More on this book
Community
Kindle Notes & Highlights
by
Jeff Booth
Read between
June 4 - June 20, 2021
Remember, the world of 2018 has approximately $250 trillion in debt to run an $80 trillion world economy. That debt in itself is a massive drag on future growth because of interest payments on it.
At approximately 1,239 kilometres per hour, the speed of a jet surpasses the speed of sound, and because the sound waves can’t push in front of it any longer, it punches through the sound barrier, causing a loud boom—and changing the rules. After that, from the ground the plane appears to outrun its sound. The metaphor of a sonic boom is akin to what we will see at some point with debt creation when the rules will change instantly.
In an August 2018 working paper titled “Monetary Policy with Negative Interest Rates” by the International Monetary Fund, the authors discuss how central banks can design and operate a system where interest rates could be far more negative than they are today. As interest rates drop too far below zero, it makes sense for deposit holders to move their money out of banks and into cash, resulting in a limit to how low central banks can reduce interest rates, as people and businesses will hoard cash. The proposed solution sees a mechanism where negative exchange rates are applied to both
...more
It is important to understand that all of that debt did have a very positive effect on our economies, jobs, and lives. As we talked about in chapter 1 on how the economy works, when asset prices rise, people feel richer and spend more, which in turn creates more jobs because their spending drives the economy. Growth would not have been nearly the same without it and, therefore, many of the benefits to society would not have accumulated as quickly without it. The number of people in the world living in extreme poverty—earning under $1.90 per day—has fallen significantly, from more than 50
...more
Easy credit resulted in a significant rise in prices across asset classes—home prices, oil prices, stock prices, to name a few—creating real wealth for the holders of assets and spurring even more growth, with countless jobs being added to growth sectors of the economy that have been aided by easy credit and low rates. Venture capital and technology companies themselves have benefitted greatly from this cheap source of capital in raising giant venture rounds, meaning that some of the technology progress and feedback loops themselves were quite likely accelerated beyond what would have
...more
What if you could buy a permanent source of electricity for your house for $2 million? What if the price dropped from $2 million to under $100? And you only have to pay that once and, forever after, all your electricity would be free—like the air you breathe?
At 9 percent of GDP, it makes up a lot of jobs around the world. In the US alone, 3.6 million direct jobs are in the traditional energy industries, including production, transmission, and storage, with another approximately two million jobs in energy efficiency.30 But the role of energy in our economies is much higher than that. We still need to factor in how much of the world’s military complex is built mainly to ensure continual access to energy at reasonable prices. Low-price, abundant energy is a critical component of any nation’s competitiveness, since it is used in every industry. Beyond
...more
Worldwide energy use has gone up almost thirteen times, from 12,100 terawatt hours per year in 1900 to 153,596 terawatt hours in 2017. The biggest drivers were cheap and abundant sources of energy: coal, crude oil, and natural gas. From 1900 to 2017, coal as a primary source of energy grew from 5,728 to 43,397 terawatt hours, crude oil grew from 181 to 53,752 petawatt hours (a petawatt hour is 1,000 terawatt hours), and natural gas grew from 64 to 36,704 terawatt hours.32
Remember, that oil as an “energy source” has only been stored. The energy in oil initially came through plants that absorbed their own energy from the sun through photosynthesis, and through animals that absorbed their energy by feeding on the plants. All of that energy originally came from the sun. That oil needs to be pumped from the ground (requiring energy) and transported (requiring energy) to an oil refinery, where it undergoes a conversion (requiring energy) to gasoline. That gasoline then needs to be transported (requiring energy) to a regional gas station where you fill your car. Even
...more
Through this example, you can see a fuller cost of the energy to move your car. It includes a staggering amount of inefficiency... and drives countless jobs.
By getting our energy directly from the sun instead of a circuitous route of digging things up that originally got their energy from the sun and transforming and re-transforming them, we remove an entire supply chain of inefficiency and cost. By converting energy from the sun directly, we can get an almost-free lunch... without the corresponding damage to our ecosystem.
In less than two hours, more energy from the sun hits the Earth than the yearly worldwide consumption of energy.33 It’s just a question of putting it to use.
Another sixty years passed before Bell Labs invented the modern solar cell in 1954. Made from silicon, this breakthrough cell had 6 percent efficiency in converting sunlight to energy, which was a huge improvement from previous technologies. It allowed solar to be used for about $256 per watt. Even with that huge leap forward, $256 per watt was far more expensive than other sources of energy at the time, so it is easy to see why a transition from lower-cost sources of energy to solar power didn’t take place. As technology has improved, though, that rate has dropped precipitously, from $256 per
...more
Swanson’s law (named after Richard Swanson, founder of SunPower) states that the price of solar tends to drop 20 percent for every doubling of shipped volume.
