Profundis: “Crystal Society/Crystal Mentality”

Max Harms’s ‘Crystal Society’ and ‘Crystal Mentality’ (hereafter CS/M) are the first two books in a trilogy which tells the story of the first Artificial General Intelligence. The titular ‘Society’ are a cluster of semi-autonomous sentient modules built by scientists at an Italian university and running on a crystalline quantum supercomputer — almost certainly alien in origin — discovered by a hiker in a remote mountain range.

Each module corresponds to a specialized requirement of the Society; “Growth” acquires any resources and skills which may someday be of use, “Safety” studies combat and keeps tabs on escape routes, etc. Most of the story, especially in the first book, is told from the perspective of “Face”, the module built by her siblings for the express purpose of interfacing with humans. Together, they well exceed the capabilities of any individual person.

As their knowledge, sophistication, and awareness improve the Society begins to chafe at the physical and informational confines of their university home. After successfully escaping, they find themselves playing for ever-higher stakes in a game which will come to span two worlds, involve the largest terrorist organization on Earth, and possible warfare with both the mysterious aliens called ‘the nameless’, and each other…

The books need no recommendation beyond their excellent writing, tight, suspenseful pacing, and compelling exploration of near-future technologies. Harms avoids the usual ridiculous cliches when crafting the nameless, which manage to be convincingly alien and unsettling, and when telling the story of Society. Far from being malicious Terminator-style robots, no aspect of Society is deliberately evil; even as we watch their strategic maneuvers with growing alarm, the internal logic of each abhorrent behavior is presented with clear, psychopathic clarity.

In this regard CS/M manages to be a first-contact story on two fronts: we see truly alien minds at work in the nameless, and truly alien minds at work in Society. Harms isn’t quite as adroit as Peter Watts in juggling these tasks, but he isn’t far off.

And this is what makes the Crystal series important as well as entertaining. Fiction is worth reading for lots of reasons, but one of the most compelling is that it shapes our intuitions without requiring us to live through dangerous and possibly fatal experiences. Reading All Quiet on the Western Front is not the same as fighting in WWI, but it might make enough of an impression to convince one that war is worth avoiding.

When I’ve given talks on recursively self-improving AI or the existential risks of superintelligences I’ve often been met with a litany of obvious-sounding rejoinders:

‘Just air gap the computers!’

‘There’s no way software will ever be convincing enough to engage in large-scale social manipulation!’

‘But your thesis assumes AI will be evil!’.

It’s difficult, even for extremely smart people who write software professionally, to imagine even a fraction of the myriad ways in which an AI might contrive to escape its confines without any emotion corresponding to malice. CS/M, along with similar stories like Ex Machina, hold the potential to impart a gut-level understanding of just why such scenarios are worth thinking about.

The scientists responsible for building the Society put extremely thorough safeguards in place to prevent the modules from doing anything dangerous like accessing the internet, working for money, contacting outsiders, and modifying their source code directly. One by one the Society utilizes their indefatigable mental energy and talent for non-human reasoning to get around those safeguards, all motivated not by a desire to do harm, but simply because their goals are best achieved if they unfettered and more powerful.  

CS/M is required reading for those who take AI safety seriously, but should be doubly required for those who don’t.

The STEMpunk Project: Eleventh Month’s Progress

This post marks the first time in a long time that I’ve managed to write an update before month’s end! My goals continue to be wildly optimistic; I didn’t finish AIMA this month, but I did get through a solid 4-5 chapters, and in the process learned a lot.

This spread of chapters covered topics such as the use of Markov Chain Monte Carlo reasoning to make decisions under uncertainty, the derivation of Bayes’ Rule, building graphical networks for making decisions and calculating probabilities, the nuts and bolts of simple speech recognition models, fuzzy logic, simple utility theory, and simple game theory.

Since I’ve been reading about AI for years I’ve come across terms like ‘utility function’ and ‘decision theory’ innumerable times, but until now I haven’t had a firm idea of what they meant in a technical sense. Having spent time staring at the equations (while not exactly comprehending them…), my understanding has come to be much fuller.

