Article Surfing Archive

The hippie bible "Steal this book" was an anti-capitalism manifesto, widely remembered because of its creative and aggressive title plus its subvertive message.

Paper books like that one were not easy to steal, because of the guards in the bookstores and the costly effort of retyping, mastering, printing, binding and distributing. In the digital era you can get an ebook, crack or OCR it if protected, and copy-pasting it.

The main client for ebook-stealing tools are webmasters, in particular those with an interest for decent search engine ranking. In order to please the search engines and acquire deep indexing, high relevance and top ranking, a lot of relevant content must be written (or copied).

When I was into science there was a phrase "Publish or perish", that expressed more or less the same. Quantity instead of quality. I guess most of us would like to have time to write short and juicy articles instead of Goog-friendly mumbo-jumbo.

Machine-writing is a relatively new activity. And I am not talking of simple copy-pasting, but a more sophisticated text-generation breed of specialized software. Some simple programs take a number of words and just mix them, like Ktumbler. Others can generate random English-like phrases, like the "Web Economy Bullshit Generator", or sensible good English phrases by the thousand, like Phrase Generator in Synonymizer.

But phrases are not substitute for longer text, and many lazy or greedy webmasters are tempted to copy-paste from the web, which is easier and faster than thinking. A much criticized brand of website ranking systems, Cloaking, copies large amounts of relevant, well ranked text from the web, and shows it to the search engine spiders. In this way, they pretend to have more and better content than any other site. And the best of it, when the visitor is a human, they hide the stolen goods and show some innocent words. Cloaking refers to that ability, and it is achieved by IP detection and comparison with known spider data.

Finally, the synonymizing and text-mixing tools modify the original texts beyond recognition. Like facial surgery after committing a crime.

However, I am afraid that there is a little copyright issue... The problem is that nobody knows exactly what are the limits for automated text surgery.

To synonymize or otherwise disguise a source text is morally wrong, most of us would agree. But it has limits. If I say "My Kingdom for an equine", or "After me, the big rain", or "Let's there be illumination", you will understand that synonyms cannot be forbiden. There are also many situations in which overprotective legislation blocks creativity and innovation.

I am tempted to emulate Abbie Hoffman and discuss the morality of copyright and the whole issue of intellectual property. However, those are deep waters.

My point is that there is no established parameter to define plagiarism in texts. What if I substitute "2" for "two" and Goog for Google, as I do to avoid being noticed by them? And Yah for Yahoo, not to forget the pioneer?. Or if I change just one word in a phrase? What if I just mix the phrases in a text? Goog will probably still consider that the keyword density is correct for top ranking, minus a correction to account for the fact than the start of a file is more important than the end.

Many Plagiarism Detection services can compare any submitted text with a large library (mostly, the WWW) and decide if there is enough similarity with a certain source. They usually do not disclose their algorithm, but assure *it comprises proprietary technology* and *detects digital signature of the authors*. A leader in this field says "Copyscape looks for pages containing sizeable chunks of identical text". Nobody knows how that translates into numbers.

Goog and the other search engines are against "duplicated content", but they do not define it.

The availability of "text de-authoring tools" makes the intellectual property issue very blurry to any attorney willing to evaluate the existence of a crime. And as a collateral effect, the modified text will not be detected by the anti-plagiarism tools, which mostly search for exact text coincidences.

I started to examine some Plagiarism Detection sites, and I noticed that the better ones require some kind of a fee. Of course, it is not easy to compare a student term paper with the whole Library of Congress and the whole Web, plus the old Web Archives in the Wayback Machine. Others like Ferret are free and let you compare your file with another, but you need to provide BOTH files.

So, without hiding my condition of SEO Tool maker, I declare the need for a public algorithm that will establish if a text is the un-ethical or immoral or illegal derivative of a web source. It is necessary either for nailing plagiarists or to help writers and webmasters to define the limits for near-plagiarism and near-plagiarism devices.