Mechanical Minds – Nuclear Fruit, Part One

Mechanical Minds – Nuclear Fruit, Part One


They say knowledge is power – and this is
seldom more true than in war. To know your enemy’s plans and deny them the
same privilege is to stack odds in your favour. Military cryptography has a history as long
as any signal corps: from the simple ciphers used by the ancients – to more complex electromechanical
devices of the 20th century. One such machine was the German Enigma. It resembles an overengineered typewriter,
but the encryption it provided for the Nazis proved tough to break. Plaintext is passed through a scrambled pathway:
a series of rotors set to a secret configuration each day. Orders, locations or mission details could
all be safely transmitted over radio, with no fear of interception. Without knowing the machine’s state, translating
a coded message back into a readable form would have been almost impossible. Almost. Its complexity was such that mere manpower
could not solve it within any reasonable time: to crack the code would require a mechanical
mind instead. The first successes in cracking Enigma belong
to the Polish Cypher Bureau – but much of the wartime glory is attributed to one man: The brilliant mind of Alan Turing. Known for his pioneering work in the emerging
field of computer science, he found himself at Britain’s codebreaking centre at Bletchley
Park. It was here that he and his team achieved
the impossible – a machine was made that could unravel the German ciphers, leaving their
private plans laid bare. No doubt, the effort spent breaking Enigma
saved lives and shortened the war. A secret success that shaped the Allied victory. The world would never be quite the same after
1945: nuclear punctuation marked the start of the modern age. The computer’s role in war had only just begun. Many of us have grown up with video games: but video games grew up during the cold war. This is a story about how the modern age took
shape – and how new technology and political tension gave rise to the games we play today. From mechanical minds, to the pursuit of other
worlds; huge nuclear arsenals and their alarming potential: Video games and war have more in common than
you think. The aftermath of World War II saw Europe in
ruins, and two new superpowers emerged in the stead of the old: the Soviet Union, and
the United States. Former allies left standing in a divided world:
rife with paranoia and espionage; An inevitable struggle for power had just
begun. Having proven their wartime worth, computers
found their way into an academic setting: governments were keen to invest in technology,
lest they be left behind. Turing’s work on Enigma was shrouded in secrecy,
and so he quietly resumed his role in computing research. With no codes left to break, he instead sought
an answer to a question that has dogged the mind of philosophers and engineers alike: Can machines think? Turing’s approach to artificial intelligence
was a pragmatic one: a convincing opponent need not think like a human, only behave like
one. He devised a method to evaluate AI ability:
a blind test in which a neutral party poses questions to both a computer and human participant,
in an attempt to discern which is which. He called it an Imitation Game: but we know
it better as the Turing test. The game a veil between machine and mind. The Mechanical Turk was a marvel of a machine
built by Hungarian engineer Baron Wolfgang von Kempelen. A formidable opponent, this metal man best
flesh with a clockwork efficiency. It was all a fake, of course – but the idea
of an automaton opponent – a thinking machine – was fascinating. Artificial intelligence is a natural application
for a game like chess, albeit a very challenging one. Simulating a chess board is easy – 64 squares,
32 pieces, 16 for each each of the 2 players. Implementing the rules is a little trickier
– but still feasible on early machines. Grids and rules are concrete. Logical. A natural
fit for a computer’s memory. More difficult is a simulation of not just
the game board – but an opponent. The game grows in complexity as it is played:
Decisions. Strategy. Hundreds of possible moves compound into trillions in just a few
turns. The first chess-playing programs appeared
in an academic setting: Alan Turing wrote one during his time at The University of Manchester. Dubbed ‘TurboChamp’, it started life as an
algorithm without a computer: a theoretical implementation only, but a working one nonetheless. In 1951, Christopher Strachey developed a
program to play the simpler game of draughts, and Dietrich Prinz implemented a practical
chess algorithm that could solve mate-in-two problems. Computers of this era were slow, taking minutes
or hours to deliberate all available permutations – but these early programs sowed the seed
for later AI routines, including those that found their way into video games. Hunt The Wumpus was originally written in
the BASIC programming language in 1972, and was later adapted for a number of other platforms. The premise is straightforward: you are in
a labyrinth comprising multiple chambers, one of which is occupied by a monster. As the title of the game implies, you must
navigate the maze, avoiding hazards such as bottomless pits and bats – and hunt the Wumpus. The catch is that you must do it by smell