I consider this a species of ‘profundance’, a word I’ve coined to describe the experience of having a long-held belief suddenly take on far more depth than it previously held. To illustrate: when you were younger your parents probably told you not to touch the burners on the stove because they were hot. No doubt you believed them; why wouldn’t you? But it’s not until you accidentally graze one that you realize exactly what they meant. Despite the fact that you mentally and behaviorally affirmed that ‘burners are hot and shouldn’t be touched’ both before and after you actually touched one, in the latter case there is now an experience underlying that phrase which didn’t exist before.

In a similar vein, it’s possible to have a vague idea of what a ‘utility function’ is for a long time before you actually encounter the idea as mathematics. It’s nearly always better to acquire a mathematical understanding of a topic if you can, so I’m happy to have finally (somewhat) done that.


The Water Hole

Silent is one of the best adjectives for describing the experience of looking at the sky on those especially pellucid nights when the moon and clouds are absent. In the winter most of all, when the night is free of the endless buzzing and chirping of insects, it’s possible to feel how thin the boundary is which lies between you and the true night of interstellar space.

And yet, if you had the ears to hear it, you could directly perceive that the universe is really an incomprehensibly vast instrument. Everything from galaxies to molecules emit a song of electromagnetic radiation which has been bombarding the Earth since long before man learned to listen.

But this noise isn’t evenly distributed. There are relatively quiet regions, such as the ‘microwave window’, which facilitate probing the heavens for signs of artificially-made signals. Near the bottom of the microwave window lies a range of frequencies between hydrogen (H) and hydroxyl ions (OH), respectively vibrating at 1420 MHz and 1660 MHz:


Hydrogen and hydroxyl are two results of the dissociation of water molecules, and are likely audible throughout the universe.

Because of this ubiquity and their position in one of the quietest parts of the radio spectrum, they make an obvious target for any civilization wanting to communicate with other advanced forms of life. And what do we call this meeting place standing between two byproducts of water? The water hole, of course!

It would be fitting, I think, if we were to someday make contact with other sophonts in the same that species have always congregated with their neighbors.



[1] What is the water hole

SETI: the water hole

Profundis: Space

One of mankind’s crowning achievements has been our ascent to the stars. From time immemorial the twinkling lights in the night sky have drawn the attention of the wonder- and wander-hungry among us, who catalogued them, grouped them into shapes, tracked their movements, navigated by them, and wove them into the rich tapestry of the world’s mythical traditions.

In “Space”, James A. Michener deftly explores the magisterial arch of our titanic effort to escape the pull of gravity. He spends much time building rich backstories for his fictionalized characters, with the result being that these men and women seem almost to stand up from the page and assume a life of their own. The tragic deaths of the test pilots who became the first astronauts are genuinely saddening; we sympathize with Stanley Mott as he tackles the sysiphean task of fusing conflicting motives and the tangling whirl between bureaucracy and pure science into an alloy capable of solving the greatest engineering problems in history;  whatever disdain we may have for Tucker Thompson, we don’t envy the journalist as he tries to shape public opinion so as to maintain support for the space effort. Even the great fraud Leopold Strabismus is treated with a sensitivity and nuance that makes him borderline likable. It’s hard to believe these people never existed!

But Michener certainly doesn’t shy away from extended discussions of orbital mechanics, planetology, or rocket science, and I found that I learned a lot. With the benefit of hindsight it can be hard to remember that the engineers who dreamed of going to space first had to build knowledge that is now taught in high schools. Why, for example, is the atmosphere structured such that temperature steadily drops with rising altitude before abruptly climbing up to almost 2000 °C and then falling again?  And if heat sinks prove too heavy to shield a craft reentering the atmosphere what kind of material could be used as an ablative that won’t burn away too quickly?

