The Computers of Star Trek (15 page)

Think of the current as air. Think of the two points on the circuit as two tires. Think of the arrow as a hose. Clearly, if one tire (Point A) has twice as much pressure as the other tire (Point B), the air in the hose flows from the first tire into the second tire. Air flows until the pressure in both tires is equal.

Current is the same. It moves from the circuitry point with more pressure to the circuitry point with less pressure. One volt forces one amp of current through one ohm of resistance. (Resistance opposes the flow of current/electrons.) In 1827, Dr. Georg Simon Ohm defined what became known as Ohm's Law: The amount of current in a circuit is directly proportional to the voltage applied to the circuit and inversely proportional to the resistance. In addition, we have Watt's Law: The power (as measured in watts) in a circuit equals the current (amperes) multiplied by the voltage (ohms). If, for example, we have a 150-watt light bulb drawing 1.5 amps, then we require 100 volts.

Now let's try to determine how much current is implied by half a million amps, the amount that melts Data. We don't know Data's resistance, only that his positronic brain has “several layers of shielding to protect [him] from power surges.” The average computer today might use 5 or 3.3 volts. Voltages in computer systems are shrinking. We're told that Data runs on “microhydraulic power.” Either he uses microvolts (10-6) or nanovolts (10-
⁹
).

Because Data's built from nanocircuits, he probably requires nanovolt components. To keep things simple for now, let's assume that when we add up all the resistance in Data, we have a full 1 volt. Remember that our 150-watt light bulb uses 100 volts. Let's apply the numbers to the algorithm:

So based on our 150-watt light bulb, we figure that Data short circuits if we apply power roughly equal to that of 3,333.33 ordinary 150-watt light bulbs (500,000 / 150 = 3,333.33 light bulbs). Data supposedly fries at 500,000 watts. The typical person might generate a few hundred watts just typing a Data chapter on the keyboard. The human brain consumes approximately the energy of an ordinary light bulb.

Something is wrong with our assumption about the 1 volt. We must move to micro or nanovolts to reduce the amount of energy Data needs and generates. At 1 microvolt, Data would generate 500 watts, which is closer to the human body.

Still, it's puzzling. It seems that Data would have a total system meltdown long before the current hit half a million amps. And if he somehow runs on close to half a million amps, he must glow in the dark; or as our friend Bill Tate says, “Data must have one heck of a glowing personality.”

These are very crude guesses, of course. We'd like to know how much power Data generates autonomously to run himself. And given that Data has only some undefined shielding to protect his positronic brain and only breathing as an internal cooling system,
how does he withstand power surges of more practical amounts, such as any number below 500,000 amps?

Perhaps Data does generate 500 watts of power using a total of 1 microvolt of pressure. Perhaps it requires four cooling-fan noses to keep his system from shutting down. Perhaps he needs giant nostrils to suck in the air to keep his legs from turning off during system overheating. If he has giant nostrils, though, nanotech diseases could enter and destroy him. Not good.

Data's system cooling probably will occur through skin pores, not nostrils; but we must take his nostrils and lungs as concrete system designs. His humanoid composition is a given of
Star Trek.

We're told that Data's circulatory system distributes biochemical lubricants throughout his body. But we don't know what the lubricants are or what they do. We can only assume that they somehow oil his parts. Perhaps they are released periodically by nanomachines in Data's body. But what if the nanomachines are off slightly? Would the lubricants slosh throughout Data's internals, causing major system damage? He hasn't any method to repair himself aside from running internal diagnostics of his positronic brain. Even Geordi, who knows better than anyone else how to repair Data, would surely be clueless when it comes to a catastrophe such as an over-internal-lubrication job by some malfunctioning nano-equipment. Geordi uses large wrenches to repair Data, not nanosurgical devices.

Besides, why does Data require biochemical lubricants? What does the bio aspect bring to the mix? Data comments that he rarely needs the services of Dr. Crusher (“Data's Day,”
TNG
), which implies that sometimes he does need medical help. In what cases, and under what conditions, does Dr. Crusher service Data? These are things that we'd love to see explored in future
Star Trek
movies.

According to Data, his circulatory system regulates his microhydraulic power and provides him with a human pulse. We're clueless why Data needs a pulse. If it's solely to make him fit into human crowds without being noticed, then why is his skin greenish? We'd notice a man with green skin more quickly than we'd feel his pulse.
^o
Yet Data claims that nobody has ever asked him if his hair grows, and indeed, nobody's ever noticed that he breathes. What were those computer scientists doing when they first found Data and sent him to Starfleet Academy? Didn't they notice that the only android in existence breathes and regulates the growth of his own hair? Or that he has a pulse?

While it's clear that a machine like Data can't be constructed today, if the science of robotics continues to advance at the speed of general computer technology, then creating a robot like Data should be possible within the next century or two. But as more than one mad scientist in old SF movies has learned to his despair, creating the body isn't enough. It's the brain that matters.

In
Star Trek
, mankind's greatest scientists have been unable to produce functioning androids for hundreds of years. They have one amazing android called Data, but they don't understand the technology that created him. We're asked to believe that the reclusive Dr. Noonien Soong was the only scientist who ever perfected a positronic brain. That no other cybernetist among the many billions of inhabitants of the Federation has ever been able
to duplicate his work. And that androids other than Data don't exist.
^p

In “Inheritance,” Juliana Tainer says that Data is the fifth android built by Dr. Soong.
^q
The first three artificial beings were total or partial failures. The fourth was named Lore, but he was taken apart by Dr. Soong when he exhibited strong antisocial behavior (“Brothers,” TNG). Since creating an android body three centuries from now doesn't appear to be an insolvable task, the problem obviously was in their brains.

