How long does it take you to hand-spin an article?

[gmarkusson]

It takes me (or my writers) 1 week to spin a 6 paragraph article. Each paragraph gets rewritten 2 more time, each sentence 2 more times, then word/phrase synonyms.

I have longer article seeds that are about 80,000 words long and spun on 8 levels deep. I also have a madlib system with spun text that is about 400,000 words and allows me to produce massive amounts of content in my niche whenever i need them. With all the seed text I have I can generate tens of thousands of articles to post wherever I want. And they are about 99% readable, make sense and are actually useful (as opposed to superficial posts).

wow... 1 week to spin a article, that's some advanced spinning you are doing man. Kudos to you!

 

[Expertpeon]

Some of you really amuse me with your X% numbers. First of all the uniqueness % provided by any spinning software is INCORRECT/BAD. The article you get is not 100/50/15/1000% unique it is the seed that is % unique compared to the original. That means if you have 100 words and for each you add a synonym you get something like 100% unique. BUT if you spin that seed 10000 your average uniqueness will be very very low because you will get lots of articles that are not that unique compared to some other article that was generated. In order to correctly compute uniqueness you have to actually spin the articles X times and then compare each spin/article with every other generated. While some will be 100% unique, most will be much lower. The more spins you generate the lower the uniqueness for the whole batch is.

I've done a copy by copy comparison of uniqueness line by line (once... cause i was curious about this exact thing)... 40% uniqueness on my spun articles is reached after around article 800
This is with 150% in AMR articles.

I then take those articles (once I amass around 20) and dig out one paragraph from one and another from a separate article etc. and then combine them into new tier 2 garbage level articles.
In fact this works so well, that I find it difficult to imagine a more effective cost per time vs readability.
Per article I usually can get 8 or 9 spun versions approved at ezine... that's an impressive feat and something I've yet to see someone else accomplish with spun content.

 

[cybersage]

Wow wish I could find someone doing a service like what Madoct does. The value of a nicely spun article can allow one to generate a massive number of backlinks.

My articles tend to be in the law niche, which may add to the time it takes, but generally I'll spend 8 hours to spin one article. I usually do sentence variations and word variations within that. If I had better seed articles I would probably whore my articles out even further than I currently do. My problem is that I'm picky with regard to how the article should read, and especially more so because lawyers are supposed to have good grammar. I have been looking for a spinning service that meets the quality I'm looking for, but my guess is that those people are few and far between.

 

[madoctopus]

I've done a copy by copy comparison of uniqueness line by line (once... cause i was curious about this exact thing)... 40% uniqueness on my spun articles is reached after around article 800
This is with 150% in AMR articles.

I then take those articles (once I amass around 20) and dig out one paragraph from one and another from a separate article etc. and then combine them into new tier 2 garbage level articles.
In fact this works so well, that I find it difficult to imagine a more effective cost per time vs readability.
Per article I usually can get 8 or 9 spun versions approved at ezine... that's an impressive feat and something I've yet to see someone else accomplish with spun content.

What algorithm have you used for comparison?
Have you done 800*800=640000 comparisons?

 

[Orbify]

I don't handspin often, but when I do, I make it 100% unique.

 

[mdsurf]

just read this thread and there is some great info in here! looks like i need to step my game up with it comes to spinning. No more "Auto Select Synonyms" for me haha!

 

[Frostedbutts]

It takes me $1.50

 

[Expertpeon]

What algorithm have you used for comparison?
Have you done 800*800=640000 comparisons?

I've used Senuke's tool back in the day when you could define the # to compare.
The only other way is through copyscape premium and batch compare.

Despite all of this, if google indexes it I'm happy.

 

[Longover]

1. Write a 400-500 word article
2. Make 2 more variations of each paragraph
3. Manually select synonyms and makes sure that it is readable
4. Test multiple spins... takes awhile to notice mistakes and strange sentence structure.

Total results is...
spun article with 3 variations for each paragraph and about 2-3 variations for most of the words in the article.

Works really well and Google loves it!

est. about 2 to 3 hours

 

[Expertpeon]

1. Write a 400-500 word article
2. Make 2 more variations of each paragraph
3. Manually select synonyms and makes sure that it is readable
4. Test multiple spins... takes awhile to notice mistakes and strange sentence structure.

Total results is...
spun article with 3 variations for each paragraph and about 2-3 variations for most of the words in the article.

Works really well and Google loves it!

est. about 2 to 3 hours

sentence by sentence is better than paragraph spin by far... and it's roughly equivalent to the work you did rewriting paragraphs anyway.

 

[jb2008]

Some of you really amuse me with your X% numbers. First of all the uniqueness % provided by any spinning software is INCORRECT/BAD. The article you get is not 100/50/15/1000% unique it is the seed that is % unique compared to the original. That means if you have 100 words and for each you add a synonym you get something like 100% unique. BUT if you spin that seed 10000 your average uniqueness will be very very low because you will get lots of articles that are not that unique compared to some other article that was generated. In order to correctly compute uniqueness you have to actually spin the articles X times and then compare each spin/article with every other generated. While some will be 100% unique, most will be much lower. The more spins you generate the lower the uniqueness for the whole batch is.

Most people overspin. They add just one-level spinning for words and then generate hundreds of spins. That produces low uniqueness. That's why I spend a whole week to spin one article, so I can spin it 1000-2000 times and still get high uniqueness. And that's why I spin it on multiple levels. The more you spin it the more unique spins you can get out of it, but it is not linear. The more you spin the higher the return for your effort is.

Then there's the problem of redundant spinning. Lots of people that offer spinning services (actually I think all of them) produce redundant spinning. That is, they spin it on multiple levels but the result (spun article) ends up similar to original or other spun article because it had lots of redundant spinning. What is redundant spinning? Let's say you have a sentence:

I was going to the market to buy some fruits.