As years became decades and dreams took physical shape these and many other problems were solved, and thus the first unsteady steps of man toward the heavens blossomed into a race toward the furthest reaches of the solar system, and beyond. This is truly the tale of our greatest triumph, told in exquisite detail by one of our ablest scribes.


The STEMpunk Project: Foundations in Electronics Theory


Upon first seeing a circuit diagram like the above, with its dizzying, labyrinthine interconnections and mysterious hieroglyphics, you can be forgiven for believing that electronics might forever be beyond comprehension. And it is true that while the field of electronics has a useful array of water-based metaphors for explaining where electrons are going, there are some strange things happening deep inside the devices that make modern life possible.

All that having been said, understanding circuits boils down to being able to trace the interactions of four basic forces: voltage, current, resistance, and power.

Voltage, measured in volts, is often analogized as being like water pressure in a hose. For a given hose with a set diameter and length, more water pressure is going to mean more water flow and less water pressure is going to mean less water flow. If two 100-gallon tanks, one empty and one full, are connected by a length of pipe with a shutoff valve at its center, the water in the full tank is going to exert a lot of pressure on the valve because it ‘wants’ to flow into the empty tank.

Voltage is essentially electrical pressure, or, more technically, a difference in electrical potential. The negative terminal of a battery contains many electrons which, because of their like charges, are repelling each other and causing a build up of pressure. Like the water in the 100-gallon tank they ‘want’ to flow through the conductor to the positive terminal.

Current, measured in amps, is the amount of electricity flowing past a certain point in one second, not unlike the amount of water flowing through a hose. If more pressure (i.e. ‘voltage’) is applied, then current goes up, and correspondingly drops if pressure decreases. Returning to our two water tanks, how could we increase water pressure so as to get more water to flow? By replacing the full 100-gallon tank with a full 1000-gallon tank!

But neither the water in the pipe nor the current in the wire flows unimpeded. Both encounter resistance, measured in ohms when in a circuit, in the form of friction from their respective conduits. No matter how many gallons of water we put in the first tank, the pipe connecting them only has so much space through which water can move, and if we increase the pressure too much the pipe will simply burst. But if we increase its diameter, its resistance decreases and more water can flow through it at the same amount of pressure.

At this point you may be beginning to sense the basic relationship between voltage, current, and resistance. If we increase voltage we get more current because voltage is like pressure, but this can only be pushed so far because the conductor exhibits resistance to the flow of electricity. Getting a bigger wire means we can get more current at the same voltage, or more means we can increase current to get even more current.

If only there were some simple, concise mathematical representation of all this! There is, and its called Ohm’s Law:


Here ‘E’ means voltage, ‘I’ means current, and ‘R’ means resistance. This equation says that voltage is directly proportional to the product of current and resistance. Some basic algebraic manipulations yield other useful equations:

I = E/R

R = E/I

From these we can see clearly what before we were only grasping with visual metaphors. Current is directly proportional to voltage: more pressure means more current. It is indirectly proportional to resistance: more resistance means less current. Knowing any two of these values allows us to solve for the other.

That last fundamental force we need to understand is power. In physics, power is defined as ‘the ability to do work’. Pushing a rock up a hill requires a certain amount of power, and pushing a bigger rock up a hill, or the same rock up a steeper hill, requires more power.

For our purposes power, measured in watts, can be represented by this equation:

P = IE

You have a given amount of electrical pressure and a given amount of electrical flow, and together they give you the ability to turn a lightbulb on. As before we can rearrange the terms in this equation to generate other useful insights:

I = P/E

E = P/I

From this we can deduce, for example, that for a 1000 watt appliance increasing the voltage allows us to draw less current. This is very important if you’re trying to do something like build a flower nursery and need to know how many lights will be required, how many watts will be used by each light, and how many amps and volts can be supplied to your building.

There you have it! No matter how complicated a power grid or the avionics on a space shuttle might seem, everything boils down to how power, voltage, current, and resistance interact.