alone – if you stumble into the monster, it’s game over. Instead, you must fire your arrows blindly
into an adjacent chamber: if you strike the beast, you win – but if you don’t, the Wumpus
will move. A very simple program, but from its few lines
of code emerge a virtual adversary: governed by simple rules yet affording an interesting
challenge to the player. 1980’s Pac-Man was a hugely popular arcade
title: and much of its appeal stems from its character. The yellow circle-section was the star, but
the 4 ghosts who chase him were each given names – and a different behaviour. All operate under 3 different modes: chase,
scatter and frightened, the latter limited to a brief duration after a power pill is
collected. Scatter mode sends each of the ghosts to a
separate corner, and occurs at preset times during a level. By default, the ghosts will chase Pac-Man:
and it’s this mode where the individual AI routines start to make the game interesting. The red ghost, also known as ‘Blinky’, is
the most aggressive – making a bee-line directly for Pac-Man’s position, and speeding up as
dots are consumed. ‘Pinky’ attempts to ambush the player, targeting
the position 4 squares ahead of Pac-Man’s direction of travel. Light-blue ghost ‘Inky’ has the most complex
targeting, seeking the tile opposite the red ghost’s position relative to Pac-Man – effectively
a pincer manoeuvre designed to trap the player. Finally, the orange ghost ‘Clyde’ will target
Pac-Man directly – until he gets too close, in which case he’ll retreat to his maze corner
instead. These four independent AIs give each of the
ghosts their own behaviour: and by extension, their own personality. Today, computer opponents are a little more
developed – far from perfect, but still capable of some surprisingly human manoeuvres. They perform best in games with a rigid ruleset,
with discrete turns: scaling from the simplest, such as the perfectly solved Tic-tac-toe;
to the greater scope of Grand Strategy games. A computer opponent can prove formidable to
play against, but a human player can exploit an AI player’s predictability to their own
advantage. Some games feature more adaptive intelligence:
shaping their own algorithms based on previous input – allowing for growth over time, and
the potential to overcome previously failed objectives. However, such techniques are difficult to
implement, and come with potentially unwanted side effects: there’s nothing worse than a
neurotic machine. Similarly challenging is parsing natural language:
computers struggle to understand English, much less compose a cogent response. This is fatally apparent in early text adventures:
only a sanctioned list of keywords are permitted, and attempting to interact with characters
in any way outside these bounds is often met with disappointment. Real-time decision making is an essential
part of first-person shooters – with reactive enemies more satisfying to kill than static
targets. Smaller touches, like when the monsters fight
each other in Doom, can be quite effective – a reaction to unexpected circumstances,
and a break from relentlessly targeting the player. The radio chatter of the enemy soldiers in
Half-Life gives an insight into their mind: and while such a stream of conciousness isn’t
entirely realistic, it’s nonetheless satisfying to hear their panic when you toss a grenade
in their direction. Like Pac-Man’s ghosts, diversity in the enemies
you face can make for more interesting battles: the varied makeup of the Covenant in Halo
helps communicate expected behavior. The smaller grunts are tough in larger groups
when led by an Elite: but decapitate their command and they’ll disperse, rendering them
less effective. Even the stalwart elites can be broken: get
too close and they might reposition – a contrast to the relentless assault more mindless enemies
might offer. AI remains an emerging field, and there are
many challenges left to overcome – but within the context of games, the ability to provide
a simple opponent is very valuable. The best AI is invisible: the illusion of
intelligence only breaks down when an opponent does something stupid. Without it, single player games would be a
hollow experience: No challenge, no surprise – no real opposition. Silicon versus flesh, man versus machine. In 1997 IBM’s Deep Blue – a supercomputer
– was able to best chess grandmaster Gary Kasparov. This was the first time a world champion was
defeated by a computer, under regular time restraint – and it wouldn’t be the last. Upon its victory, Deep Blue ranked amongst
the fastest machines on the planet: But today, you can find a similar amount of
number-crunching power in your pocket. With the right application, artificial intelligence
is smart – And it’s only going to get smarter. The impact of the work Turing did cannot be
understated. He defined the very essence of computing; Considered how they might think, And proved their military worth. His death a tragedy. His treatment, an embarrassment. As one war ended another began. A power vacuum fuelled by paranoia; Two nations
found themselves perpendicular: American individualism versus Soviet collectivism. Blue versus Red. Us versus Them. Coming up in part two: dreams of a distant
world, a scramble for technological superiority – and a huge thematic influence on early video
games. Thanks for watching – and until next time,
farewell.