0
0
Imagine yourself at the head of a successful, multinational corporation. You call all the shots, you bear all the responsibility. You are at the pinnacle of your career and you have earned your way up there through hard labour and good decisions. Clearly, your excellent ETH education has provided you with the basis for success, and the rest you were smart enough to pick up along the way. Clearly, right?
Since the days of Newton, human knowledge has doubled roughly every 17 years. This means that by the time you have climbed the corporate ladder, the knowledge needed to run this company will have roughly quadrupled compared to when you joined. This assumes you stay with the same company for your entire career. In effect, you will answer hardly
any of the questions you ask yourself as CEO with the knowledge you learned during your university degree. This does not even take into account all the training which by then has become obsolete. In short, why not let the pizza-delivery guy run the company for a fraction of your salary?
Against the Obsolescence of your Knowledge
First of all, you will have probably gained at least some inside knowledge during your time at the company that will have escaped even the most astute pizzaiolo. Ideally, you will have continued to learn well into your middle age and, at least when employed by any sort of technical company, will have kept up with most of the developments in your business area. Secondly, the particular knowledge you have acquired here at ETH (ideally) will have fared much better than the mean of knowledge overall. Note that while most of what was known at the time of Newton is now obsolete, Newton’s own laws remain as relevant as ever. By focusing on fundamentals, you will not only increase the half-life of your knowledge, but also develop the ability to pick up emerging fields once you encounter them. But do not be fooled: The simple and brutal truth of how knowledge accumulates means that most of the dominant technologies during your heyday do not yet exist today.
The Future comes with Change
This is no simple task. During your career you will inevitably encounter a large number of new ideas, most of which will appear absurd, some of which will come to substantially impact society. Given this, you have neither the luxury of ignoring all of these ideas outright nor of pursuing every such “crackpot” idea, lest you never get anything done at all. Fortunately, the future is not entirely arbitrary. It is both true that we can shape what the future will look like and that much of these developments will be determined by trends beyond our control. In hindsight, the past always seems obvious. Obviously, smartphones did replace landlines. Obviously, transistors did replace vacuum tubes. Obviously,
cars did replace horses. These developments seem obvious because we understand their underlying mechanism, we understand why one technology is superior to another and how this technology could become more relevant than its predecessor. At least some of this knowledge we have now also must have been available back when these technologies were first discovered. By learning what mechanisms have lead to technological change in the past and by understanding which fundamental constraints exist today, the loop invariants of history so to speak, we may have a chance at finding these few, life-altering technologies that are emerging at any point in time.
Possible? Likely? Desirable?
Equipped with our first principles, we can make some informed decisions today about what is to come. Richard Hamming (1915–1998), an important early computer engineer, dedicated 10% of his working time (Friday afternoon) to understanding the future of computing. As he describes in his 1996 book The Art of Doing Science and Engineering, when Hamming was asked to participate in a seminar on the topic of fibre optics in the 50s, he considered whether he should attend at all. In the end, Hamming found five reasons to attend the
seminar:
- Optical frequencies are much higher than electronic ones which equates to a higher bandwidth, i.e. more information per second.
- Alexander Graham Bell himself had already sent telephone signals through a light beam, so the technology is fundamentally feasible. Glass fibres are also relatively easy to manufacture.
- As the speaker at the seminar remarked “God loved sand, he made so much of it.” Glass is a cheap material compared to copper which was becoming scarce at the time.
- In Manhattan, the telephone wire ducts were already running out of space. Because fibre optics are smaller, you could replace the telephone wires with smaller cables to increase bandwidth without having to dig up the street.
- Optical fibres are difficult to tap and can withstand an atomic bomb much better than electronic wires, therefore, the US military is likely to put its support behind this technology.
When looking at the future, Hamming begs us to ask three questions: What is possible? What is likely? What is desirable? Optical fibres hit all three marks and Hamming (correctly) predicted that fibre optics would become the most common type of information-carrying wire. Hamming also immediately identified a fundamental challenge: How can glass fibres be joined together without loss of signal compared to the easy soldering required for copper wires? Over time, the sheer number of possible technical solutions proposed to “splice” two glass fibres together convinced Hamming that at least one solution would also work in the field, where it had to be done by technicians under adverse conditions and not by scientists in a controlled experiment.
During his career, Hamming had to repeatedly keep up with such developments as his lifetime not only coincided with the most rapid advancements in computer hardware, but also because he spent much of it at a place where a particularly large number of these innovations where made, the Bell laboratories. Hamming might have been uniquely good at this. His 1996 book even contains two, now prescient-seeming, chapters on AI. The amount of time he could (and should) dedicate to predicting the future was probably both the privilege and the curse of his position at Bell laboratories. Hamming actually repeatedly turned down promotions into managerial positions, which he later came to regret: “I was not doing my duty by the organization. That is one of my biggest failures.”
The Exponential Growth in Everything
The Exponential Growth in Everything Predicting the future is evidently impossible in many cases. Often, we can at least give a range for what might happen. Some things do however generalize, such as the famous Moore’s law, which states that the number of transistors on a computer chip doubles every two years. This might seem like a rare finding, but actually most new fields experience such exponential growth. The production of beer in Japan, for example, or the cost of offshore gas pipelines. Much of this can be understood by the much less famous Wright’s law which states that every time the production of some product has doubled, the cost of producing it has increased by some fixed percentage. Wright, who (confusingly) is unrelated to the Wright brothers, found this result while analysing the price of airplanes, where every doubling of the number of airplanes reduced the labour requirement by 10-15%. Essentially, the more you build, the more you learn about your product and the cheaper it is to produce it. Such exponential scaling laws are so ubiquitous that Sam Altmann, the founder of OpenAI, even published a blog post entitled “Moore’s Law for Everything” in 2021 outlining his view of a world in which AI accelerates most tasks. It is possible, somewhat likely and, as this column hopefully convinced you, desirable that ETH will in the future hold a lecture on “Deriving the Future from First Principles.” As a humble suggestion from our current state of things, a tentative syllabus could look something like this: To Hamming, the purpose of this exercise was ultimately to get a sense of direction while we are inevitably stumbling into the future. Without a vision, one is forced to simply react to what is happening. As he himself put it: “The drunken sailor who staggers to the left or right with n indepen- dent random steps will [on average] end up about √n steps from the origin. […] In a lifetime of many, many independent choices, small and large, a career with vision will get you a distance proportional to n.” Those who look ahead at least know where they are going.
