Chat-GPT spinning webs of words from virtually nothing at all; Bret Stephens is a liar; equity premium; fundamental values win in the end; extra value of in-person contact; tacit knowledge...
I commented over on Andrew Gelman's blog 'Statistical Modelling' about the chat-GPT stuff. I was reminded of the old Memorex cassette television commercial with Ella Fitzgerald, "...is it live or is it Memorex?"
Trevor Logan is correct, of course, and it is sadly not astonishing how often we hear the "markets kill discrimination" argument, which fails on both logical and empirical grounds.
I would go further than Logan, and point out that similar considerations apply to employment discrimination. If your workers won't cooperate with a Black employee it is irrational to hire one, no matter how diligent and capable. And there were in fact hardly any Jim Crow laws about employment, yet discrimination was rampant.
That Rohit piece made a good argument for industrial policy in support of domestic manufacturing. It isn't the actual production so much as the ability to turn something from an idea or prototype into a practical product. It's a form of capital accumulation where the capital is a community of human knowledge, embodied in humans, of how to make things. This is a major point of Bunnie Huang's work; his view is that as someone who wants to make things but cannot do so in the US. It's also a major issue in our problems with soaring infrastructure costs. We no longer have an engineering civil service capable of specifying and designing things even at a local level, so the work goes to consultants who have entirely different agency and incentives.
My personal feeling is that this change was a result of the rise of the MBAs in the 1960s which viewed workers and knowledge as fungible. One could hire knowledgeable people as needed, but a lot of knowledge is community knowledge and requires an ongoing community to sustain it. The Navy flirted with this MBA mentality in its nuclear submarine program, but since that debacle has maintained two competing ship builders, providing them with regular work, to maintain the necessary community. I could cite Boeing and its more recent problems. There's a tendency to ignore the stroma and concentrate on the parenchyma which leads to missing half the picture.
One point I quibble with is Rohit's argument about bringing in outside talent. There have been many failures, but there have also been many successes. Look at Qian Xuesen in China who was critical to China's becoming a nuclear and aerospace power but greatly influenced other industries there as well. There are countless more examples involving US companies bringing in talent to build new business areas, many of them very successful.
Self driving cars was a bad example. It wasn't a Levandowski problem. No one has made any real progress on self driving cars. The problem is much harder than it seemed. Self driving cars now are like gene modification therapy in the 1990s, great ideas with inspiration from recent breakthroughs but likely decades from implementation.
I commented over on Andrew Gelman's blog 'Statistical Modelling' about the chat-GPT stuff. I was reminded of the old Memorex cassette television commercial with Ella Fitzgerald, "...is it live or is it Memorex?"
I have always found the fact that Ben Graham said "weighing machine" instead of scale annoying.
Trevor Logan is correct, of course, and it is sadly not astonishing how often we hear the "markets kill discrimination" argument, which fails on both logical and empirical grounds.
I would go further than Logan, and point out that similar considerations apply to employment discrimination. If your workers won't cooperate with a Black employee it is irrational to hire one, no matter how diligent and capable. And there were in fact hardly any Jim Crow laws about employment, yet discrimination was rampant.
Do electrons yield to Ampere pressure?
Ohm my Gauss! Watt a re-Volt-ing pun.
That Rohit piece made a good argument for industrial policy in support of domestic manufacturing. It isn't the actual production so much as the ability to turn something from an idea or prototype into a practical product. It's a form of capital accumulation where the capital is a community of human knowledge, embodied in humans, of how to make things. This is a major point of Bunnie Huang's work; his view is that as someone who wants to make things but cannot do so in the US. It's also a major issue in our problems with soaring infrastructure costs. We no longer have an engineering civil service capable of specifying and designing things even at a local level, so the work goes to consultants who have entirely different agency and incentives.
My personal feeling is that this change was a result of the rise of the MBAs in the 1960s which viewed workers and knowledge as fungible. One could hire knowledgeable people as needed, but a lot of knowledge is community knowledge and requires an ongoing community to sustain it. The Navy flirted with this MBA mentality in its nuclear submarine program, but since that debacle has maintained two competing ship builders, providing them with regular work, to maintain the necessary community. I could cite Boeing and its more recent problems. There's a tendency to ignore the stroma and concentrate on the parenchyma which leads to missing half the picture.
One point I quibble with is Rohit's argument about bringing in outside talent. There have been many failures, but there have also been many successes. Look at Qian Xuesen in China who was critical to China's becoming a nuclear and aerospace power but greatly influenced other industries there as well. There are countless more examples involving US companies bringing in talent to build new business areas, many of them very successful.
Self driving cars was a bad example. It wasn't a Levandowski problem. No one has made any real progress on self driving cars. The problem is much harder than it seemed. Self driving cars now are like gene modification therapy in the 1990s, great ideas with inspiration from recent breakthroughs but likely decades from implementation.