Panda 3.3

[crazedspyker]

"Better to remain silent and thought a fool, than to speak out and remove all doubt"?

lol jk but sometimes its good to read before calling out an entire community.

 

[meatro]

it seems like they have figured out a way to tell spun content and trained their bot read like a human

They are nowhere near this and I highly doubt that even the next 15-20 years will bring around any type of technology like this.

Google is a machine, and the way it works is very similar to the CTRL+F find function in most document editors (Notepad, Word, etc.) It can find exact instances of your search term and about as smart as Google is, it can also display related terms and misspellings.

However, it is still very basic in the way that it handles data.. Which is why you heard a while back about Google employees actually checking sites manually. Because Google (the algo) is capable of FINDING the term, but how is a machine to know whether or not what its found the term in is legible, or makes sense?

In my opinion, the reason that people end up with garbage spun content is because they don't fully understand the impact of Google STOP WORDS!

Here's a good list of Google Stop words: http://www.link-assistant.com/seo-stop-words.html

What a stop word is, is a word that basically is completely ignored by Google. And you'd be surprised how many of these words are in your thesaurus!!

{none|not any} of {these|what I'm saying} {is|will be} {seen|regarded} as something {we've|I've} made {myself|ourselves}.

Every single one of those words is a stop word. So no matter how you spin that sentence, no matter how that sentence ends up... It doesn't matter to Google, because in its mind, those words are useless and mean nothing.

So if you are spending your time deciding between {I have|I've|I tend to} etc, stop. Because your time is being wasted, Google does not give a shit how you refer to yourself, it cares about the remainder of your content.

Another crafty way to unique up your content is through punctuation.. Google does not read punctuation, but it surely knows that it's there. For example...

"thank you for your time dr"
"thank you for your time, dr"
"thank you for your time, dr."
and even
"[email protected],your;time:dr?"

are all pretty much the same. Why? Because Google is a machine. And in order to FIND THE TERM, it must disregard the punctuation. Why? Because Google is a machine and does not understand the CONTEXT in which the term is being used!

All it knows is "ok, I've located the search term here.. To the left is this other term, the two are relevant to each other. 4 sentences in, I've found another term, relevant to the previous 2... etc. etc."

It does not know, "oh, wow, what a great article about fruit bats. I love how it's all laid out and their grammar is perfect! This definitely deserves top spot."

 

[Getwhatchuwant]

Interesting point about stopwords meatro, do you think that applies to domain names too?

 

[crazedspyker]

Maetro is right, machines are still a long way from this. However, most spun garbage not only has stop words, but most have horrible grammar! It's easy for google to flag pages with extremely bad grammar and send it to a manual team to check out.

You can see this happening on google webmasters. Google employees will sometimes actually point out spun content publicly for everyone to see. (this usually happens when people complain about their rankings. I've seen it happen a few times. Google will mention that the source is a bad rewrite/spin of a useful text and suggest the user to fix it up)

What surprises me is the employees were able to list the "origin" the spin came from even though the spin was "at least 80% unique"

I spend a lot of time on my spun content, most of my spun articles are 40,000 words pre-spun (1 problem I encountered is, some services/tools don't let you input spun articles this big) and come out to 400-700 words after spinning with virtually no grammatical mistakes. It's created from 20 unique articles spun at word, sentence and paragraph levels. Takes a long time to piece together but you can use that article for you niche almost indefinitely.

Dupe free pro shows most articles have 0.6% to 3% duplicate content from each other. The trick is to "quality check" it as you go. Check for grammar mistakes of each article individually at each spin level, fix up the mistakes, then combine all 20 articles together. (if all the spun articles are grammar free, its common sense the final article will be grammar free) End result is each sentence has about 60+ (assuming average 3 sentence variations per unique articles) sentence variations spun at word level.

What I'm saying is people need to start putting more time into creating high quality spun articles for their niches. It takes a few days to put together but you can be confident knowing your spun articles will pass all duplicate and grammar checks.

That being said, a bunch of my web 2.0s got hit hard by this panda(high quality spun as well as original unique web 2.0s), from page 1 to page 10-17 . Makes me wonder really hard if its the content or the scrapebox blast going into them..

 

[googlemonster]

I don't agree at all with grammar, as in that Goog-all knows what good grammar is, what about a 10 year old writing, or even a genius who can't spell properly etc. Since when can a computer read and critique somebodies' work? Hell even my math teacher couldn't spell for shit. People need to get over the fact thinking Google is the borg, they simply like to create this impression.

Let's take two shit sites ehow and livestrong, both appear on multiple queries, even though their content has spelling errors, is thin, and is written by people with passing knowledge on subjects.

The reason is because of their algorithms which determine importance. If these two sites were downgraded for literary prose, they'd rank 100+, but they don't because of Google's ranking factors. Just look closely. Their 'computer scripts' are not that hard to beat son.

