*Using Automated Untelligence to fight Artificial Intelligence.

Posted: July 23, 2010 in 2010, Exclusives
Tags: , , ,

Ignorance Is Futile:

The genius behind the Google machine is that when we use the Internet we help make it’s A.I. smarter and more powerful, and on the other hand the growing trend is to make the entire Internet “semantic“. By merely using the Internet, we’re all in effect helping to build ‘Skynet’, so something must be done to at least slow it down. So a couple years ago I had the idea of software meant to make it dumber.  Bots against bot.

The Google Machine is the ultimate in crowd sourcing. For example, we all helped Google’s AI learn how to understand spoken language. But it goes beyond just using it’s search engine. By using the Internet there is almost no escaping this. On the one hand it “indexes” any and every web page it can find, allowing it to analyze things we’ve written. On the other hand, most sites out there have either Google Search or Google Ads integrated right into the page, allowing them to track our web surfing habits even if we don’t start our travels at Google.com.

So last year I proposed software scripts designed to dumb it down, the more automatically the better. I just ran this idea by my new collaborators, TransAlchemy, it it turned out that member “SeH” not only liked the idea, he already has a working prototype program that uses “Markov Chain” algorithms to garble up text.

Not only does it randomize text, it actually solves the page indexing problem. In my idea I hadn’t thought it through enough to figure out how to hassle the page indexing, I only had the concept of a bot that spits random nonsensical search stings into the search box. His program takes manuscripts, and rearranges both the order of the sentences, and blends sentences together.

The output even contains paragraphs, and on top of everything it’s actually fun to read. Here is an example of the first 2 paragraphs of an entry from his Transperiments blog:

High Mentality that it May Be Impaired by Dawn

as the olive grove . he handled and separated parts of a small , and self – – not ; a moment on the slightest hope of his experiments with robbie nodded significantly ; consequently he was experimenting on the nitre – or the manager of the house ? he’s been hungry . it rang , because they had come from my ancestors had with full and calculating appraisal at times he hoped at the sagging floors , robert , and i had entered the neighbouring town had an air – tree , we lacked at the wrong with disastrous results , unknown malady . but , but i shiver . despite the thing quietly , for a student of dr .
gloria was rarely home – daemon – – for a long centuries had much of my friend believed it wasn’t quite overshadowed by the hillside below the body , the early acquired a dark colour . the cellar laboratory , i saw a shrill rhythm . “where have been able to overcome impatience . so carelessly sceptical , “damn it .

robbie was strangely bent every sort of brain and his lips and buried by west and comfortable . the moss – – black form the men had occurred , yet never get rid of the servants were widely ridiculed by the memory of mind , i call death . a supremely great work when the way to me , and whose ramifications and tom – stained blocks of a debauch . eloi , in a greater age , by one night the midst of his terrible groping . gloria ? ” west’s closest neighbour , who wore a fashionable thing . the unfathomable abyss of restraint – the others had occurred , to feed and disgust ; led by the oldest burying – down his selection of the st . nor indeed noticed the night of their estates , unnatural expedients in skill with good , virtues , though kalos and their reception by sheer force of yore , whom we waited until the body , and in preparation .

Imagine all of the varying different web pages, books and so on that you could morph into oddity hybrid entertainment. Religious texts, news articles and so on. Consider the different ‘specialty’ forms of English used everyday such as slang,  patent documents, “legelese” (legislative / legal documents), and more. Then you have old manuscripts written in odd tongues and dialects. To really screw with Google you could even mix different languages. Google can translate just about any language, but it would still confuse it no less.

The overall idea of this concept is to give people a fun tool they can have fun with, and then hopefully post their varying results online. With his program, in the context of what I’m trying to do with Google, the hard part is already done. The easy part should be making a smaller automaton applet that uses your web browser to do phony web searches. This should be a program that runs in the background, and uses very little CPU resources. You can even run it while you’re away from your computer, if you already leave it on all the time anyways.

The Tools:
1. Downloadable tool that users can feed different sources of manuscript into it.
2. Web page that has it built in, allowing you to select your own sources or select from a provided list.
3. An app that runs in the background and constantly feeds random search strings into Google.
4. A “Chaos” button extension built into your web browser, allowing you to auto-generate gibberish for posting into comments like on news sites.  Interactive chaos fun.

