Randomness does not exist
Reification
This is part 2 of a series of essays on the nonexistence of
things that are commonly treated as real. Reification is the treatment of an
abstract idea as though it were a real thing. The modern world is full of
reifications, but people seem to be unaware of their true nature.
The idea of randomness occupies a special place as a
quasi-reification. This is because those who study probability theory and are
deeply involved with probabilistic models understand the true nature of
randomness. They might call it “chance,” or “uncertainty,” which are good ways
to think about “randomness.”
While probability theory serves as a valuable tool in
scientific inquiry, it is essential to recognize that the concept of randomness
within this framework should not be conflated with the uncertainty stemming
from our incomplete understanding and control over natural phenomena.
Probability theory offers a systematic way to quantify and manage this inherent
uncertainty, enabling scientists to test hypotheses and make informed decisions
in the face of incomplete information.
Then there are the quantum mechanics, or rather, people who
use the phrase “quantum” to describe anything that they don’t understand. To
these people, randomness is a fundamental property of the universe, and instead
of being governed by physical forces and attributes, they believe the universe
is governed by mathematical models. While I cannot delve deeply into the topic
of quantum mechanics, I can relate how the general absorption of the theory has
percolated through society, so that people ladle out the phrase “quantum
mechanics” with a fairly liberal, handwaving indulgence. As a result of this,
and perhaps other misconceptions about chance and randomness, people have become
accustomed to the idea that “random” stuff just happens for no reason.
It is true that very often stuff happens and the reasons are
unknown or obscure. A black cat walks across our path and we find that we didn’t’
win the lottery for some reason. People get struck by lightening. Some people
actually DO win the lottery. We chalk all these things up to a property called “randomness,”
and dismiss it and admire it as some magical property that the universe conveys
upon certain things. We flip coins, shuffle cards, put different colored
marbles into black bags, draw lots, read entrails, and so on because we believe
that there is a kind of mystical process at work. However, what we are really
doing when we randomize something is not confer upon the thing a release from
the rules of causation, but making ourselves ignorant of the outcome.
Randomness is the shadow
Randomness is really our inability to predict the outcome of
something because we do not have enough information. For example, if I put five
black balls and three white balls into a bag and drew one without looking, I
would have a 3/5ths chance to pull a white one. However, if the bag is clear
and I can see which one I am choosing, then the “randomness” vanishes. Randomness
is uncertainty, nothing more.
Scientific experiments often use the concept of randomness
to prove their points. After we isolate a variable, we try to control for all
the other variables, and so the independent variable (the one we manipulate) should
be entirely responsible for the outcome. Experiments are never quite this
simple, though, because when we manipulate the independent variable, we usually
find that the dependent variable does not change on a one-to-one relationship,
but that the independent variable “has an effect” on the dependent variable,
usually expressed as a percentage or a probability. For example, an increase in
the amount of sunlight was responsible for a 40% increase in plant growth. The
other 80% of plant growth was caused by “randomness,” in other words, noise.
However, noise is really just other independent variables that we have not
controlled for. If we take rainfall into the equation, we find that sun and
rain both account for 80% of the growth rate of the plant.
So, in this example randomness was initially responsible for
60% of the plant growth when only sunlight was taken into account, but when we took
into account sunlight and rain, the randomness was only responsible for 20% of
the growth. By extension, the more certainty we have about the causes of the
effects, the less uncertainty, meaning the less we can attribute to “randomness.”
From this, we conclude that randomness is not a fundamental attribute
of the universe, but an absence of control. The more we control and account for
independent variables, the less is left to chance, the less wiggle room
randomness has to play around with the outcomes. Randomness is like the shadow
of control. It is not a thing in itself, but the absence of a certainty, like a
shadow is the absence of light.
Infamous P-value
This brings us to a conundrum, though. Science requires the
idea of “chance alone,” to prove hypotheses. We express our certainty of the effect
of an independent variable on the dependent one as a probability of having an effect
compared to “chance alone.” Frequently, the validity of a test is expressed by
its “p-value,” usually in the format (p < 0.05). The value 0.05, or 5% was
chosen arbitrarily. The idea is that if something can happen by “chance alone,”
more than say 5% of the time (p > 0.05), then we cannot rule out the null
hypothesis, which presumes that there is no relationship between the
independent and dependent variables. Anything more than that, and we hesitate
to say for sure that the variable had an effect. Interestingly, this means that
approximately 5% of all studies are false positives, but that is a tangent.
If we are defining the success of a scientific study by
comparing the probability of our results to “chance alone,” and we know that “chance
alone,” means nothing more than unaccounted variables, what are we saying? The
problem is the conflation of the mathematical ideal of a random variable with the
reality of unaccounted variables that would otherwise determine the outcomes of
real-life events. Although the mathematical world of probability theory might
be a good way to manipulate uncertainty in our equations so we can approximate
predictability, there is a danger of confusing the abstract mathematical
principles with the unknown, complex, and uncertain real variables that we are
unaware of acting in the real world. As symbols that we can manipulate, random
variables can be useful, but as proxies for the complexity of the real world,
they are the shadows of certainty, concealing the vast complexity of the
problem behind the veil of ignorance or lack of control.
Death of science
Everyone knows how tossing a coin works. The goal is to toss
the coin and expect a 50/50 chance of getting a head or a tail. It works
because we lack the ability to flip the coin precisely enough to predict the outcome.
However, suppose I was able to flip a coin so that it always flipped exactly 3
times, and so I could predict the outcome? Is it random? No. What if I could
flip it 4 times, but still predict the outcome? Still not random. As long as I
can exert control of the number of flips, it’s not random. So where does
randomness come in? When I lose track of the number of flips, or lose control
of the number of flips? My losing track does not create randomness, though: it
creates ignorance. Were I to practice enough, I would be able to predict the
outcome of a coin toss after any number of flips.
Another example is a game of pool. Most of us break the triangle of balls and they spin off in various directions and we look at the scattering as random. However, a pool trick-shooter can get the balls to go in precisely predetermined directions and velocities. The only difference is the skill and knowledge of the trick shooter. Randomness is eliminated by knowledge and control.
The point here is that randomness shrinks as certainty
grows, and the goal of all sciences is to increase certainty. As we increase
certainty, the idea that anything is random naturally shrinks, so it is possible
to extrapolate our thinking to a scenario in which certainty is absolute and
randomness is nonexistent. In fact, randomness IS nonexistent, because it doesn’t
exist: it is the lack of certainty, and nothing more. The irony is that science
depends on the concept of randomness to function. We randomly assign test
conditions. We use p-values to measure the validity of our tests. In this sense, scientists use the mathematical concept of the random variable, which is distinct from the real-world meaning of random. This is why it is important not to conflate the abstract concept of randomness with the real-world version. Even as
science destroys randomness it must rely on randomness to advance in an
ever-tightening circle of annihilation. Science
ends when everything is known, and randomness is a memory of our ignorance.
The theoretical limit to our knowledge, though, is absolute,
because the map cannot be the territory. Our model of the universe can never be
as complex as the universe itself, because then it would be the universe. Hence,
we must rely on heuristics, approximations, and randomness for the areas where
we cannot or choose not to know for certain.
No comments:
Post a Comment