THE MOMENTOUS QUESTION
“Tell me, oh tell me, dearest Albert, have you any Railway Shares?”
For decades now, techno-optimists have been prattling about a “Singularity” that will immeasurably increase the growth rate in human productivity and thereby living standards. This Singularity may either come through Artificial Intelligence, as it outpaces the human brain and begins to make its own technological advances, or through other technologies that combine to produce a similar growth acceleration. It is a nice theory, an attractive counter to the innumerable environmental doom-sayers. However, we have already enjoyed such a Singularity. It occurred when Thomas Newcomen (1664-1729) produced his first commercial steam engine in 1712 and liberated Mankind forever from the limitations of human muscle. It is quite unlikely that we will ever benefit from another Singularity of equivalent potency.
Before Newcomen, all human societies had been limited by their inability to move great masses without recruiting the entire population to do so (as in the case of the Pyramids) or to move at faster speeds and for longer than a horse (if they had stirrups) could carry them. As a result, the vast majority of humanity toiled daily in activities of sheer drudgery, wearing out their bodies and condemning themselves to an arthritic and early old age. Inventions such as the lateen sail and the compass allowed great bodies of ocean to be crossed, but only very slowly as the winds took you, so that Commodore George Anson’s (1697-1762) circumnavigation of 1740-44 took 45 months to complete. Human strength was augmented by windmills and water power, but the Netherlands’ economically sluggish experience after 1700 suggests those power sources were inadequate; too intermittent in the case of wind and incapable of providing speed in the case of water power.
Newcomen’s invention changed all that. Initially, it was used for a limited industrial activity: pumping water out of mines, a process that could add much value to the mine but without steam required huge amounts of unremitting human and animal labor. James Brindley (1716-72) designed a water-powered mine pump that worked ingeniously, as did all of Brindley’s inventions, but it required a water source separate from and higher than the mine’s contents, so could be adopted in only a few cases. Brindley’s canals also pre-dated steam transportation, but movement on them was achingly slow.
Improvements were needed before Newcomen’s engine could power railways or steamships. The crucial development was not James Watt’s condenser but Jonathan Hornblower’s (1753-1815) 1781 compound engine, which could not be marketed until 1800 because of Watt’s over-broad patent protection. Hornblower’s engine, unlike Watt’s, used pressurized steam, whereas Watt’s was an “atmospheric” engine like Newcomen’s. Pressurized steam was needed to provide a high enough power-to-weight ratio to move transportation, whether trains or steamboats – the original 1807 Robert Fulton “Clermont” with a Watt engine worked fine on the slow-flowing Hudson but Watt-engined boats were useless upstream on the Mississippi, so the first Mississippi steamboat captains had to scrap their river fleets in New Orleans. The main problem that delayed the advent of pressurized steam engines was metallurgical technology, in particular cylinder boring tolerances, which were nothing like good enough in Newcomen’s time for a pressurized steam engine to be built.
Since the advances made by steam power were initially limited to mining and a few other sectors, it had little initial effect on productivity growth, although other advances such as Brindley’s canals made 1790’s Britain much richer than 1700’s. Only by 1815 were steam engines ubiquitous enough in Britain that their productivity effect was noticeable (by Jean-Baptiste Say, among others). Thus, the first industrialization surge in the early 1820s was a true wonder for both foreign visitors and British statesmen – the King’s Speech opening Parliament highlighted it every year. Other innovations followed; the Industrial Revolution spread first to Europe, then worldwide and new technological advances, notably the Bessemer steel process, the chemical industry and above all electric power compounded the effect of the initial surge. But the Singularity came in 1712; every other technological advance from then to today followed on Newcomen, which was the first true step change in power technology and the world we live in.
For a Singularity to occur, the technological advance that causes it must be as fundamental as Newcomen’s engine, enabling Mankind to do some new thing that had never been possible before, not merely to improve upon existing capabilities.
It is increasingly clear that Artificial Intelligence will not produce a Newcomen-type Singularity, at least not within the next couple of decades. The capabilities of ChatGPT and its competitors clearly increase our capabilities and will put millions of degree-credential drones out of a job –according to an FT story this week, the big consultants with staffing in the hundreds of thousands are already cutting back sharply on entry-level hiring. However, anyone who has used current AI services knows that they are very limited, prone to invent “facts” out of thin air and not capable of writing anything that sensible people would want to read.
