Jason, is your view that the industrial revolution is not uniquely significant? Meaning, it just happens to be at the bend in the exponential curve when we look backwards from this point, but since it’s been an exponential all along (or super exponential) then plenty of things before it and after it contributed just as much to the increasing rate of progress?
It was uniquely significant in that it was a fundamental shift in the basis for the economy, enabling a new regime of faster growth. I think the only other equal-magnitude shift was the transition from nomadic hunter-gatherer societies to settled agricultural ones. It just wasn't the beginning or the end of acceleration, and it didn't alter the super-exponential trend.
I also wouldn't say it “just happens to be at the bend” … the curve bends at that point *because* of the Industrial Revolution!
Maybe this framing will help: Previously I had a model that was like: the world was plodding along, not really making much/any progress, and then there was this exogenous shock which led to the Industrial Revolution, and that started real progress.
Now my model is more like: Progress was building up slowly for a long time. Once it built up to a certain point, the prerequisites were in place for an Industrial Revolution, and that was the next thing that made sense to do. Then, that allowed progress to accelerate even more.
Maybe even more concisely: the IR is the *form* that acceleration took at that point in time. But the IR isn't the most fundamental explanation for acceleration. There's a deeper process going on, and in 10k BC that process takes the form of agriculture, and in the 1400s it takes the form of navigation and the printing press, and in the 1700s it takes the form of steam engines and textile machinery, and in 2025 it takes the form of AI.
"The best minds of my generation are thinking about how to make people click ads. That sucks." - Jeff Hammerbacher
Given that Google and Meta are almost entirely based on ad revenue and they are the most sunk into AI investment, I am experiencing some bitter cynicism that AI will be used to manipulate the public into improving the effectiveness of ads before it is used to solve genuine human problems. Shareholder pressure for these big tech companies and VCs will decide where the capital goes.
It's crazy to think how the entire AI revolution was based on the invention of GPU hardware used to make video games a little prettier in the 70's. What else are we missing out on today because the niche inventions that precipitate them are slightly unprofitable? It's hard to talk about unknown unknowns, but I think this is perhaps the most severely underexamined area of human inquiry, despite entire ecosystems of entrepreneurs and industrial engineers. ARPANET started out as a niche, too. What is our best course of action for cultivating these history-defining drivers of progress when they are still in their infancy?
Progress refers to humanity in general. As you said, "tribe(s) died out instead of spreading." Isn't humanity now one big "tribe" and therefore subject to "die out"?
This was a great chapter Jason! The point you made at the end stuck out to me: if technological progress in the far past was similar to biological evolution in that inventions stuck around if a tribe was able to pass down the knowledge to their descendants, then it makes sense that progress was so slow. Would it therefore make sense to split the talk of technological progress in two eras: before and after writing / more durable ways to spread inventions? Instead of one super-exponential curve, it would be a piecewise function.
I am unclear whether progress is a smooth super-exponential curve, or whether it is piecewise exponential, or some other piecewise function. In any case, it approximates a smooth curve if you zoom out enough. Robin Hanson has an interesting paper trying to approximate long-run growth as a series of exponential modes with CES transitions between them
Jason, is your view that the industrial revolution is not uniquely significant? Meaning, it just happens to be at the bend in the exponential curve when we look backwards from this point, but since it’s been an exponential all along (or super exponential) then plenty of things before it and after it contributed just as much to the increasing rate of progress?
It was uniquely significant in that it was a fundamental shift in the basis for the economy, enabling a new regime of faster growth. I think the only other equal-magnitude shift was the transition from nomadic hunter-gatherer societies to settled agricultural ones. It just wasn't the beginning or the end of acceleration, and it didn't alter the super-exponential trend.
I also wouldn't say it “just happens to be at the bend” … the curve bends at that point *because* of the Industrial Revolution!
Maybe this framing will help: Previously I had a model that was like: the world was plodding along, not really making much/any progress, and then there was this exogenous shock which led to the Industrial Revolution, and that started real progress.
Now my model is more like: Progress was building up slowly for a long time. Once it built up to a certain point, the prerequisites were in place for an Industrial Revolution, and that was the next thing that made sense to do. Then, that allowed progress to accelerate even more.
Related: https://slatestarcodex.com/2019/03/13/does-reality-drive-straight-lines-on-graphs-or-do-straight-lines-on-graphs-drive-reality/
Maybe even more concisely: the IR is the *form* that acceleration took at that point in time. But the IR isn't the most fundamental explanation for acceleration. There's a deeper process going on, and in 10k BC that process takes the form of agriculture, and in the 1400s it takes the form of navigation and the printing press, and in the 1700s it takes the form of steam engines and textile machinery, and in 2025 it takes the form of AI.
"The best minds of my generation are thinking about how to make people click ads. That sucks." - Jeff Hammerbacher
Given that Google and Meta are almost entirely based on ad revenue and they are the most sunk into AI investment, I am experiencing some bitter cynicism that AI will be used to manipulate the public into improving the effectiveness of ads before it is used to solve genuine human problems. Shareholder pressure for these big tech companies and VCs will decide where the capital goes.
It's crazy to think how the entire AI revolution was based on the invention of GPU hardware used to make video games a little prettier in the 70's. What else are we missing out on today because the niche inventions that precipitate them are slightly unprofitable? It's hard to talk about unknown unknowns, but I think this is perhaps the most severely underexamined area of human inquiry, despite entire ecosystems of entrepreneurs and industrial engineers. ARPANET started out as a niche, too. What is our best course of action for cultivating these history-defining drivers of progress when they are still in their infancy?
Progress refers to humanity in general. As you said, "tribe(s) died out instead of spreading." Isn't humanity now one big "tribe" and therefore subject to "die out"?
This was a great chapter Jason! The point you made at the end stuck out to me: if technological progress in the far past was similar to biological evolution in that inventions stuck around if a tribe was able to pass down the knowledge to their descendants, then it makes sense that progress was so slow. Would it therefore make sense to split the talk of technological progress in two eras: before and after writing / more durable ways to spread inventions? Instead of one super-exponential curve, it would be a piecewise function.
I am unclear whether progress is a smooth super-exponential curve, or whether it is piecewise exponential, or some other piecewise function. In any case, it approximates a smooth curve if you zoom out enough. Robin Hanson has an interesting paper trying to approximate long-run growth as a series of exponential modes with CES transitions between them