In “The Offspring,” Data builds another android like himself. This new android, whom Data considers his child, is named Lal and is similar to Data but in some ways more advanced. Still, her positronic brain fails and she dies. Not even Data seems to know exactly how his mind works.

In “Evolution” (
TNG
), Captain Picard permits an alien intelligence to enter Data's brain. Apparently, it's simple for the alien nanites to figure out exactly how Data's brain operates. As Data tells Picard, “I can easily furnish the nanites with the schematic design of my neurological structure. Entering my neural net would require no more than their most basic skills.” It's odd that a microscopic alien intelligence can penetrate, understand, and
use Data's brain instantly, whereas neither human scientists nor Data can reproduce it.
^*

It's pretty clear that all of Data's knowledge is in his skull, and not dispersed throughout his body, even though that would make more sense. Geordi often opens Data's head, never, for instance, his stomach to fiddle with his positronic net, his “brain.” Nor does Data seem to have the same redundancy that's built into the ship's computer. When his head's gone, he's inoperable (“Time's Arrow,”
TNG
).

Data is more than a mere computer walking about in an artificial humanoid body. As an android, he is more than a machine, though less than human. This fact is demonstrated quite clearly in “The Schizoid Man.” In that adventure, Data meets Dr. Ira Graves, who was Dr. Soong's mentor. In a sense, Graves is Data's grandfather. Unfortunately, when the two meet, Graves is dying. He deactivates Data and transfers his mind into Data's positronic brain. Though Data has no capacity for emotion (this is before he retrieved the emotion chip from Lore), Graves in Data's mind exhibits a full sweep of emotions, ranging from love to anger. Picard and Graves/Data get involved in a philosophical argument about whether Data, technically a machine, has the right to exist if it means Graves will die. Later, when Graves transfers his mind into the ship's computer, there's no evidence of any emotions; the human element is gone, implying that Data's positronic brain is quite different from the
Enterprise
computer core.

According to Data, his brain possesses a storage capacity of 100,000 terabytes of memory, or 10
¹⁷
bytes. He also states that his positronic brain processes 60 trillion computations per second. Checking back to Chapter 2, we note that Data thus has enough memory to hold 1,000 Libraries of Congress. It's not close to the memory capacity of the
Enterprise
, but it's still a lot of memory. When Geordi and Data discuss letting Data run some ship subroutines through his brain, it's not that unbelievable (“A Fistful of Datas,”
TNG
). Except perhaps for the subroutines they test—weapons systems, for example, not a prime choice.

Many scientists feel the human brain has a maximum memory capacity of three terabits. Which would mean Data has the memory storage of more than a quarter-million people. With that much memory, there's plenty of room for subroutines to coordinate all the necessary movements Data needs to handle such activities as walking, talking, even playing the violin. But all that knowledge can't make him creative. He still has to be taught what to do before he can do it. While Data's positronic brain can process 60 trillion computations per second, it's estimated that the human brain can process 10 trillion bytes per second. He thinks faster than people, but not that much faster.

Data's positronic brain is supposed to be a neural network with lots of parallel processing. Let's try to figure out what this means.

We talked a little about neurology and artificial intelligence in Chapter 5. The human brain contains approximately 100 billion to 200 billion neurons that fire about 10 million billion times per second.
^r
Each neuron connects to roughly 10 thousand other
neurons. This is how the brain handles trillions of operations per second. It's an extremely complex neural network.

A computer-neural network is a simplified version of a biological neural network. In the biological form, a neuron accepts input from its dendrites and supplies output to other neurons through its axons. The neuron applies weights to the connections, or synapses, between dendrites and axons. A higher weight might be applied to a synapse related to touching fire than to a synapse about seeing the pretty color of fireball orange.

In the computerized version, each “input neuron” feeds information into every neuron in what is called the hidden layer, which may have one or multiple layers of neurons. If the hidden layer has two layers of neurons, for example, then every neuron in the first hidden layer feeds into every neuron in the second. Every neuron in the last hidden layer feeds into neurons in the output layer.

The designer of a neural net provides different weights for the connections among neurons. While our brain receives input from many sources, such as sensations on our skin, what we hear, what we smell, and so forth, an artificial neural network takes input only from values we provide, and then it weighs everything and supplies a best-guess answer. We know, for example, that 1.00 + 1.00 equals 2.00. An artificial-neural net may not find the problem so easy. It may guess that the answer is 1.98, or perhaps 2.04. But the artificial brain will do quite well in guessing correctly between say, a nerf football and a soccer ball. Or even a nerf soccer ball versus a hard soccer ball. Both are spheres. Both are the same size. One is soft, the nerf; one is hard, the soccer ball.

Various methods exist for applying weights to artificial neurons, and for assembling the input, hidden, and output layers into network architectures. A neural network learns by adjusting the weights given to its neurons. A very common neural net architecture
is called back propagation, which compares forecasts to actuals, then adjusts the weighted interconnections among neurons. Over time, as it compares more forecasts to actuals, the neural weights become more accurate. In a sense, the neural net itself has learned and adjusted to its environment.

The big question: is Data's positronic brain possible? It seems highly likely.