Here are three things I'm thinking about:
- The fractal nature of violence
- Why there will be no "next Microsoft" of AI
- Galloping Gertie and the recent North American heat wave
Let's go.
01. The Worse Devils of Our Nature
More than two decades ago Steven Pinker wrote a book (The Better Angels of Our Nature) about how we live in the most peaceful period in human history. This argument pleased many people, and, from his data about the "long peace", it wasn't wrong. Violence, wars, and related strife had generally declined.
Critics, however, rightly pointed out that he was conflating various things. While major conflicts were less frequent, they were even more severe. The two great wars of the 20th century, for example, killed more people than all conflicts of the prior two thousand years combined.
I'll quote Pinker-critic Nassim Taleb and a co-author here in a rebuttal working paper:
We show that the inter-arrival times among major conflicts are extremely long, and consistent with a homogenous Poisson process
To translate, major conflicts are as large as ever, even if they don't come as often, but they do come. The long peace may be real, but it could destroyed and made irrelevant by even one spasm of violence, given the intensity and technology of modern conflict.
And this brings us to today. As NYT columnist David Wallace-Wells points out in a recent piece:
In 2011, when Pinker published 'Better Angels,' there were nearly 40,000 deaths from warfare worldwide. In 2022, the number was above 238,000 — a nearly sixfold increase. It had nearly doubled in a single year.
While that hasn't completely reversed Pinker's peace, it shows how spasms of violence in modern war can reverse periods of seeming calm. Further, and this is more insidious, you could equally argue that the long peace created the tensions that made the latest explosion of deaths so horrible, like slip-strike faults of stored and angry human energy. The human tendency to confuse temporarily calm seas with a regime change remains as vexing as ever.
02. There is No "Next Microsoft" of AI, Not Even Microsoft
I regularly speak with some of the largest investors in the world about the shifting economic and technology ground on which we stand, as waves of AI make it all seem less permanent than it once did. One of the questions I get a lot—and I got it this week in a private meeting with some executives at a major U.S. bank—is, "Who is the next Microsoft or Google or whatever?" They want to know, in short, who wins?
My answer, in general, is that they are thinking about it all wrong. In its current form, AI is job replacement technology, something that will remove humans from the workplace, not augment them. Economist Daron Acemoglu at MIT, among others, has written extensively about this, about augmenting vs replacing technologies, and he is worth reading on the topic.
We can turn to recent headlines to get a sense. This week there was a somewhat overheated story about a European company that had replaced something like 700 call center workers with bots, algorithm technology entities that combined various technologies to provide the same services at a fraction of the cost. While there is still some confusion about exactly when the jobs were lost and the role of bots therein, that didn't stop public markets from slamming call center stocks the next day, with some off 25% or more. Who needs call centers in a world of AI?
A naive way of thinking about this, unhappy human consequences aside, is that maybe bot companies will do well in the future. After all, you can see how bots will quickly replace humans in most call centers, customer service jobs, booking centers, and so on. Go long bots, right?
Not so quickly. These are commodity technologies already, even if they must be made specific via being trained on a company's internal data. There are dozens of bot companies, all clamoring to replace your expensive humans cheaply, and the price keeps falling.
Ah-ha, you think. Maybe not bots, but all these companies rely on large language models for their interactive versatility, like OpenAI's GPT-4, and so won't the owners of those LLMs be the winners (and perhaps even Microsoft, given its OpenAI holdings)?
Again, not so fast. While OpenAI is apparently generating nearly two billion dollars in revenues, its costs are immense and growing, given training needs, GPU purchases, data center growth, and soaring energy costs. Further, it's not clear that the market will tip to a single LLM provider anyway in a natural AI monopoly. While no LLM provider is on par with GPT-4, they are getting closer, and open-source technologies offer the advantage of being free, even if more complex to manage. It's as if Microsoft launched Windows, immediately had credible open-source competitors, and needed billions of dollars in ongoing costs to keep Windows working, rather than just relying on your home PC and its Intel Inside(tm).
What I generally tell people is that the winners—and this is the wrong and sociopathic word—will be dominated by companies using these technologies to slash costs. We will see firms in people-heavy but rule-bound fields, like insurance, airline booking, call centers, and so on, discover they can run with a fraction of their current employees. Shareholders will reward this as margins rise, competitors and boards will take notice, and every other company will follow, quickly shedding costs (read: people) and driving up margins.
Consumers will benefit from this, of course, in the same way they benefited from the rise of Costco and Walmart. Efficiencies, consolidation, and lower prices have diffuse and immense benefits, even if they have concentrated costs. The next Microsoft is a diffuse and fast-moving anti-Microsoft of cost-cutting.
03. Galloping Gertie and the recent North American Heat Wave
It is hard to put into words the immense scale of the heat records this week in North America. More than 1,000 records have been set, both daily and monthly, with many stations beating prior records by more than 10 C (close to 20 F). And it has been followed by a cold plunge, which fill be followed by more heat. Even the extremes are extreme.
Most people will have heard of Galloping Gertie, the Washington bridge that found a resonant frequency in a steady wind, oscillated in larger waves, and then fell to pieces.
Humbled, we have learned a great deal about engineered design, resonant frequencies, and systems since then—at least in constrained systems like bridges. We still know very little about the topic in world systems, like climate, as we are constantly reminded. The hard-earned humility in engineered systems is absent in natural ones, having become inured to the costs, normalizing them as they come at us.
For example, recent heatwaves in southern Europe have been a puzzle, larger than expected, but can be explained in part, a new paper suggests, by understanding the coupling between current climate and prior year Arctic melting, spilling freshwater into the northern Atlantic. Systems within systems.
While natural systems are not bridges, coupling and oscillations exist in similar fashions. Systems repeatedly pushed to extremes may not fall into a river obligingly and on camera, like Galloping Gertie, but they do "sample" new equilibria, and, if pushed there often enough, can restabilize there, impossible to push back.
There is a growing body of research on this topic. A recent paper shows that a slow reaction to initial, less severe changes is its own warning. The system, lacking the ability to adapt and respond, becomes more fragile, not less, eventually leading to a catastrophic transition. We are slowly incorporating these new approaches to precursors to critical changes in models, and the results are impressive, even if deeply disturbing, given the state of the world.
Back again this weekend with more things to think about.
