Abstract Poetry fascinates me.
There is some beauty to be found in the subjectivity that is indigenous to every poem. It leaves itself open to a hundred different interpretations, and it makes you feel special for coming out with your own unique version.
On a few instances it may appear that the poet is just playing on you, with no meaning in itself, it leaves on the audience to generate sense out of pure gibberish.
Isn't it a possibility that a person out of sheer randomness, starts typing words mindlessly, and these words are interpreted by others as a work of pure genius, a work that would take hundreds of hours to come up with ?
But random strokes of music dont make a masterpiece.
Or do they?
In his book, "Fooled by randomness", Nissim Nicholas Taleb talks about "Monte Carlo Poetry", and elegantly summaries the beauty of chance and numbers by suggesting that "if one puts an infinite number of monkeys in front of a typewriter, there is certainty that one of them would come out with an exact version of the Iliad."
Repeat the procedure a hundred times and fit the words in a syntax.
What are the odds that the words you'd have written down by now would make any sense, leave alone being poetic?
Infinitesimal.
Well that's not discouraging at all.
The point to be noted is that it's not zero. It's infinitesimal.
Infinitesimal, but not zero.
So if you generate a million poems in such a way, you might expect to come up with at least one meaningful poem.
Meaningful, but not a masterpiece.
A million poems? We are humans, not monkeys on typewriters.
Let's consider the following rules:
We feed a software with a set of random verbs, nouns, adjectives and prepositions, identify a syntax, say Preposition-verb-article-adjective-noun. example : He-like-the-great-place. ( Caveat: the sentence may not be grammatically correct, but rules can be fed to eliminate the problem. )
The software randomly and independently selects a word from the set provided, and voila, we have a simulated sentence. To take things to another level, we can also randomly select the syntax of the sentence.
I took inspiration from the idea, and using a word data base of commonly used nouns, verbs and adjectives, I tried to simulate a cinquain using Microsoft Excel.
Now, a cinquain is as simple as a poem can be.
In consists of five lines ("cinq" uain).
The first line consists of one word, generally a noun. Lets call this word the subject.
The second consists of two adjectives describing the subject.
The third line contains three verbs related to the subject.
The fourth line is a four word general statement.
And the final line, again consists of one word, generally the subject itself.
This simple structure of a cinquain, makes it very easy to replicate the rules, and hence simulate the poem.
Here are some of the outcomes that came out within a few clicks of the mouse button :
Interesting, aren't they?
The same technique would have done equally well, if not better in generating ( simulating ) metaphors.
My friends weren't impressed though, but I guess they would be had i told them that these were a product of pure randomness and chance.
For the initiated:
Can more complex and real masterpieces be simulated?
Of course they can, but they would require much more efficient set of rules and the knowledge about functioning of the English language. In the cinquain, the nature and the number of the words to be used was a constant, which made the work much much easier. However, in the case of conventional poetry, this may not be the case and combinatorial rules may need to be framed for the occurrences of different words or alphabets. Further, several diagnostics algorithms can be structured to maximize the efficiency of the simulations.
More on this will be covered in a later post.
There is some beauty to be found in the subjectivity that is indigenous to every poem. It leaves itself open to a hundred different interpretations, and it makes you feel special for coming out with your own unique version.
On a few instances it may appear that the poet is just playing on you, with no meaning in itself, it leaves on the audience to generate sense out of pure gibberish.
Isn't it a possibility that a person out of sheer randomness, starts typing words mindlessly, and these words are interpreted by others as a work of pure genius, a work that would take hundreds of hours to come up with ?
But random strokes of music dont make a masterpiece.
Or do they?
In his book, "Fooled by randomness", Nissim Nicholas Taleb talks about "Monte Carlo Poetry", and elegantly summaries the beauty of chance and numbers by suggesting that "if one puts an infinite number of monkeys in front of a typewriter, there is certainty that one of them would come out with an exact version of the Iliad."
Well.
Consider it, close your eyes, randomly open a page in the Oxford English Dictionary, open your eyes, and write down the first word you see,Repeat the procedure a hundred times and fit the words in a syntax.
What are the odds that the words you'd have written down by now would make any sense, leave alone being poetic?
Infinitesimal.
Well that's not discouraging at all.
The point to be noted is that it's not zero. It's infinitesimal.
Infinitesimal, but not zero.
So if you generate a million poems in such a way, you might expect to come up with at least one meaningful poem.
Meaningful, but not a masterpiece.
A million poems? We are humans, not monkeys on typewriters.
Let's consider the following rules:
We feed a software with a set of random verbs, nouns, adjectives and prepositions, identify a syntax, say Preposition-verb-article-adjective-noun. example : He-like-the-great-place. ( Caveat: the sentence may not be grammatically correct, but rules can be fed to eliminate the problem. )
The software randomly and independently selects a word from the set provided, and voila, we have a simulated sentence. To take things to another level, we can also randomly select the syntax of the sentence.
I took inspiration from the idea, and using a word data base of commonly used nouns, verbs and adjectives, I tried to simulate a cinquain using Microsoft Excel.
Now, a cinquain is as simple as a poem can be.
In consists of five lines ("cinq" uain).
The first line consists of one word, generally a noun. Lets call this word the subject.
The second consists of two adjectives describing the subject.
The third line contains three verbs related to the subject.
The fourth line is a four word general statement.
And the final line, again consists of one word, generally the subject itself.
This simple structure of a cinquain, makes it very easy to replicate the rules, and hence simulate the poem.
Here are some of the outcomes that came out within a few clicks of the mouse button :
"Child
Cunning, Intelligent
Question, Elucidate, Alter
Bring back to life
Child!"
"Woman
Temperate, Talented
Form, Disguise, Contradict
Wings past the world
Woman!"
"Human
Foolish, Incorrigible
Recall, Renounce, Study
Loves its own kind
Human"
Interesting, aren't they?
The same technique would have done equally well, if not better in generating ( simulating ) metaphors.
My friends weren't impressed though, but I guess they would be had i told them that these were a product of pure randomness and chance.
For the initiated:
Can more complex and real masterpieces be simulated?
Of course they can, but they would require much more efficient set of rules and the knowledge about functioning of the English language. In the cinquain, the nature and the number of the words to be used was a constant, which made the work much much easier. However, in the case of conventional poetry, this may not be the case and combinatorial rules may need to be framed for the occurrences of different words or alphabets. Further, several diagnostics algorithms can be structured to maximize the efficiency of the simulations.
More on this will be covered in a later post.