At sentence level you add an equivalent for the above as:

I will go to the store because I want to buy a few fruits.

Now, when you do word-level spinning, you will use "will go" as a synonym for "was going", "store" for "market" etc. So the result is like you didn't even done the sentence level spinning. You might even change from active voice to passive voice or reverse the order of the sub-sentences (I bet I lost you here... google "english active passive voice" and also find out about how a sentence can be comprised from multiple shorter sentences, structurally - read on linguistics). The sentence ends up ONLY LOOKING different but at n-gram level (google n-gram too) they remain very similar. Text comparison algorithms (google, copyscape, etc) are based on n-gram comparison and n-gram statistical analysis. That's why even with a super spun article done at top quality by my spinners I still get 10-15% similarity though mathematically/statistically it should be zero for a number of spins below 1000.

OK, I'll stop here, I probably lost you completely.

No , you didn't lose me, that was very good, and spot on.

You are right, when I do multi level sentence level and word level spinning and I generate 1000 articles, they are about 85% unique compared to eachother.

I am curious to see, what spinners / spinner services do you use? Because I am SICK of spinning everything manually. My end articles all make sense but god damn, it takes me ages! I need to outsource but I don't trust anyone to do it properly, especially given the example above. When I rewrite my sentences (I do 3 versions of each sentence), I do them almost COMPLETELY different but with as similar meaning as possible. Manual Spinning services are lazy and do that crap that you rightly pointed out. I need a service that can do this shit PROPERLY.

 

[madoctopus]

I've used Senuke's tool back in the day when you could define the # to compare.
The only other way is through copyscape premium and batch compare.

Despite all of this, if google indexes it I'm happy.

that was my point. it is incorrect. it is the "easy" way out. to do 640000 comparisons on a dual core will take you days of processing. on average using a PHP script it takes me 1 second/comparison. In a compiled language could be a lot faster but sttill very slow. I think the way senuke compares is by comparing the original vs the spin which is incorrect. hence you think it is very unique when in fact it could be very likely 10% unique.

The uniqueness plotted as a graph is a Bell curve / normal distribution. However if you spin to much it gets skewed towards 0% uniqueness (but never reaches it, the peak will be at about 10-15%).

The following is a graph plotted using data from 1,000,000 comparisons of a 400 word article. That's 1000 articles generated, each compared with each-other. The seed was 320,000 words long spun at 6-8 levels deep. As you see from such a huge seed text I still get "only" 50% uniqueness on average.

First for comparison using an n-gram size of 3:

Note: The axes are incorrectly labeled. Spins should be uniqueness and uniqueness should be spins. Spins is the number of COMPARISONS that produced that uniqueness, not the number of articles.

Next is same thing but using comparison set for n-gram size of 4:

If I would generate much more spins for example, the graphs would not look like that anymore. The peaks around 50-60% would get skewed towards zero as I generate more and more spins.

Algorithm used is similar to the one used by CopyScape and the similarity to CopyScape is +-15%.

Things are not as easy as they seem and certainly uniqueness computation is not as easy as made by the spinning software. The mathematics behind are much more complex and to this day I was unable to find somebody to help me develop an algorithm that computes optimal number of spins for a given seed for a given desired uniqueness. Even less success with a spinning algorithm that doesn't rely on randomness but produces spins optimally for uniqueness.

 

[madoctopus]

I have 2 full time writers who spin using TheBestSpinner but BY HAND.

No , you didn't lose me, that was very good, and spot on.

You are right, when I do multi level sentence level and word level spinning and I generate 1000 articles, they are about 85% unique compared to eachother.

I am curious to see, what spinners / spinner services do you use? Because I am SICK of spinning everything manually. My end articles all make sense but god damn, it takes me ages! I need to outsource but I don't trust anyone to do it properly, especially given the example above. When I rewrite my sentences (I do 3 versions of each sentence), I do them almost COMPLETELY different but with as similar meaning as possible. Manual Spinning services are lazy and do that crap that you rightly pointed out. I need a service that can do this shit PROPERLY.

 

[jb2008]

How much do they cost? I know this is a long shot but you wouldn't happen want to share them privately of course? I would really appreciate it because I am dying here spinning by hand lol!

 

[madoctopus]

I pay them $600/mo full time.

 

[jb2008]

Thanks mate. How much would they charge to spin per article?

 

[madoctopus]

Thanks mate. How much would they charge to spin per article?

I just told you I pay them $600/mo. They are not a company offering a service. They are MY employees. I went and posted ads on the doors of journalism, English, literature colleges that's how I found them.

 

[GooglesMostWanted]

I just told you I pay them $600/mo. They are not a company offering a service. They are MY employees. I went and posted ads on the doors of journalism, English, literature colleges that's how I found them.

Hey madoctupus,

Thumbs up for the posts... got me actually thinking.

How are you able to tell if the article you wanted spun by your employee was manually spun or auto spun? I always wonder how I could make sure that I'm not being fooled...

The only way I know how is to read the spin variations after to see if they make some sort of sense.

 

[slim_dusty]

Hi madoctopus,
Good to see someone who understands spinning, and it sounds like you have found employees who really understand spinning. Proper spinning takes time. And spinning needs to be done right to make it effective. There is a lack of services for proper spinning. Have you considered providing a service using your employees? You'd be ideally placed to offer this service, and miffy help to recuperate some of your costs!

 

[inwebdev]

I hate hand spinning because I'm a perfectionist an I want the spins to be perfect. I can take hours to do a 500 word article. Seems like a complete waste of time to me.