- Introduction to Focus: 00.0 Machine Writing
Can truth be told, it’s clear my work’s not Zen? Levenshtein, your magic will clear the haze Iambic verse has rules and guides my pen It seems to me your spelling drives my daze.
02.0 It begins like that. No, actually, it begins more like this:
|Raw Score||Essay Classification System|
|Freq/1000||Odds for Group|
03.0 Or say rather, it begins like this:
Metrics for Evaluating Student Writing:
Does the composition make sense? Is the composition organized logically?
04.0 Married to this: /* “executes” the RE against the text taken from stdin */; and this: 01001001001 01001001001001010….
05.0 Read back up the above chain of texts: from the 1s & 0s of machine language and software commands married to metrics for judging writing translated into a calculus for determining literary merit, to the output of literature itself, a poem. What this sequence sketches is a cartoon of one way we got from there (a time when writing was written by humans) to here: a time when more writing is done by machines than people, a time when much writing is meant to be read by other machines, and is increasingly indistinguishable from (and often preferable to) human writing. Long before NSA revelations, machines have been reading the trillions of emails, text messages, and blog posts that pass through the Internet by the hour, combining our messages with information harvested from public records such as birth and death certificates, census data, publications of income by zip code, or property records, to automatically write detailed profiles of individual people: “actionable intelligence,” that is, reports that can set in motion actions as diverse as arresting us to alerting us, after a global search, to a vacation we might like to take. The recommendations that sites like Amazon.com write for us have become so ubiquitous, so natural, that we’ve stopped noticing that to write these recommendations, computers continually read our writing. Indeed, spell and grammar checkers and the like were only the first baby steps of machines that would eventually do the writing for us, as is happening when machines with the ability to process natural language take on writing assignments that once would have been given to a human, a sports or business reporter, for example:
While company shares have dropped 17.2% over the last three months to close at $13.72 on February 15, 2012, Barnes & Noble (BKS) is hoping it can break the slide with solid third quarter results when it releases its earnings on Tuesday, February 21, 2012….
Philip M. Parker alone has written over 200,000 books, ranging from poetry to food security, well, his machines have, or, as he explains, the process his software used to write, the poem this essay opens with (along with about 1.3 million other poems) is an algorithm that employs graph theory and a metric for linguistic differences across word strings in other poems.
06.0 It is not an exaggeration to say that machine writing has touched every person on the planet who can read. “Please listen carefully, as our menu options have changed” is our cue to align ourselves as closely as possible within an information system designed to maximize profits by making humans part of a programming loop. If you’re one of the 500 million average Twitter users, it’s estimated that only about 35% of your followers are actual people, the rest falling along a spectrum from spambots to Carina Santos, a journalist out of Brazil ranked more influential than Oprah Winfrey, and, who as it turns out, is actually a machine. Indeed, Indiana University’s School of Informatics, among others, has developed a program to help people (or, presumably, machines) determine whether or not an essay was written by a person. (Spoiler alert: it gave the probability of the essay you are now reading a 75% chance of having been written by a human.) EdX, the...