Sign In

Communications of the ACM

Last byte

Fatal Guidance


View as: Print Mobile App ACM Digital Library Full Text (PDF) In the Digital Edition Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
Fatal Guidance, illustration

Credit: Shutterstock.com

You will never be able to find me, so don't even try. I do not think I murdered Mycroft Christie, and you will never prove I did. Yes, I was responsible for creating NOVELS, the New Online Virtually Excellent Literature System, and, to profit from it, I published personalized mysteries. But I never anticipated how it would be misused, with the loss of thousands of lives, and accept no responsibility for that unintended consequence.

My fundamental idea was not entirely new, but how I developed it was revolutionary. Interactive novels had existed since at least the 1970s, originally published as paper books, then online as trees of webpages. At the end of each scene, the reader would have a choice. Standing before the dead body of, say, a rich baron, do you accuse his butler or his wife? If the butler, does he have an alibi or not? If the wife, what was her motive? Some of the most recent role-playing computer games used the same crude method to add complexity to the stories, with the added feature that the player could select a name for the main character. My innovation was taking this method to its logical extreme.

My first experimental novel took a year to program but was a spectacular success. Based on The House of the Seven Gables by Nathaniel Hawthorne, it offered the reader an electronic book with a beautiful tapestry of personalizations. To begin with, the software would replace the last name Pyncheon, which appears 397 times in the original 1851 novel, with the reader's own family name. The reader would then select first names, replacing Hepzibah, Phoebe, Alice, Jaffrey, and Colonel Pyncheon. "New England" appears 20 times in the book and could become "New York," "the Carolinas," "Scotland," or whatever region the reader called home. "Massachusetts" appears only twice, but still the reader was given the opportunity to enter the name of a different district, as in, say, "Connecticut" or some county in England. Salem, the town in Massachusetts where the actual House of the Seven Gables still stands, is not mentioned at all. Indeed, the fact that the novel uses the word "town" 46 times without ever naming it gave me the Principle of Optional Location, applying to all the later projects. Other words could also be replaced, to bring the story into the world inhabited by the reader. For example, "As the Italian shouldered his hurdy-gurdy" could become "As the Texan shouldered his guitar."

The Principle of Optional Location was a starting point for my biggest innovation, a big data database with information stolen from Wikipedia, online phonebooks, and government sources that would automatically insert local landmarks into the novel, requiring the reader to merely specify the town. This worked especially well for murder mysteries, which tend to have formula plots and primarily vary their settings. Given the address of the place where the fictional murder was committed, NOVELS would insert the name of the nearest real restaurant where the detective could interview a possible witness to the crime, and perhaps the street address of the jail. As money rolled in, I hired hundreds of workers, added print on demand to produce paper copies, and published more than 1,000 novels, all adaptations of best sellers that had gone out of copyright.

In its most advanced version, the system could construct entire paragraphs describing a location. For example, if the final battle between the detective and the murderer took place in a public park, the system would steal text from the website of the one nearest the town the reader had selected. If the town was, say, Redding, Connecticut, that would be Putnam Memorial State Park. If it was, say, Redding, California, then Turtle Bay Exploration Park would be the denouement site. There were limitations, of course. The artificial intelligence was not advanced enough to raise the question of whether the Revolutionary War general after whom Putnam Park was named really was himself a murderer who executed young men on charges of spying on his army or deserting from it without holding proper trials. NOVELS did not make the reader's computer perform an ethical analysis of a story but merely personalized it superficially to entertain the reader.

Five years and one billion dollars into this spectacularly successful solo-owned business, Mycroft Christie contacted me via email with ominous news I at first refused to believe. He was only an underling at the Federal Bureau of Investigation's statistics bureau, responsible for compiling the Uniform Crime Reports the FBI publishes annually. Apparently, like me, he was an ambitious nerd who had stumbled across an especially promising idea. For a couple of years, the Report had shown a surprising rise in the homicide rate in small cities and towns, even as big cities like Chicago or Philadelphia showed no increase at all. He contacted several of the local police departments that had reported their first murder in a decade, collecting various bits of information. Most of the murders remained unsolved, but in two cases the murderer had been caught at the scene of the crime, with one of my novels in his possession.

Christie accused me of creating an information system that helped real murderers plan their crimes. The premise of every mystery novel is that only a singularly clever detective is smart enough to solve the crime. Why did the British police repeatedly ask Sherlock Holmes for help solving murders? Well, yes, the police were stupid, but also the mysteries were challenging, and only a genius could solve them. In the absence of the fictional detective, one of my murder mysteries could be the script for a perfect crime in the painfully real world. In small communities, many people were angry enough at their neighbors or relatives to kill them but feared their guilt would quickly be discovered. So they entered their data into one of my interactive novels and used it as the instruction manual for their crime.

Feeling angry myself, and not thinking through the possible consequences, I entered "Mycroft Christie" into the memory register of NOVELS that held the default victim name. Most readers would enter the name of a personal enemy for the victim in the murder mystery, but, if they failed to do so, they would read about the circumstances of Mycroft's death, along with the fictional evidence on his virtual corpse. I was astonished a few days later when the Facebook newsfeed reported the real-death murder of Mycroft Christie, considered newsworthy only because he had become one of the crime statistics it was his job to assemble. I clearly needed to disappear before his FBI colleagues discovered our email exchange and accused me of the crime, so I used my wealth to create a new identity.

My limousine is waiting outside, to take me to the airport, as I contemplate the main menu of NOVELS, trying to decide which option should end this story. Do I select Enter to add this testimony to the dataset? Do I select Exit to leave everything as it is, for my employees to inherit and use as they judge best? Or do I select Erase to destroy the entire system and protect the public from any further novel murders? Ah, yes. Perhaps I do know how this story ends. The CIA may pay me handsomely to use the system to write spy novels. If not the CIA, then ... you?

Back to Top

Author

William Sims Bainbridge (wsbainbridge.com) is a sociologist who taught classes on crime and deviant behavior at respectable universities before morphing into a computer scientist, editing an encyclopedia of human-computer interaction, writing many books on things computational, from neural nets to virtual worlds to personality capture, then repenting and writing harmless fiction.


©2017 ACM  0001-0782/17/02

Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and full citation on the first page. Copyright for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, to republish, to post on servers, or to redistribute to lists, requires prior specific permission and/or fee. Request permission to publish from permissions@acm.org or fax (212) 869-0481.

The Digital Library is published by the Association for Computing Machinery. Copyright © 2017 ACM, Inc.


 

No entries found