 

[lancis]

Sorry for raining on your parade, mate.

Just some examples of what Natural Language Processing can do nowadays:

- Automatic summarization: Produce a readable summary of a chunk of text. Often used to provide summaries of text of a known type, such as articles in the financial section of a newspaper.

- Natural language generation: Convert information from computer databases into readable human language.

- Natural language understanding: Convert chunks of text into more formal representations such as first-order logic structures that are easier for computer programs to manipulate.

- Part-of-speech tagging: Given a sentence, determine the part of speech for each word.

- Question answering: Given a human-language question, determine its answer.

- Relationship extraction: Given a chunk of text, identify the relationships among named entities (e.g. who is the wife of whom).

- Topic segmentation and recognition: Given a chunk of text, separate it into segments each of which is devoted to a topic, and identify the topic of the segment.

- Text simplification: an operation used in natural language processing to modify, enhance, classify or otherwise process an existing corpus of human-readable text in such a way that the grammar and structure of the prose is greatly simplified, while the underlying meaning and information remains the same.

For example, try the MS Word thesaurus. Among all the possible synonyms it will present you mostly the relevant ones. Because it is able to determine correctly the semantic meaning of the word.

Analyzing text, especially English, is not that hard. And finding the common between two spun sentences is even simpler.

They are nowhere near this and I highly doubt that even the next 15-20 years will bring around any type of technology like this.

Google is a machine, and the way it works is very similar to the CTRL+F find function in most document editors (Notepad, Word, etc.) It can find exact instances of your search term and about as smart as Google is, it can also display related terms and misspellings.

However, it is still very basic in the way that it handles data.. Which is why you heard a while back about Google employees actually checking sites manually. Because Google (the algo) is capable of FINDING the term, but how is a machine to know whether or not what its found the term in is legible, or makes sense?

In my opinion, the reason that people end up with garbage spun content is because they don't fully understand the impact of Google STOP WORDS!

Here's a good list of Google Stop words: http://www.link-assistant.com/seo-stop-words.html

What a stop word is, is a word that basically is completely ignored by Google. And you'd be surprised how many of these words are in your thesaurus!!

{none|not any} of {these|what I'm saying} {is|will be} {seen|regarded} as something {we've|I've} made {myself|ourselves}.

Every single one of those words is a stop word. So no matter how you spin that sentence, no matter how that sentence ends up... It doesn't matter to Google, because in its mind, those words are useless and mean nothing.

So if you are spending your time deciding between {I have|I've|I tend to} etc, stop. Because your time is being wasted, Google does not give a shit how you refer to yourself, it cares about the remainder of your content.

Another crafty way to unique up your content is through punctuation.. Google does not read punctuation, but it surely knows that it's there. For example...

"thank you for your time dr"
"thank you for your time, dr"
"thank you for your time, dr."
and even
"[email protected],your;time:dr?"

are all pretty much the same. Why? Because Google is a machine. And in order to FIND THE TERM, it must disregard the punctuation. Why? Because Google is a machine and does not understand the CONTEXT in which the term is being used!

All it knows is "ok, I've located the search term here.. To the left is this other term, the two are relevant to each other. 4 sentences in, I've found another term, relevant to the previous 2... etc. etc."

It does not know, "oh, wow, what a great article about fruit bats. I love how it's all laid out and their grammar is perfect! This definitely deserves top spot."

 

[meatro]

Sorry for raining on your parade, mate.

Just some examples of what Natural Language Processing can do nowadays:

- Automatic summarization: Produce a readable summary of a chunk of text. Often used to provide summaries of text of a known type, such as articles in the financial section of a newspaper.

- Natural language generation: Convert information from computer databases into readable human language.

- Natural language understanding: Convert chunks of text into more formal representations such as first-order logic structures that are easier for computer programs to manipulate.

- Part-of-speech tagging: Given a sentence, determine the part of speech for each word.

- Question answering: Given a human-language question, determine its answer.

- Relationship extraction: Given a chunk of text, identify the relationships among named entities (e.g. who is the wife of whom).

- Topic segmentation and recognition: Given a chunk of text, separate it into segments each of which is devoted to a topic, and identify the topic of the segment.

- Text simplification: an operation used in natural language processing to modify, enhance, classify or otherwise process an existing corpus of human-readable text in such a way that the grammar and structure of the prose is greatly simplified, while the underlying meaning and information remains the same.

For example, try the MS Word thesaurus. Among all the possible synonyms it will present you mostly the relevant ones. Because it is able to determine correctly the semantic meaning of the word.

Analyzing text, especially English, is not that hard. And finding the common between two spun sentences is even simpler.

I did not say that it couldn't do any of that, I would very much hope that even the most basic of technology could do most of that.