AI’s capabilities will improve, of course they will. But we are still decades away from the “Artificial General Intelligence” breakthrough that might make AI a true competitor to intelligent human beings. That AGI will be able to write code very effectively, putting more modestly able drones out of work, but it will not make intellectual breakthroughs any more effectively than intelligent human beings can. Hence, there is no coming moment, at least within the next 20-30 years, at which AI might create a Singularity acceleration of human productivity growth.
There are also two countervailing factors relating to AI and current technology in general. One is its appalling cost; trillions of dollars are being spent on data centers and massive upgrades to the world’s electricity grid, to power the new systems. In this respect, AI is not the steam engine but the railway, an innovation of undoubted merit but so expensive it generally proved a very poor investment. In the cartoon attached, published as the 1845 “Railway Mania” bubble was bursting, Queen Victoria is worried that Prince Albert, with his enthusiasm for technological novelty, has sunk the Royal fortune in duff railway speculations like those of “Hudson, the Railway King.” George Hudson (1800-71) was the Elon Musk of his day, projecting new railways all over the North and Midlands of England; needless to say, his empire came to a miserable end, although some of his railways still exist. For the rest of the 19th and early 20th century, U.S. railway bonds were a means by which good British money was poured into American scams; the combination of massive overinvestment, considerable fraud and frequent credit crises made them a very poor buy.
Regardless of any bubbles bursting, there is also the possibility that modern technology is, by taking the effort out of education, degrading human capabilities so far that technological progress must inevitably go into reverse. The recent report that a majority of college entrants at University of California, San Diego cannot handle middle school math is sufficient indication that if AI goes far enough, the magnificent machines may be talking only to each other, while the remnants of humanity grunt meaninglessly and fight each other for the remaining leftovers of package foods.
Outside AI, there are more plausible contenders for a future Singularity. The current expansion of Waymo self-drive automobiles will in the long run reduce accidents and allow mentally and physically disabled people (such as myself, marginally, with slowed reactions and fading eyesight) to transport themselves without the difficulties and dangers of public transportation. That doesn’t qualify as a Singularity – it is a development of existing capabilities — but it will expand both productivity and life experiences, while reducing the incidence of automobile fatalities. Its one weakness may be an inability to cope with Hudson Valley winters, as distinct from the endless boring California sunshine marred by the occasional earthquake.
In the long term, the most likely candidate for a Singularity will be the genetic re-engineering of Humanity itself. I first wrote on this topic almost 25 years ago (part 1 and part 2); the appallingly slow progress of genetic engineering, held back as it has been in the West by fanatical anti-eugenic sentiment, suggests that we are probably still half a century away from the point at which we can engineer truly superior human beings, with or without brain-computer interface additions that will enhance their capability even further.
Nevertheless, that ability, if sufficiently enhanced to allow mass production and the replacement of college students who can’t manage elementary algebra, would indeed represent a Singularity worthy of Newcomen. A human species of which most members have genuinely superior intellects would provide a hugely impressive acceleration in productivity growth and human achievement. Olaf Stapledon (1886-1950) in his 1930 “Last and First Men” postulated a future in which humanity would bioengineer itself into eighteen future species over the next 2 billion years, producing unimaginable civilizational advances before the heat death of the Sun made even Neptune uninhabitable. If within the next century we can produce a replica of Stapledon’s Fifth Men, we will achieve a Singularity truly worthy of the name.
Meanwhile, we should work to ensure that destructive forces of decay are fought off, so that we can continue to enjoy the long-term fruits of the Augustan Age Singularity produced by Thomas Newcomen.
-0-
(The Bear’s Lair is a weekly column that is intended to appear each Monday, an appropriately gloomy day of the week. Its rationale is that the proportion of “sell” recommendations put out by Wall Street houses remains far below that of “buy” recommendations. Accordingly, investors have an excess of positive information and very little negative information. The column thus takes the ursine view of life and the market, in the hope that it may be usefully different from what investors see elsewhere.)