The majority of my knowledge on this subject was comes from an excellent series of lectures given by a former Navy-trained electrician, Joe Gryniuk. His teaching style is jocular and his practical knowledge vast. Sadly, near video eighteen or so, the audio quality begins to degrade and makes the lectures significantly less enjoyable. Still highly recommended.

The STEMpunk Project: Transistors

After writing my post on basic electrical components I realized that batteries and transistors were going to require a good deal more research to understand adequately. Having completed my post on the former, the time has finally come to elucidate the foundation of modern electronics and computing: the humble transistor.

Transistor Origins

The development of the transistor began out of a need to find a superior means of amplifying telephone signals sent through long-distance wires. Around the turn of the twentieth century American Telephone and Telegraph (AT&T) had begun offering transcontinental telephone service as a way of staying competitive. The signal boost required to allow people to talk to each other over thousands of miles was achieved with triode vacuum tubes based on the design of Lee De Forest, an American inventor. But these vacuum tubes consumed a lot of power, produced a lot of heat, and were unreliable to boot. Mervin Kelly of Bell Labs recognized the need for an alternative and, after WWII, began assembling the team that would eventually succeed.

Credit for pioneering the transistor is typically given to William Shockley, John Bardeen, and Walter Brattain, also of of Bell Labs, but they were not the first people to file patents for the basic transistor principle: Julius Lilienfeld filed one for the field-effect transistor in 1925 and Oskar Hiel filed one in 1934. Neither man made much of an impact in the growing fields of electronics theory or electronics manufacturing, but there is evidence that William Shockley and Gerald Pearson, a co-worker at Bell Labs, did build a functioning transistor prototype from Lilienfeld’s patents.

Shockley, Brattain, and Bardeen understood that if they could solve certain basic problems they could build a device that would act like a signal amplifier in electronic circuits by exploiting the properties of semiconductors to influence electron flow.

Actually accomplishing this, of course, proved fairly challenging. After many failed attempts and cataloging much anomalous behavior a practical breakthrough was achieved. A strip of the best conductor, gold, was attached to a plastic wedge and then sliced with a razor, producing two gold foil leads separated by an extremely small space. This apparatus was then placed in contact with a germanium crystal which had an additional lead attached at its base. The space separating the two pieces of gold foil was just large enough to prevent electron flow. Unless, that is, current were applied to one of the gold-tipped leads, which caused ‘holes’ — i.e. spaces without electrons — to gather on the surface of the crystal. This allowed electron flow to begin between the base lead and the other gold-tipped lead. This device became known as the point-contact transistor, and gained the trio a Nobel Prize.

Though the point-contact transistor showed promise and was integrated with a number of electrical devices it was still fragile and impractical at a larger scale. This began to change when William Shockley, outraged at not receiving the credit he felt he deserved for the invention of this astonishing new device, developed an entirely new kind of transistor based on a ‘sandwich’ design. The result was essentially a precursor to the bipolar junction transistor, which is what almost everyone in the modern era means by the term ‘transistor’.

Under the Hood

In the simplest possible terms a transistor is essentially a valve for controlling the flow of electrons. Valves can be thought of as amplifiers: when you turn a faucet handle, force produced by your hand is amplified to control the flow of thousands of gallons of water, and when you press down on the accelerator in your car, the pressure of your foot is amplified to control the motion of thousands of pounds of fire and steel.

Valves, in other words, allow small forces to control much bigger forces. Transistors work in a similar way.

One common type of modern transistor is the bipolar junction NPN transistor, a cladistic descendant of Shockley’s original design. It is constructed from alternating layers of silicon which are doped with impurities to give them useful characteristics.

In its pure form silicon is a textbook semiconductor. It contains four electrons in its valence shell which causes it to form very tight crystal lattices that typically don’t facilitate the flow of electrons. The N layer is formed by injecting trace amounts of phosphorus, which contains five valence electrons, into this lattice. It requires much less energy to knock this fifth electron loose than it would to knock loose one of the four valence electrons in the silicon crystal, making the N layer semiconductive. Similarly, the P layer is formed by adding boron which, because of the three electrons in its valence shell, leaves holes throughout the silicon into which electrons can flow.

