The Death of Spintax: Spun Text Detection Algorithm

Hi harvest316,

I think you are going "down the rabbit hole" with your reasoning.

Google isn't interested in detecting spun content, they are interested in filtering duplicate content.

Nearly all content is spun content. It is either spun by the human brain, or spun by software. There is nothing wrong with spun content as long as the content is of high quality, and that it is useful, relevant and used in a responsible fashion.

For the sake of diversity and purpose of conserving resources, Google will filter identical, or nearly identical results from SERPs and limit the amount of duplicate content that will be indexed. This is not a direction they have been heading in, they have been doing this since their inception. It's only your perception of this possibility that is new.

I agree that deep spinning on paragraph, sentence, phrase/word level is the correct method. but... No matter the depth of spin the first sentence would always be the most vulnerable right? Even If you have several variations of the first sentence with spun words it will start to be VERY similar. So if you blast out 100's of articles using the spun content and they all have a similar beginning wouldn't that be easy to detect? Not to mention the title. How many titles do you usually use?

Agree completely with this.

Reviving an old thread.

Spintax is fairly useless for producing spun content that is meant to be seen as different by computers. For documents meant to be read by humans, grammatically correct spintax can still be useful for marketing copy, like descriptions that will be spread across the web.

When read by software with a good LSI dictionary, it is trivial to reduce words with synonyms to canonical synonyms. Using your last example, a sentence with synonyms of "cat" can be constructed: "She saw the {cat|feline|kitten|kitty}." Spinning can produce four variations.

She saw the cat.
She saw the feline.
She saw the kitten.
She saw the kitty.

Using "cat" as the canonical synonym, all four sentences can be reduced to "She saw the cat." The differences are an illusion. Spun documents can have a large number of different words and appear different at a cursory glance by people but are easily converted to the same document by software.

To spin documents that will pass muster with both humans an computers, variations should be done at two levels, the word/phrase level and the structural level. Words and phrases can be varied with spintax. This can be seen as low level variation. High level variation can be done with sentences and paragraphs variants where one variant is chosen at random just like spintax chooses a word or phrase choice at random. For sentences this can be done with grammatical rearrangements like this:

After looking for cars, the chicken walked across the road.
The chicken walked across the road after looking for cars.

By adding or taking away information that does not change the meaning of the sentence, more variants can be produced:

The chicken walked across the road.
Across the road the chicked walked.
The chicked was wary of cars, so it cautiously walked across the road.

So a simple sentence has five variants. All variants are grammatically correct and would pass human review. Adding spintax to the sentences would produce more variation but it would be mostly for the benefit of human readers.

Paragraph variants offer the opportunity to dramatically alter the structure of a document. While spun sentences using the above technique have a one to one relationship with each other in spun documents, paragraph variants can be used to change the sentence relationships between spun documents. One paragraph variant could have three sentences while another could have five. The order ideas are introduced can often be rearranged without changing a paragraph's meaning. The result of paragraph varations will be a spun documents with a radically different document structure. They will appear this way not just to humans but also to machines.

Using these concepts I wrote software to do this type of spinning. I started out by creating an XML document specification to store documents with spintax, sentence variants, and paragraph variants. Then I wrote Java tools to spin the documents. Finally on top of the original Java code I built a GUI editor called whirlDOC. I guess I need to figure out how to do a WSO.

Exactly, using this more then 10 times already leaves enough traces for Google to detect when they have the processing power.

If you look at the effort put in to make it all readable it would be easier to just write those 10 sentences, especially when you type 100 words/minute like me and many other computer nerds

I think producing this tiny spin would take me at least 5-10 minutes and then you have the occasional error with misplacing a spin bracket. Not worth it at all.

In case you have a tool that is able to handle this flawlessly then it might be worth it.

All the stories about creating huge massive spun's based on 100+ unique articles. For who is it really useful? Most people already agree that they rather have true unique valuable content on their websites.

Everyone also seems to agree that mass link spam isn't working anymore so blasting 10.000 variations with automatic software building services is also out of the question.

There are only few people that I can think of who this might be useful for:

- massive churn & burners
- a very specific type of marketeer that I have in mind

Anyway kudos for you as you already said to take 100 unique articles and use a spun of each max 2-3 times which cuts the costs in half. Finally someone I can talk with

Paul

your right!!!!

I mean I truly don't believe its necessary to ever actually sit down and write new content when it can be..hmmm.. re-purposed.

there is a balance between university grade content and normal this is how I speak content.

so i dont see at what point google can actually penalize. you unless its blatant.. gibberish



look at wordai.. they spin content probably the same level as I would write

eddie

No I know exactly what it is... its taking someone's original article, putting it into a program and changing it around just enough so that it passes automated plagiarism detection. I also know that by the time the end product is edited enough to even make sense you could have just written the article yourself.

Its lazy, it's unsustainable as a business model and it genuinely takes away from the credit the original author should receive for actually writing the content in the first place....in a word...its theft.

HAHAHAHA

maybe if you guys dont re-evaluate what you consider "quality".

Some of the stuff Ive seen people post or link to on here would make my old English teachers have a heart attack.

I guess that discounts you from the possible candidates, being smart enough to bump a 6 month old thread.