just use the same damn words in the same damn paragraphs. It doesn't mean Google can read your mind.

Sorry if I sound mean. I am just sick of people who don't know what this word means throwing it around.

LSI is something that hasn't been effectively integrated into any modern search engine. We are years away from technology like that.

Just say what these words really are, which is related keywords. That's it. Nothing more.

If you disagree with me, please put some information directly from Google that conflicts with my analysis.

[xxxJamesxxx]

Mmmm I don't really know myself but it'll be interesting what others will have to say

James

[TZ]

Hmmm....I see someone has been paying attention.

Beautifully presented, and DEAD on the truth.

LSI is an old Red Herring thrown into a room filled with smoke and mirrors of the SEO guru world.

I'm very happy making a good living and being an SEO dunce.

Google is smart for sure, but it has it's limitations, and those limitations will always give smart webmasters a way to profit from publishing content en masse.

[Cataclysm1987]

Originally Posted by TZ

Hmmm....I see someone has been paying attention.

Beautifully presented, and DEAD on the truth.

LSI is an old Red Herring thrown into a room filled with smoke and mirrors of the SEO guru world.

I'm very happy making a good living and being an SEO dunce.

Google is smart for sure, but it has it's limitations, and those limitations will always give smart webmaster a way to profit from publishing content en masse.

Indeed. People are so afraid of Google too. It's like they're scared of this raw power that they possess, too scared to experiment or question the way things are done.

I just had to get this one off my chest. I'm really interested in hearing what people have to say about it.

[bgmacaw]

Overall, I think you're correct in that LSI isn't used in the mainstream ranking algorithms due to the data complexity and processing power issues related to it. I think they tinker around with it some as nerdy fun but it isn't something that's in production.

However, there is considerable evidence that suggests that they use more scalable statistical inference algorithms to determine ranking and related content. These are typically imperfect since they rely on prior data sets which themselves may be imperfect and changes to the data can cause unforeseen chaotic events.

[Biggy Fat]

If LSI is the clever way of ranking for another keyword in the context of your main keyword, then yes it exists.

[Cataclysm1987]

Originally Posted by bgmacaw

Overall, I think you're correct in that LSI isn't used in the mainstream ranking algorithms due to the data complexity and processing power issues related to it. I think they tinker around with it some as nerdy fun but it isn't something that's in production.

However, there is considerable evidence that suggests that they use more scalable statistical inference algorithms to determine ranking and related content. These are typically imperfect since they rely on prior data sets which themselves may be imperfect and changes to the data can cause unforeseen chaotic events.

Right, that's kind of what I'm saying. It has more to do with the referential integrity than LSI. LSI really just isn't out there.

Originally Posted by Biggy Fat

If LSI is the clever way of ranking for another keyword in the context of your main keyword, then yes it exists.

No, it doesn't. This isn't LSI. LSI specifically refers to Google processing synonyms and using them to determine rankings and keyword density by counting synonyms as a keyword you're targeting in addition to your main keyword.

What you're referring to is simply targeting more than one keyword. Totally separate from LSI.

[TZ]

Originally Posted by Cataclysm1987

Right, that's kind of what I'm saying. It has more to do with the referential integrity than LSI. LSI really just isn't out there.



No, it doesn't. This isn't LSI. LSI specifically refers to Google processing synonyms and using them to determine rankings and keyword density by counting synonyms as a keyword you're targeting in addition to your main keyword.

What you're referring to is simply targeting more than one keyword. Totally separate from LSI.

That is how I see it as well.

This is why I stopped "creating" content worrying about keywords and density years ago. It will only hurt us.

We need to create content that is natural.

One of my most lucrative posts on the web mentions the main keyword only once, and the post in over 1200 words long.

[Cataclysm1987]

Originally Posted by TZ

That is how I see it as well.

This is why I stopped "creating" content worrying about keywords and density years ago. It will only hurt us.

We need to create content that is natural.

One of my most lucrative posts on the web mentions the main keyword only once, and the post in over 1200 words long.

I've noticed the same thing on some of my sites, but then other times it looks like it's the opposite, that maybe intentionally manipulating keyword density and h1 tags is good for SEO.

I'm not sure who to side with. I think there are so many factors in SEO you could go with just about anything you prefer.