I said that they cannot read and understand what they are reading. Can they relate it to a specific topic? I sure hope so. Can they tell me who the content is referring to? Again, I sure hope so. Or else we're all wasting a lot of time writing content when we could just be doing this:

"the the the the the the the the the the ANCHOR TEXT the the the the the the the the the the ANCHOR TEXT the the the the."

But if you're trying to say that a MACHINE can read a piece of text and say, "wow, that was compelling! This is a much higher quality piece of literature than this other block of text and rightfully deserves to be above it."

No it cannot. It can break it down, find the topic, simplify it, substitute terms, but none of that is based on a method of QUALITY. Because there's not a set algorithm for quality.

How would you tell a machine to conclude that... "Tom Sawyer is a good piece of literature, however this shitty Michael Chrichton book is not."?

You can't, because the machine can't read AND COMPREHEND. What's the machine going to say?

"This book is about Tom Sawyer of St. Petersburg, Missouri. Becky Thatcher is mentioned in the book 72 times, most often in the vicinity of Tom's name. He most likely has some type of relationship with this character. Grammar appears correct as well as spelling."

It could not tell you the quality of the literature based off of anything other than grammar, topic and wording. Anybody who writes more than Post-It notes will tell you that there's much more than that which goes into a quality piece of text.

 

[lancis]

How would you tell a machine to conclude that... "Tom Sawyer is a good piece of literature, however this shitty Michael Chrichton book is not."?

You can't, because the machine can't read AND COMPREHEND. What's the machine going to say?

"This book is about Tom Sawyer of St. Petersburg, Missouri. Becky Thatcher is mentioned in the book 72 times, most often in the vicinity of Tom's name. He most likely has some type of relationship with this character. Grammar appears correct as well as spelling."

Actually for machine it would look like: "Tom Sayer book is better than Michael Crichton book". Mathematically its a binary tree with 3 nodes: object (Tom Sayer's book), operation (better) and object (some Michael Crichton book).

Word it anyway you like, at the end the machine will build a binary tree with the same structure.

 

[wingman80]

I knew it! this was the main reason.. thanks for this post!

 

[meatro]

Actually for machine it would look like: "Tom Sayer book is better than Michael Crichton book". Mathematically its a binary tree with 3 nodes: object (Tom Sayer's book), operation (better) and object (some Michael Crichton book).

Word it anyway you like, at the end the machine will build a binary tree with the same structure.

That's exactly what I'm saying, IT couldn't conclude anything based on the input. IT could only come to a specific conclusion based on preset parameters ("this makes a block of text better than another.")

Now, how do you suppose they plan on breaking down centuries of literature and the English language into a set of parameters? WHILE, trying to avoid a specific set of parameters, as that opens the doors for cats like us. (edit: and remembering not to blacklist the 98% of English speaking humans who have terrible grammar, BTW. Based on this logic, those folks would not have anything worthy to say from a machine's standpoint.)

A machine can't do that. Sure, it can return all types of nifty data relating to the content, but it can't analyze two GOOD pieces of content and come to its own conclusion of WHY one is better than the other. Unless, like I said, it was relying on a specific set of parameters to draw that conclusion; and in which case it would be completely incorrect a huge percentage of the time.

Analyzing content for quality is not a logical action, analyzing data maybe, but not the written human word. It isn't a chess move, it's not finding the answer to a question and it's not remembering where the content is stored. That is why we rely on things like back links and social signals to tell Google that our content is high quality.

Whether that content originated from another source in a similar form is a completely different topic altogether, that is back to analyzing data, not QUALITY. That type of stuff they leave up to the humans to do.

 

[lancis]

That's exactly what I'm saying, IT couldn't conclude anything based on the input. IT could only come to a specific conclusion based on preset parameters ("this makes a block of text better than another.")

Its a bit deeper than that, good machine will have a database behind it and will actually retrieve the Tom Sayer book. I.e. if additional tasks have to be done, such as understanding what type of book this is, it becomes trivial in this case.

Now, how do you suppose they plan on breaking down centuries of literature and the English language into a set of parameters? WHILE, trying to avoid a specific set of parameters, as that opens the doors for cats like us. (edit: and remembering not to blacklist the 98% of English speaking humans who have terrible grammar, BTW. Based on this logic, those folks would not have anything worthy to say from a machine's standpoint.)

The machine is designed to do specific tasks. One task is checking grammar, another task is understanding content. These are two different actions that don't have to intersect. I.e. machines today are able to tell you that this text
is about books and the sentence has a terrible grammar.

A machine can't do that. Sure, it can return all types of nifty data relating to the content, but it can't analyze two GOOD pieces of content and come to its own conclusion of WHY one is better than the other. Unless, like I said, it was relying on a specific set of parameters to draw that conclusion; and in which case it would be completely incorrect a huge percentage of the time.

