Exponential Innovation: The 4 Pillars
All this new data and all these new techniques are no good to us, unless we find ways of building the computers which will be able to teach themselves. Luckily, material sciences have advanced as swiftly, or more swiftly, than computer scientists could have accounted for 50 years ago.
As we’ve advanced in information science, we have also taken leaps in material sciences. Today, plastics and silicone seem set to be replaced by new, more exotic materials like graphenes, and fullerenes, among others. Today, industrial applications for graphenes are being heavily invested in all over the world. 200 times stronger than steel, transparent, and capable of flexing by up to 25% without any wear, acting as a semi-conductor, and being grown into novel shapes, graphenes may be to the industry of the 21st century what steel was to the 19th, or plastic to the 20th.
While this section could be a catalogue of breakthroughs on every front, we will stick to the importance of material sciences to the development of computing power. And in this realm, much has been achieved that was not seen as possible only decades ago.
IBM, among others, are developing chips now on the 7 nanometer node scale- not much larger than a strand of human DNA. Essentially, the next generation of microchips will be as information dense as human DNA itself. With advances in graphene and other exotic materials, we may see semi-conductor lengths cut to as low as .2 nanometers- over 10 times smaller still- or small enough for microchips to insert themselves directly into cells, and interact with DNA.
On other fronts, quantum computing has emerged from mere theory, into its first practical demonstrations in recent years. While it’s still early to speculate as to quantum computing’s impact on computer science in the short term, many computer scientists see it as the answer to many of traditional computing’s material and theoretical limitations.
It seems likely now that quantum computing will continue to develop in parallel to traditional computing, slowly assuming responsibility for many of the tasks that traditional computers are ill-equipped to perform.
For context, here is a great explanation of how quantum computing would work:
So, for a concrete example, suppose you wanted to model the processes inside a dying star or a black hole. Transistor architecture is useful in modeling newtonian physics, in which assumptions about gravity are simple, and physics are not affected by high energy states. This is why Nasa still uses Newtonian physics for things as complicated as interplanetary missions, or even interstellar probes like Voyager. A transistor based computer can accurately model what will happen, in broad terms, when two planets pass each other, or when an asteroid hits a planet, or even exactly how to get a 2-ton probe to pass within a few hundred kilometers of Pluto.
However, because the inside of a black hole or a dying star is affected by the rules of high energy states, in which quantum fluctuations become as powerful as gravity and the electromagnetic force, the number and speed of calculations necessary to model what will happen becomes overwhelming for a traditional computer. Indeed, a traditional computer might be entirely incapable of modeling the outcomes of certain events, because no algorithms based on semiconductor architecture could ever be fast and accurate enough, or simple enough to run in any human time scale.
Computers are binary, but the physical universe is not. Using binary computing to model the universe, or many of the processes that happen inside it, is like using a pencil and paper to recreate a 3D image. It can be faked, and it can be sufficient for some purposes, but it will never be completely realistic.
Obviously these new material sciences, combined with our growing knowledge of how to build and “train” neural networks, are now set to deliver powerful changes to the capabilities of computers in the very near future.
And here’s why it matters: our ability to understand higher energy states (like the collision of atoms or the process of nuclear fusion), will drastically affect our ability to manipulate physical reality, including finding new ways to produce energy and materials, among much else.
Which brings us to our final category:
In 1964, Soviet astronomer Nikolai Kardashev was considering a major problem in cosmology. If the universe is populated by other intelligent beings, how would humanity recognize them?
The result of his thought experiments, which came to be called the Kardashev Scale, postulates that interstellar civilizations can be recognized according to the patterns of their energy use. As a technological society advances, he observed, it typically harnesses ever increasing amounts of energy: from resource gathering to farming, to coal and oil, to solar and nuclear power, humanity has rapidly increased its control over energy resources.
Kardashev speculated that if we wished to identify an alien civilization, we should look for the signs of three different stages of evolution in energy use, which became Kardashev Civilization Types: I, II, and III (a highly speculative type IV might not be observable using current methods).
A type one civilization is able to harness most of the energy that it receives from solar insolation, and from the core of its own planet. A type two civilization would be capable of harnessing most of the energy of its adjacent star, using, for example, a Dyson Sphere, or a cloud of solar arrays. A type three civilization would be technically capable of controlling most of the energy in its galaxy, applying the traits of a type two civilization on a cosmic scale, exploiting cosmic rays emanating from supermassive black holes at the centers of galaxies.
While the Kardashev Scale might seem like science fiction, it does provide a framework for actual observation of the known universe. Using these predictions, it’s possible to actively look for stars that appear to be surrounded by Dyson Swarms, or clouds of energy collecting satellites that redirect large amounts of energy to a planet or other location. We may even have already uncovered a possible extraterrestrial civilization of some kind.
In the future, it may be possible to identify alien civilizations, or alien superstructures by noting their absorption of cosmic rays at the centers of galaxies. The evidence for such advanced civilizations may in fact already be before us- we simply lack the tools to measure it.
It’s important to note that the Kardashev scale doesn’t require, for example, that a type one civilization should capture all sunlight as it reaches a planet. The scale deals with the amounts of energy being consumed- some of which may come from sources more energy dense than sunlight. Astronomer Carl Sagan speculated in the 1970s that humanity was already at the equivalent of 0.7 on the Kardashev scale, using about 2 terawatts of energy per day. That number has increased upward, though not as fast as it had in the century before Sagan made his calculations. It has been widely speculated that we will reach Kardashev’s level one within the next century or two.
In fact, we have already discovered most of the basic processes and methods needed to reach this level of energy manipulation. Humans first fused atoms in 1944, as part of the Manhattan Project to build the first atom bomb. In the 1950s, we began to fuse hydrogen to create hydrogen bombs, and controlled fusion reactions now occur in fusion chambers around the world. The problem in applying this technology on a planetwide scale has been energy efficiency: the amounts of energy needed to create controlled fusion reactions have so far been too high to deliver a net energy gain- meaning that it takes more energy to create controlled fusion than we can actually get from fusion. So far.
But advances in our understanding of quantum physics and material sciences mean that this barrier may soon be lifted. If it is, that will mean the eventual replacement of all extractive energy collection (such as coal, gas, and even nuclear fission), with clean, abundant nuclear fusion. And meanwhile, battery and solar technology are quietly outstripping previous expectations of increased efficiency, and are set to deliver massive changes to our energy infrastructure.
In material sciences, computing, energy, and artificial intelligence, we now stand at a crossroads. How soon will each of these areas of science yield material results that will dwarf those of the past century? Taking history as our guide, human civilization moved from human labor to animal labor over a period of 100,000 years. We moved from animal labor to mechanical labor (think water wheels) in a period of some 3-5,000 years. Shifts from mechanical labor to steam, and from steam to nuclear fission took less than two centuries.
These are trends that are hard to assess, but impossible to ignore. And as our ancestors of the 18th century could not have imagined life in the nuclear age, it’s entirely possible that we too lack the imagination necessary to picture our own lives 20 years from now.