According to a research report by financial think tank Carbon Tracker in November 2018, 42 percent of the world’s coal plants are already running at a loss, and it costs 35 percent more to keep existing coal plants running than to build new renewable energy generators.34 If these numbers are true, due to economic realities and competition, the days of coal as a source of energy are numbered.
How much area would be required to build solar farms to generate all of our needed energy? Without taking into consideration any improvement in technology, according to the renewable energy advocacy group Land Art Generator, the surface area required is 496,805 square kilometres.38 That may sound like a lot of land, but the land leased to the oil and gas industry in the United States alone covers 104,177 square kilometres.39 If you used that land for solar power, you could provide more than one-fifth of the world’s entire energy needs.
In 2000, solar only accounted for 1.15 terawatt hours of electricity; by 2017, that had grown to 443 terawatt hours. Solar is still a very long way away from producing 100 percent of the 153,596 terawatt hours energy needed today, but with lower price on its side, and even lower pricing on the horizon, that gap will close quickly.
One such solution—a flywheel—converts electricity to kinetic energy for storage and then converts kinetic energy back to electricity when needed. There’s a new race to control key pieces of technology to enable a shift to abundant renewable power. For example, Temporal Power of Mississauga, Ontario, Canada, a leading innovator in the space, was recently acquired by the Chinese flywheel technology company BC New Energy.
Clean energy will stop adding more carbon dioxide to our environment, of course. Carbon dioxide emissions from the burning of fossil fuels have taken levels of greenhouse gases in our atmosphere to levels never before seen in human history: 415 parts per million. Ice core samples confirm it is a concentration not seen for more than 800,000 years.
Like most predictions of technology, Minsky’s proved to be early. The doubling-up examples from chapter 4 show why: it is as easy to overestimate the impact of exponential growth in the early doubles, as it is to underestimate it in the later ones.
Our intelligence—our ability to master the world around us—is actually derived from other people: their thoughts, inventions, and science, which we have in turn continued to build upon. Without that information and knowledge, most of our limited time would go into providing basic human needs. Throughout our history, it is our collective growth of knowledge that is the real driver of what we deem “intelligence.”
Error correction is the basis of all intelligence.
As Karl Popper (1902–1994), one of the great twentieth-century philosophers of science, said, “All of our knowledge grows only through the correcting of our mistakes.”43 Some of the biggest revolutions in science actually come from small refinements of existing theories.
The printing press recorded and stored information and, with it, delivered the ability to correct errors to a much wider audience. This gave rise to the Age of Enlightenment—also known as the Age of Reason.
In a world that seems more divisive with each passing day, it is worth remembering that intellectual debate to find better answers is the goal of science and the very thing that has allowed great leaps forward for mankind. To quote Karl Popper again, “True ignorance is not the absence of knowledge, it’s the refusal to acquire it.”46
Because of the combined ability to both make a permanent record of our knowledge and have our ideas continually questioned and built upon, humanity’s ability to understand our world has seemed to change overnight on the evolutionary scale. Remember, our brains have been almost the same for around 300,000 years, but we’ve had the printing press for just under 600 years.
In 1950, he published a paper titled “Computing Machinery and Intelligence” where he proposed a test called the imitation game, now commonly referred to as the Turing test. In the test, a human evaluator would have a conversation with two others, one being a machine and one a human, and the test would be passed when the human evaluator could not distinguish between the human and machine—in short, when humans can’t distinguish artificial from real intelligence.
A message that doesn’t reduce the possibilities for the receiver transmits zero bits of information. Because of Shannon’s information theory, for the first time, information became quantifiable.
What is a Bayesian probability machine? A computer that works on Bayes’s theorem, named after Thomas Bayes (1702–1761). Bayes’s theorem assesses the probability of an event based on prior information.
Through this Bayesian method, you could imagine learning any problem as long as you had a starting probability and enough cycles to update the probabilities. Similarly, a computer could solve any problem if it had a prior probability and enough data and compute power to continually adjust that probability—in other words, error correction and refinement of hypothesis through iterations. Intelligence.
Just one year later, in 2017, Google launched a newer version called AlphaGo Zero that beat AlphaGo 100 games to zero.
Understanding only the rules of the game, AlphaGo Zero became its own teacher, playing itself millions of times and through deep reinforcement learning getting stronger with each game. No longer constrained by human knowledge, it took only three days of the computer playing itself to best previous AlphaGo versions developed by top researchers and it continued to improve from there. It mastered the masters, then mastered itself, and kept on going.