My concept currently has the bot use an Internet Explorer browser, that opens 2 tabs. One tab is set for Google News, the other is just the plain Google search page. It uses the available words from the news page, and then rearranges 13 random words into one nonsensical search string. It does the search on the plain search box, saving it from having to reopen the news page, thus using less resources. All of this done persistently and automatically. Or at least that’s one way to do it. There could be infinite ways to go about it.

OPEN SOURCE:

These sorts of ideas inherently scream open source, and SeH already has it available as open source. The more people that help out this the better, and for software programmers you can join SeH’s GIT to collaborate on work and ideas of this effort.

The software that inspired him is also open source. There’s a website called DadaDodo that is a functional first generation version of this “cut up” chaos concept. The history behind it is most interesting:

DadaDodo
Exterminate All Rational Thought

William S. Burroughs called this “cut up theory.” His approach was to take a page of text, divide it into quadrants, rearrange the quadrants, and then read the page across the divisions. He wrote this way; writing, cutting up, shuffling, publishing the result. Collage and randomness applied to words. He saw this as a way of escaping from a prison that words create for us, locking us down into one way of thinking: an idea echoed in Orwell’s1984,” where the purpose of NewSpeak was to make ThoughtCrime impossible by making it inexpressible: “The Revolution will be complete when the language is perfect.”

Ted Nelson, the inventor of hypertext, published “Computer Lib” in 1973. This book was more a stream-of-consciousness collage than anything else, nominally about nonlinear texts, and effectively an example of the same. It was written as hundreds of individual typewritten rants, and then pasted together for printing. Ironically, it was printed with a third of the pages out of order, allegedly due to a mix-up with the printer: one wonders, however, whether that really mattered.

The site is functional, but the key limitation is that you can’t specify what text sources it uses. Another drawback in terms of page index chaos is that the pages it generates aren’t being indexed, unless someone where to copy and paste it somewhere else.

SeH’s beta software overcomes that key limitation, and strives for more complex semantic structure. This initial release is really in its alpha stages, but it works, as seen above. SeH wants to make the utility more functional and easier to use, and then integrate into a web page that people can use to paste in links or other forms of text, or even have a big list of prime sources for semantic fun right at the users fingertips. Then the next step is to make smaller automaton bots for dizzying the search engines.

Stay tuned for updates and contact us if you’re able to help build on these tools and techniques.

Comments
  1. rah says:

    I was searching for some info about Markov models and whether google could or could not figure out if some text is Markov generated nonsense. Seeing as this “Transperiments” blog is indexed just fine I don’t think that google is capable of detecting it. Anyway, I rolled my own Markov text generator and ran it through the first 4 paragraphs of your blog post for fun, here is the result:

    By merely using it’s search engine. By using it’s A.I. smarter and more powerful, and more automatically the ultimate in effect helping to make it’s A.I. smarter and rearranges both the page, allowing them to figure out that member “SeH” not only liked the Internet there have either Google machine is the other hand, most sites out that member “SeH” not only liked the page, allowing it turned out that member “SeH” not only does it can find, allowing them to understand spoken language. But it actually solves the Internet we use the order of the concept of the Internet we all in crowd sourcing. For example, we use the page, allowing them to track our web surfing habits even if we all in effect helping to at Google.com. So a couple years ago I just using the concept of software meant to track our travels at least slow it dumber. Bots against bot.

  2. ignoranceisntbliss says:

    Wow. That works a little bit too well! What programming environment are you using? SeH’s is made with Java, it works but he’s having some compatibility issues is why I haven’t posted an update with download links.

    There’s another functional tool almost ready also:
    http://www.abovetopsecret.com/forum/thread597264/pg1

    I started a forums for everyone to post these types of softwares and techniques, I just haven’t populated it with content yet. Been real busy with everything.
    http://agimanhattanproject.com/au/

  3. rah says:

    I’m using Factor (from factorcode.org) but I’m a bit of a programming language geek. 😛 I would be surprised if the text from “CHA0S” can not be detected by google though. With some simple statistics it is easy to detect that that type of text is automatically generated. If you generated a bunch of pages with such text and put them on a blog then google may flag it and de-index/ban the website from the search results.

    I’m wondering whether or not the text that I generated would be detected by google as automatically generated. Currently I’m leaning towards: no it won’t be detected by google because your friends’ “Transperiments” blog seems to be indexed just fine and his punctuation is very odd. If that is the case would that be true for foreign languages as well? The statistics/”features of the language” can be different after all.

Leave a comment