Autograph writer

The work of a mediocre writer can, and therefore, must be automated.

For people who are far from setural , it may not be obvious how many works of art have already been written, or will be written in the near future. Sites like Samizdat or Proza.rucontain volumes of text that programmers have not dreamed of even the most complex and heavily loaded systems. Even the seediest literary contest in the network gathers from tens to hundreds of participants, and it should be noted that the uniqueness of the texts is a condition for participation in most of them. That is, the production of the new is spurred on by life, but the quality of this new one is comparatively low and not in demand: barely one hundredth of those who manage to publish a book for money, win the competition in the same way.

This is due to limited resources (purchasing power and demand among readers, printing and production capacities of publishing houses, prize pool, time of the jury members), and, to a large extent, the quality of the works. If we discard frankly illiterate and inadequate texts — and, according to conservative estimates, no less than a quarter of any flow to publication or victory — there will still be too many standard, hackneyed, boring, secondary, uninteresting texts in our time that imitate the achievements of writers of the past and each other. Increasingly, assumptions are heard that even a robot could write this - competent, but absolutely inexpressive, differing from similar ones only in their own names, a work. So why doesn't the robot really do this?


The theory of literature has accumulated enough information about the structure of a literary work so that it is possible to compose an algorithm for generating text from large and medium blocks: parts and chapters.

The simplest is name generation . Leonid Kaganov, a well-known writer who is well acquainted with programming, at one time wrote an article " How to name your book ", containing an exhaustive list of possible schemes. In the simplest version, “Noun in the nominative case + noun in the genitive case” (“Dragon Sword”, “Game of Thrones”, “King of the Ring”) is realized in half an hour. For example, on the Kohana framework with its QueryBuilder:

	public function random_textname ()
	{
		// Get random text name part 1
		$offset = DB::select(DB::expr('MIN( `id` ) AS min, MAX( `id` ) AS max'))->from('story_names')->where('part', '=', 1)->execute()->as_array();
		$RID = mt_rand($offset[0]['min'],$offset[0]['max']);
		$random_name = DB::select('story')->from('story_names')->where('id', '>=', $RID)->and_where('part', '=', 1)->limit(1)->execute()->get('story');			
		// Get random text name part 2
		$offset = DB::select(DB::expr('MIN( `id` ) AS min, MAX( `id` ) AS max'))->from('story_names')->where('part', '=', 2)->execute()->as_array();
		$RID = mt_rand($offset[0]['min'],$offset[0]['max']);
		$result = DB::select('story')->from('story_names')->where('id', '>=', $RID)->and_where('part', '=', 2)->limit(1)->execute()->get('story');		
		$random_name .= ' '.$result;
		return $random_name;
	}

It is enough to have a base of the first and second parts with such a structure:

id | part | story

Next, you should think about the plot . There is a well-known statement by Borges that there are only 4 subjects in the literature :

• On the assault and defense of a fortified city (Troy)
• About the long return (Odyssey)
• About Search (Jason)
• On the suicide of God (Odin, Atis)

But for the modern reader this causes some bewilderment. Much closer to life is the theory of “Thirty-Six Dramatic Situations” by Georges Polty, with the help of which, with some assumptions, any literary work can be classified.
That is, having created only 36 plot construction schemes, we can already - in draft form - compose absolutely any work in the world that is not inferior to either proven classics or modern creations.

But what does the text consist of, regardless of the plot? Many even remember from the school course of literature that in every real work one can distinguish exposure, complication, development of action, climax, denouement and postpositiontherefore, the work must have at least 5 parts, each of which will represent one of the classic stages in the development of the plot.

Each part, in turn, can consist of some semantic blocks: narratives, descriptions and reasoning :

STORY TEXT DIAGRAM

1. The beginning of the event.
2. Its development.
3. The end of the event.

Note. In literary texts, there is also a climax preceding the end of the event.

TEXT-DESCRIPTION SCHEME

1. General impression of the subject (phenomenon).
2. Signs of the subject (phenomenon).
3. Attitude to the subject (phenomenon).

TEXT-DISCUSSION SCHEME

1. Introduction.
2. Thesis.
3. Evidence (explanation).
4. Conclusion.

Note. In a reasoning text, sometimes the introduction and thesis may coincide. In addition, in a text of this type, in some cases, the conclusion may be missing.

If in each of the 36 stories we write an algorithm for sequentially constructing narratives, descriptions, and reasoning, then we get. almost the entire "skeleton" and the main "muscles" of the work.

In turn, within a block there may be several paragraphs with a certain order of sentences within them, which - sentences - in turn consist of words in a certain order and forms. Although it is believed that the grammar of the Russian language is less enslaved than, for example, English, the standard literary style implies the direct construction of sentences - almost like in English - and any change of place of the “terms” of a sentence already implies an emotional coloring or individual characteristics of speech that we are now not interested.

And finally, we got the full text, which has a randomly generated name, compiled according to the scheme of one randomly selected from 36 subjects, with one of the random schemes, the arrangement of paragraphs and sentences. Plain mediocre text. It remains to add to it what distinguishes the work of the present author from the robot. Proper names.

Parsing lines is not such a difficult task. It is enough to place a marker like "# Main_Char #", "# Main_Enemy #", etc. in the original text instead of specific names and titles. Complicating the situation a little is the need for the Russian language to change the endings of words in different cases, but, I believe, adding another flag that clearly indicates the use of a specific case ending will not be difficult.

Thus, several months of work of a competent programmer will be able to replace hundreds of years and thousands of units of human labor. I dare to notice absolutely useless and sad.

The platform for the auto-writer, obviously, should be the web:

1. It has a low entry threshold for both the user and the developer
2. Good potential for modifications
3. Highest Availability in the Modern World
4. Relatively high reliability by storing information in the clouds

But one question remains: why is it necessary? Well, I believe that if mankind is saved from the need to write a lot and badly to shade off rare talents, there will be more of them.