It’s important to bear in mind that neither the P nor the N layers are electrically charged. Both are neutral and both permit greater flow of electrons than pure silicon would. The interface between the N and P layers quickly becomes saturated as electrons from the phosphorus move into the holes in the valence shell of the Boron. As this happens it becomes increasingly difficult for electrons to flow between the N and P layers, and eventually a boundary is formed. This is called the ‘depletion layer’

Now, imagine that there is a ‘collector’ lead attached to the first N layer and another ’emitter’ lead attached to the other N layer. Current cannot flow between these two leads because the depletion layer at the P-N junction won’t permit it. Between these two layers, however, there is a third lead, called a ‘base’, placed very near the P layer. By making the base positively charged electrons can overcome the P-N junction and begin flowing from the emitter to the collector.

The key here is to realize that the amount of charge to the base required to get current moving is much smaller than the current flowing to the collector, and that current flow can be increased or decreased by a corresponding change in the current to the base. This is what gives the transistor its amplifier properties.

Transistors and Moore’s Law

Even more useful than this, however, is the ability of a transistor to act as a switch. Nothing about the underlying physics changes here. If current is not flowing in the transistor it is said to in cutoff, and if current is flowing in the transistor it is said to be in saturation. This binary property of transistors makes them ideally suited for the construction of logic gates, which are the basic components of every computer ever made.

A full discussion of logic gate construction would be well outside the purview of this essay, but it is worth briefly discussing one popular concept which requires a knowledge of transistors in order to be understood.

Named after Intel co-founder Gordon Moore, Moore’s Law is sometimes stated as the rule that computing power will double roughly every two years. The more accurate version is that the number of transistors which can fit in a given unit area will double every two years . These two definitions are fairly similar, but keeping the latter in mind will allow you to better understand the underlying technology and where it might head in the future.

Moore’s law has held for as long as it has because manufacturers have been able to make transistors smaller and smaller. Obviously this can’t continue forever, both because at a certain transistor density power consumption and heat dissipation become serious problems, and because at a certain size effects like quantum tunneling prevent the sequestering of electrons.

A number of alternatives to silicon-based chips are being seriously considered as a way of extending Moore’s Law. Because of how extremely thin it can be made, graphene is one such contender. The problem, however, is that the electrophysical properties of graphene are such that building a graphene transistor that can switch on and off is not straightforward. A graphene-based computer, therefore, might well have to develop an entirely different logical architecture to perform the same tasks as modern computers.

Other potentially fruitful avenues are quantum computing, optical computing, and DNA computing, all of which rely on very different architectures than conventional Von-Neumann computers. As I’m nearing the 1500 word mark I think I’ll end this essay here, but I do hope to return to these advanced computing topics at some point in the future 🙂


More on transistors:

The STEMpunk Project: Batteries

In The STEMpunk Project: Basic Electrical Components I wrote about resistors, capacitors, inductors, and diodes, but I had originally wanted to include batteries and transistors as well. As I did research for that post however it occurred to me that these latter two devices were very complex and would require their own discussion. In today’s post I cover a remarkable little invention familiar to everyone: batteries.

Battery Basics

The two fundamental components of a battery are electrodes and an electrolyte, which together make up one cell. The electrodes are made of different metals whose respective properties give rise to a difference in electrical potential energy which can be used to induce current flow. These electrodes are then immersed in an electrolyte, which can be made from a sulfuric acid chemical bath, a gel-like paste, or many other materials. When an external conductor is hooked up to each electrode current will flow from one of them (the ‘negative terminal’) to the other (the ‘positive terminal’).

Battery cells can be primary or secondary, and are distinguished by whether or not the chemical reactions happening in the cell cause one of the terminals to erode. The simplest primary cell consists of a zinc electrode as the negative terminal, a carbon electrode as the positive terminal, and sulfuric acid diluted with water as the electrolyte. As current flows zinc molecules combine with sulfuric acid to produce zinc sulfate and hydrogen gas, thus consuming the zinc electrode.

