Putting the science in fiction - Dan Koboldt, Chuck Wendig 2018

Information theory: deep thoughts on building hal
Things to know for when Skynet takes over

By A.R. Lucas

Good science fiction starts with the question “what if?” It asks us to take what we know and stand at the crossroads of a myriad of possible futures. Technology is crucial to science fiction because it represents change. Sometimes it’s a simple plot device, a way to get from A to B. Often it’s a catalyst, something fundamental to the world we’re about to explore, or even a character in its own right. And at technology’s core is a fundamental but slippery concept: information.

Okay, don’t leave yet. I know, it’s not spaceships or killer mutant viruses, but it can be something far stranger. Don’t believe me?

Take a number called Champernowne’s constant. In Base 10, the number is a concatenated list of all numbers (12345678910111213…). It’s what’s called a normal number, which means its digits contain all numerical combinations equally, but that’s not the cool part. The cool part is that Champernowne’s constant, when translated into letters or bits or any other code, contains your next unwritten novel, the copy for this week’s grocery flier, and Jorge Luis Borges’s entire Library of Babel. You know, because of that whole all numerical combinations thing.¹

Information theory is less than a century old, but it has changed our understanding of everything from physics to biology to computer science. It deals with the representation, communication, processing, and utilization of information.

Drowning in data and creating context

Information grows exponentially. This means that the current volume is always greater than the sum of everything before it. But too much information can overwhelm us, like the town of Ersilia in Italo Calvino’s Invisible Cities (Harcourt Brace Jovanovich, 1974), which must be abandoned due to a lack of physical space.

The recent explosion of data has led many information experts to coin variations of the phrase “We are drowning in data, but starved for information.” Data can multiply like the Tribbles on Captain Kirk’s U.S.S. Enterprise, but without context, it’s meaningless. Or worse, we can ascribe false meaning to it (google “spurious correlations” for a laugh), or make false assumptions.

In the 1983 movie WarGames, the protagonist hacker, David Lightman, believes the computer system WOPR to be a simulation game instead of a military supercomputer. In fact, a military program and a computer game representing the same simulation may have the same programming, even though the consequences of engaging with one versus the other are very different.

The problem is Lightman interprets his interaction within the context he is expecting. He finds WOPR when asking for games and assumes that it is a game. False assumptions can enhance plot, and such problems may be unavoidable as information, like time, is relativistic.

Damn you, Heisenberg, the answer is 42

Heisenberg’s uncertainty principle states you can know either the exact speed or the exact position of an object, but not both. In fact, what you are trying to measure does not exist until you try to measure it. This means to get an answer, you must first ask the right question.

In Douglas Adams’s The Hitchhiker’s Guide to the Galaxy (Harmony Books, 1994), the supercomputer Deep Thought is built to solve the Answer to the Ultimate Question of Life, the Universe, and Everything. After 7.5 million years, the computer arrives at the answer: 42.

Deep Thought claims the answer seems meaningless because the programmers didn’t know what the question was, and without the right question, quantum mechanics dictates that the answer is not just unknown, it is unknowable.

Entropy as a measure of ignorance

Probabilities are essential to information theory, but they often play tricks on us. Take the following example: I have two children, born two years apart, one of whom is a boy. What is the probability the other is also a boy?

Did you answer one-half to the question above? Unfortunately, that’s incorrect. The answer is one-third. The possible combinations are BB, BG, GB (GG is not possible). Only one of these has a second B.²

Tricky, isn’t it? Well, what if I tell you the taller of my two children is a boy? Height seems irrelevant, but it’s not. It changes our probabilities. Now we only have two combinations: TB, SB and TB, SG. Probability is essential to information and information affects probability in surprising ways.

Entropy is a measure of ignorance in information theory. An entropy of zero would represent complete knowledge, and an entropy of one complete ignorance or true randomness. (This should not be confused with probability itself, where both zero and one are less entropic states, and true randomness, the coin flip or 50/50, represents the state of highest entropy.) In thermodynamics, entropy can never be decreased, only held constant or increased.

Which unfortunately means the search for Deep Thought’s right question may be impossible. If we knew everything, the information entropy would be zero and the universe itself might cease to exist. Yeah, it’s a bit of a buzzkill. Blame Heisenberg.