Unlike the things we know well where the neural connections are strengthened, the brain has to rewire itself through repetition and error correcting. And that becomes the trap—when new thinking is needed, it is very easy for us to remain entrenched.
Patterns can be seen without knowing that you’re seeing them. Because those patterns are now committed to your unconscious, conscious energy is freed up for more important moves or decisions, as any elite athlete and many others will tell you about a state of flow.
We don’t actually hear or perceive what others “say” in the way they mean it; rather, we “hear” them through our own filters of previous information encoding. Computers are not bound by that thinking. Computers do not attach emotion to storage of information in the way that humans do. They do not have a bias problem (unless programmed in by a human). They recall data exactly as received.
It is not therefore a difficult leap for the imagination that—with enough data, compute power, and storage—almost any problem that could be solved by a human could be better solved by a computer.
Don’t forget how rapidly this is accelerating now. The first Homo sapiens emerged over 300,000 years ago. The alphabet, which enabled writing, was invented approximately 3,000 years ago. The printing press was invented almost 600 years ago. The first mechanical computer was envisioned (not built) 170 years ago. The first ideas around artificial intelligence were developed seventy years ago. The first AI to beat a grandmaster in chess was developed twenty-three years ago. The first AI to beat Jeopardy! was eight years ago. The first to beat a grandmaster at Go was three years ago. The growth of
...more
Do you have an Apple Watch? If you do, you have an example of the future of healthcare on your wrist. Apple’s watch already collects heart rate information, ECG information, exercise patterns, and sleep patterns. By detecting heart rate abnormalities or elevation in heart rates, it has already saved many lives.
It will also, again, be bad news for jobs. Why? Look at how many jobs come from waste in the system caused by information asymmetry. Think of a case where you go to multiple doctors—family doctor, radiologist, gastroenterologist, other specialists—who each have their own staff and each only have a part of your information. As you are treated, repetitive trips, further specialization, and often misdiagnoses are all part of the overall health budget. When artificial intelligence reduces that waste and increases the benefits to society, as a by-product of removing the waste in the system, it
...more
For instance, as reported in a May 2019 Nature Medicine article, researchers created a 3D volumetric deep learning model to screen for lung cancer.55 When comparing a single image, the deep learning model outperformed six radiology experts, with an 11 percent reduction in false positives and a 5 percent reduction in false negatives.
All of today’s top companies globally are data companies that enjoy network effects, capturing more data as they grow, which in turn creates better systems. They are creating data monopolies where vast data sets are being combined to produce impressive results. The more data, and the more data velocity, the better the artificial intelligence becomes and the better outcomes from it. Top researchers in AI are attracted to companies that have these data sets because of the faster rate of experimenting. As the holders of our data amass giant data sets, they ultimately control the world.
That is the real race today, a race that is geopolitical in scope and scale. Look no further than Vladimir Putin’s comment about artificial intelligence in 2017: “Whoever becomes the leader in this sphere will become the ruler of the world.”56 It is not only companies but countries that are investing heavily to win this race.
That race for AI superiority may be behind some recent high-profile events between governments. The Huawei case, where the United States government charged China’s Huawei and its officers of intellectual property theft as well as sanctions violations, offers clues to that race. It is no secret that Huawei has ambitions to build an infrastructure backbone to capture data...
This highlight has been truncated due to consecutive passage length restrictions.
other good deeds. The idea behind the original plan might not seem bad at a high level; it was targeting dishonesty in government affairs and encouraging commercial and social integrity. But it is easy to see where such a system could be open to error and manipulation. The rules of the systems are not universal, and the systems are not interconnected. In some regions, listening to music too loudly deducts points; in others, jaywalking or playing video games does. It would be hard to imagine any system in China that rewarded opposition in any form to the Communist Party. The system is already
...more
Underlying our intelligence as a species always has been, and still is, fundamentally a collective growth of information.
Error correcting on our information gave rise to a world of science and discovery that led to many of the advances we take for granted today,
It is logical that the flood of information and knowledge is now transferred to computers because of their ability to “see” and correct patterns in massive data sets better and faster than we can.
But whether we embrace it or not, the genie will not be put back in the bottle. These things are true: 1) error correction is at the heart of all of our “intelligence”; 2) information is growing at an exponential pace; 3) that information is being transferred to computers that can gain knowledge and correct errors faster than human brains can; and 4) Every one of our jobs is a function of our intelligence.
The lack of an ability to form bonds is associated with some of the most severe personality disorders, such as paranoia and schizophrenia, and much research has been done on the need for humans to belong.
“The need to belong is a powerful, fundamental, and extremely pervasive motivation,” they write, and they discuss the possibility that “much of what human beings do is done in the service of belongingness.”59
