[ExI] The Chess Room
steinberg.will at gmail.com
Sun Feb 28 19:17:57 UTC 2010
In the near future, America's greatest game theorists develop Benthic
Violet, the world's most complicated chess machine. Aside from playing
exceptionally well, BV has an astounding extra property. If any player
faces BV enough times, its algorithms will eventually be able to mimic this
player perfectly--from any initial move or set of moves, BV knows with
certainty exactly how the game will play out. It is infallible. Chess
players everywhere are aghast. Their game, it seems, has been solved. So a
team of chess masters resolves to set things straight. They form a group
composed of themselves along with other chess players along the skill
gradient. Every play style is represented, from crafty to frank, from
superb to awful. The players christen their group "The Last Stand."
The day has come for the match to begin, and TLS has devised a system of
play. Each player is given a number in accordance with their skill level,
with the best players being awarded the lowest numbers. Play begins
randomized across all players, each one having an equally likely chance to
play the next move. They have also written an ingenious algorithm to rate
how well the group seems to be doing, and when the algorithm reports that
things are starting to go sour, lower numbers are given a probabilistic
weight in accordance with just how sour the scene is. The match starts, and
the players all wait outside a secure room with a single door for their
chance to play. On entering, they see the board in front of them, along
with a list of previous moves, should they want to check the development.
This list is also broadcast on a screen outside the room. The games grind
on and on, and BV, as perhaps was expected, wins every time. But this is
not TLS's goal. After each game, technicians ask BV whether it is able to
emulate its opponent yet, and the answer is always the same: INSUFFICIENT
DATA FOR MEANINGFUL ANSWER. While BV is able to face TLS more and more
efficiently over time, it seems never to be able to predict the next move.
TLS is ecstatic. Champagne bottles are popped and flutes are filled.
Party horns are blown, shirts removed and swung over heads. As expected,
humanity has beaten the devil machine.
But, in the midst of this mirth, BV begins to emit some odd noises. Sparks
fly about and the humans take cover to observe. After a while, BV stops
making any noise. The air is still, and then, out of its print slot comes a
seemingly endless ribbon of paper. Finally this stops too, and all of TLS
goes over to observe. Kasparov looks at the readout, puts his hand to his
head, proclaims that he seems to have come into a case of the vapors, and
promptly faints. A less technically inclined player picks up the sheet and
reads off it. The first line, beginning with "M0" states a few possible
moves. The second line, which starts off "M0-1," again lists a few moves.
Upon closer examination of the list, the players find headings like
"M0-1-1", "M0-2-1-5-2-3", and so on. Each of these can be traced back to
the beginning in a backwards-branching fashion. At the very end of the
printout, a line states: LIKELY DEVELOPMENTS; P>.5. Under this are long
lists, each with a probability next to it:
"M0-12314231423255142314234231....123123: P>.72" A player sees it fit to
check these moves against the last game. A horrified look comes over him
and the paper falls out of his hands. The moves match well into the list.
More and more of these movelists are checked, and, eventually, the exact
playout of the game is found, P>.57. What does this mean? Has BV won? It
can't emulate the player perfectly, but it seems like, but for a different
path being taken, it could have. One would expect that, running more games,
eventually BV would get it right. At this moment of realization, a bright
light begins to flood out from under the door of the chess room. Suddenly,
an explosion ravages the area, destroying BV and TLS, yet, magnificently,
the final readout is carried on a thermal column up into the atmosphere,
where it flits into a cave somewhere in the Siberian mountains.
Millennia pass. In the time between then and now, a group of scientists,
having delved deeper and deeper into understanding our brains, have written
a formula that can take someone's chess history and compose a rough AI about
it, consciousness and all. While memories are lacking, this AI will act
very much like its complementary player. (This might be preposterous, but
it would have been the same if the original chess room instead analyzed
every action of all its players and attempted to create the consciousness
directly based on this. I am simply engaging in a lengthy chronological
syllogism.) In a neat coincidence, one of the scientists who wrote the
algorithm is an avid mountain climber, and, while hiking in the Siberian
mountains (which are now known for their beautiful flora, fauna, and general
verdant nature,) stumbles into that fated cave and finds the ancient
readout. He is confused, but takes the paper back to his laboratory and
runs it through the machine. The algorithm, like it often tends to do,
tells the scientist that it has found a number of probable consciousnesses
for the movelist. He tells it to choose the most probable one. A voice
ekes out of the machine: "Hello..."
And this, unfortunately, is where our story ends. A shampoo bottle inspired
me to write it. In the shower, I realized that the placement of the bottle
and the amount of shampoo inside and other factors are determined by the
actions of my family. The history of the bottle is dependent on many
consciousnesses, and so "understanding" of it cannot exist on one mind. The
bottle is a piece of external information that relies on multiple actions by
different people, and, perhaps if we were able to understand a person based
on their use and placement of this bottle, we could extend this to roughly
understand *the family* that uses the bottle. Then do the actions revolving
around it contain, in a sense, some sort of consciousness than cannot be
understood without taking multiple minds into question?
I thought about it more. The placement is in essence a computation by the
brain (for a moment ignore the questions of traditional computability and
say it may be a "non-computable computation.") If our brains are a network
of interconnected computations leading somehow to consciousness, might any
actions requiring the input of multiple minds require a sort of emergent
consciousness? When events take place on Earth, humans put malleable
information into the environment, be it in the form of money, television,
war, shampoo, talking, sex, mountain climbing, shampoo. Other humans
receive and re-emit this information. Humanity as a whole, it seems, could
have this (pardon any panpsychic jargon) "overmind" associated with it. In
fact, humanity seems to be following, roughly, a calculable progression.
Trends exist; this is undeniable.
Now, if one wants, this could be extended to places further than Earth.
Earth receives inputs in the form of events and produces outputs, hopefully
many more in the future in the form of interstellar travel. Maybe,
(probably?) alien species with minds of their own exist that will eventually
come into contact with us. Essentially, what I am asking is: does a system
which contains consciousness, as a simple result of this consciousness being
there, have a slight consciousness of its own, though it may act in a much
more convoluted way, over a much longer period of time? Does the universe,
as the "Old One," perform large-scale mind-like calculations at a very slow
scale? If any of you remember Hofstadter's Siamese twins, you understand
the idea of a mind existing over what seem to be multiple systems. Maybe
the distances involved here are too large for something to emerge, but this
might be compensated by the inordinate amounts of time needed for these
calculations to take place.
What say you?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the extropy-chat