Transmitting a message

Current science dictates that information transmission speeds cannot exceed the speed of light. This means your deep space communications could take a long time to reach home unless your faster-than-light (FTL) space travel technology extends to information, too. The Shannon-Hartley theorem tells you how much signal power you’d need and what the maximum amount of information you could send per unit of time without it becoming error-riddled noise.

Recently, there’s been talk about using quantum entanglement to send FTL communications, but beware. The no-cloning theorem says you can’t do this, and that even if correlation exists between two states it remains unknown until the states are verified against each other, which requires communication at the speed of light. For now, intergalactic snail mail is here to stay.

A common language

A sender must represent information such that it conveys meaning to the receiver. Think of the prime numbers embedded in the alien message in the 1997 movie Contact. That string of numbers established the message as an intentional communication (not just noise) and mathematics as the common language. Communication imbues information with meaning and when technology is a character in its own right, developing systems of shared meaning can be a fundamental story challenge.

Returning to WarGames, WOPR is a doomsday machine. It relies on automatic fulfillment and cannot be unplugged or it will trigger nuclear war. This forces Dr. Stephen Falken and Lightman to find a basis for communication. The solution they arrive at is to have WOPR play tic-tac-toe against itself to teach it the concept of futility, as tic-tac-toe, when played without errors, always results in a draw (and in the case of nuclear war, the stalemate of mutually assured destruction). If you haven’t heard of game theory, it’s a useful tool to understanding how computers—and Vulcans—think.

Empirical to abstract: in our own image?

In the 1968 movie 2001: A Space Odyssey, the computer HAL 9000 tries to destroy the ship’s crew after overhearing that they plan to disconnect him following a series of malfunctions. Many viewers interpret the plot as a degradation of HAL’s programming into human-like madness, but that assumes HAL is anthropomorphic. Instead, perhaps he is the opposite of mad, and his behaviors are pure logical self-preservation.

When developing artificial intelligence (AI) and other technologies, it is important to consider how they gather, process, and interpret information. The same image can mean disparate things to different people, and they share the common experience of being human. Push the boundaries when trying to devise logical frameworks for information processing and for communication between frameworks, and understand the limitations of your systems.

Complexity

Information theory tells us that the more complex a system is, the more entropy it has. If more entropy equals greater randomness, then the more complex our computing systems become, the less we can predict them. When they interact with a complex world, it is impossible to program responses for every potential scenario (too much data) or develop blanket rules that address every possible negative outcome.

Isaac Asimov’s Three Laws of Robotics and his many writings about human/robot interactions and their unintended consequences demonstrate this. It is impossible to build complex systems of complete predictability. In recent literature, a hard takeoff (fast-learning) Artificial Superintelligence (ASI) and its potential unintended consequences are sometimes referred to as “the busy child.” The unintended consequences may be malevolent or benign, but they must be there.³ Entropy demands it.

The universe says, “Hold my beer”

In The Matrix, the majority of humans go about their daily lives in a mundane, massive computer simulation while machines harvest the energy from their neural activity, sparking many late-night debates as to whether we are already living in the Matrix.

Turns out this may be true, on a far grander, though less nefarious, scale. Physicists are exploring the possibility that the fundamental building block of the universe may be not the quark or the lepton but the qubit, the base unit of quantum information. Space, time, and the universe as we know it may all emerge from the interaction of these qubits.

And remember, even though your next novel is lurking somewhere in Champernowne’s constant, the pattern that makes it relevant doesn’t exist until you write it.

Conclusions

Information theory is about interaction, and our interaction with information is changing our world every day, from artificial intelligence to quantum physics. However, if information theory is correct, as the complexity of our universe expands, so does its uncertainty, creating plenty of twists and turns for science fiction writers. The question is: Will any of them be stranger than the truth?

1 von Baeyer, H. C. (2003). Information: The New Language of Science. Boston, MA, USA: Harvard University Press.

2 von Baeyer, H. C. (2003). Information: The New Language of Science. Boston, MA, USA: Harvard University Press.

3 Barrat, J. (2013). Our Final Invention. New York, NY, USA: Thomas Dunne Books