This is Artificial Intelligence you're referring to, mate. May be when quantum computers hit the market, we'll see something alike. Although I find it hard to believe.

But today's machine doesn't have to understand what you're speaking about, it just have to understand that two texts are the same. Its really a basic task in NLP, and like any basic task it has a good theoretical and algorithmic background.

 

[DutchTrafficService]

Sorry for raining on your parade, mate.

Just some examples of what Natural Language Processing can do nowadays:

- Automatic summarization: Produce a readable summary of a chunk of text. Often used to provide summaries of text of a known type, such as articles in the financial section of a newspaper.

- Natural language generation: Convert information from computer databases into readable human language.

- Natural language understanding: Convert chunks of text into more formal representations such as first-order logic structures that are easier for computer programs to manipulate.

- Part-of-speech tagging: Given a sentence, determine the part of speech for each word.

- Question answering: Given a human-language question, determine its answer.

- Relationship extraction: Given a chunk of text, identify the relationships among named entities (e.g. who is the wife of whom).

- Topic segmentation and recognition: Given a chunk of text, separate it into segments each of which is devoted to a topic, and identify the topic of the segment.

- Text simplification: an operation used in natural language processing to modify, enhance, classify or otherwise process an existing corpus of human-readable text in such a way that the grammar and structure of the prose is greatly simplified, while the underlying meaning and information remains the same.

For example, try the MS Word thesaurus. Among all the possible synonyms it will present you mostly the relevant ones. Because it is able to determine correctly the semantic meaning of the word.

Analyzing text, especially English, is not that hard. And finding the common between two spun sentences is even simpler.

Cool story bro, but i'm ranking thousands and thousands of pages with content you couldn't even read if you tried REALLY hard....that's how 'smart' google is.

As i said before, google's algorithm is a totall lame ass joke, and will remain so untill they get acces to cpus from outerspace.

 

[lancis]

Cool story bro, but i'm ranking thousands and thousands of pages with content you couldn't even read if you tried REALLY hard....that's how 'smart' google is.

As i said before, google's algorithm is a totall lame ass joke, and will remain so untill they get acces to cpus from outerspace.

I wasn't speaking about Google. I have no idea how 'smart' their algorithms are. All I'm saying is that if they wanted really bad to do something smarter than keyword analysis - the technology is there.

And if we are back to Google, their problem is not content analysis. Their problem is backlink based algorithms. They can solve some of the problems by better content analysis, but its like treating the symptoms instead the illness.

 

[meatro]

Well then I apologize. I never left the topic of Google, everything I've said so far has been regarding their abilities. But still.. As I said, you couldn't hand Google two books and tell it to analyze them and form an 'opinion' about which is the better. Unless it was based entirely on a database (parameters as I referred to it).. And they're simply not able to break down the history of the English literature, language and vernacular into a set of parameters.

Just over the last 100-200 years the English language has changed dramatically. It still continues to do so as the younger generations always tend to be coining new terms.

I also was not saying that they cannot detect spun text, I'm sure they can. The simple solution to that is write/spin better. Or just write unique text, that works well, too.

 

[kideze]

this update canned my auto-blog completely

 

[zmoney]

Interesting thing here is this domain I have, only has 10 links too it, it has 40 pages all Unique content image rich, White hat stuff and it got hit.

Remember, 40 pages, and 10 links mostly just to the homepage. Why would all of my page views drop? Can't be from link evaluation, because I don't have any links to my articles...

???????

 

[thedon23]

Interesting thing here is this domain I have, only has 10 links too it, it has 40 pages all Unique content image rich, White hat stuff and it got hit.

Remember, 40 pages, and 10 links mostly just to the homepage. Why would all of my page views drop? Can't be from link evaluation, because I don't have any links to my articles...

???????

Well, since you say it only has 10 links to it, I figure it's a fairly new site? If so, it may just be the G dance.

 

[zmoney]

1.5+ years I just never got around to promoting it lol..

MAYBE the link evaluation thing on the sites above me was hurting them and turning it off boosted em... lol who knows...... Just thought I would share some info.. maybe help to put some pieces together.

 

[madsem]

1.5+ years I just never got around to promoting it lol..

MAYBE the link evaluation thing on the sites above me was hurting them and turning it off boosted em... lol who knows...... Just thought I would share some info.. maybe help to put some pieces together.

Maybe your site ranked because of those 10 links and now G decided to not factor anchor text so much anymore instead WHERE the link is coming from.

I'm not sure yet but from what I can see in the serps I've checked is that many sites rank now that have links from thematically relevant domains. meaning where the actual domain is about something related to the page on your domain you are linking to.

This is also how i read the text of the part where google writes how they changed how link topics are being evaluated.