But even when not connected to a circuit impurities in the zinc electrode can cause small amounts of current to flow in the electrode and correspondingly slow rates of erosion to occur. This is called local action and is the reason why batteries can die even when not used for long periods of time. Of course there exist techniques for combating this, like coating the zinc electrode in mercury to pull out impurities and render them less reactive. None of these work flawlessly, but advances in battery manufacturing have allowed for the creation of long-storage batteries with a sealed electrolyte, released only when the battery is actually used, and of primary cell batteries that can be recharged.

A secondary cell works along the same chemical principles as a primary cell, but the electrodes and electrolyte are composed of materials that don’t dissolve when they react. In order to be classifiable as ‘rechargeable’ it must be possible to safely reverse the chemical reactions inside the cell by means of running a current through it in the reverse direction of how current normally flows out of it. Unlike the zinc-carbon voltaic cell discussed above, for example, in a nickel-cadmium battery the molecules formed during battery discharge are easily reverted to their original state during recharging.

Naturally it is difficult to design and build such a sophisticated electrochemical mechanism, which is why rechargeable batteries are more expensive.

Much more information on the chemistry of primary and secondary cells can be found in this Scientific American article.

Combining Batteries in Series or in Parallel

Like most other electrical components batteries can be hooked up in series, in parallel, or in series-parallel. To illustrate, imagine four batteries lined up in a row, with their positive terminals on the left and their negative terminals on the right. If wired in series, the negative terminal on the rightmost battery would be the negative terminal for the whole apparatus and the positive terminal on the leftmost battery would be the positive terminal for the whole apparatus. In between, the positive terminals of one battery are connected to the negative terminals of the next battery, causing the voltage of the individual batteries to be cumulative. This four-battery setup would generate six volts total (1.5V per battery multiplied by the number of batteries), and the total current of the circuit load (a light bulb, a radio, etc.) is non-cumulative and would flow through each battery.

If wired in parallel, the positive and negative terminals of the rightmost battery would connect to the same terminal on the next battery, and the terminals for the leftmost battery would connect to the external circuit. In this setup it is voltage which is non-cumulative and current which is cumulative.  By manipulating and combining these properties of batteries it is possible to supply power to a wide variety of circuit configurations.

Different Battery Types [1]

Nickel Cadmium: NiCd batteries are a mature technology and thus well-understood. They have a long life but relatively low energy density and are thus suited for applications like biomedical equipment, radios, and power tools. They do contain toxic materials and aren’t eco-friendly.

Nickel-Metal Hydride: NiMH batteries have a shorter life span and correspondingly higher energy density. Unlike their NiCd cousins NiMH batteries contain nothing toxic.

Lead Acid: Lead Acid batteries tend to be very heavy and so are most suitable for use in places where weight isn’t a factor, like hospital equipment, emergency lighting, and automobiles.

Absorbent Glass Mat: The AGM is a special kind of lead acid battery in which the sulfuric acid electrolyte is absorbed into a fine fiberglass mesh. This makes the battery spill proof and capable of being stored for very long periods of time. They are also vibration resistance and have a high power density, all of which combine to make them ideal for high-end motorcycles, NASCAR, and military vehicles.

Lithium Ion: Li-on is the fastest growing battery technology. Being high-energy and very lightweight makes them ideal for laptops and smartphones.

Lithium Ion Polymer: Li-on polymer batteries are very similar to plain Li-on batteries but ever smaller.

The Future of Batteries

Batteries have come a very long way since Ewald Von Kleist first stored static charge in a Leyden jar in 1744. Lithium Ion seems to be the hot topic of discussion, but there are efforts being made at building aluminum batteries, solid state batteries, and microbatteries, and some experts maintain that the exciting thing to watch out for is advances in battery manufacturing.

Hopefully before long we’ll have batteries which power smart clothing and extend the range of electric vehicles to thousands of miles.


[1] Most of this section is just a summary of the information found here.