At the Helm, Kirk or Spock? The Pros and Cons of Charismatic Leadership
That is a new paper (AEA gate) by the very smart Benjamin Hermalin, here is the abstract:
Charismatic leaders are often desired. At the same time, experience, especially with demagogues, as well as social science studies, raise doubts about such leaders. This paper offers explanations for charismatic leadership’s “mixed report card.” It offers insights into why and when charismatic leadership can be effective; which, when, and why certain groups will prefer more to less charismatic leaders; and how being more charismatic can make leaders worse in other dimensions, particularly causing them to work less hard on their followers’ behalf.
And here is one important part of the model:
…a charismatic leader can get away with concealing bad news without triggering overly pessimistic beliefs.
I like the paper, but isn’t Spock super-charismatic, and all these women want to sleep with him? And isn’t there a long article about all the terrible decisions and advice stemming from Spock in various Star Trek episodes?
What should I ask Reid Hoffman?
I will be doing a Conversation with him, and he has a new book out Impromptu: Amplifying Our Humanity Through AI, co-authored with GPT-4. But I can ask him about anything!
Wednesday assorted links
A Day in the Life
…of the Bank of England, in the eighteenth century. Brought to you by Anne L. Murphy in her new book Virtuous Bankers: A Day in the Life of the Eighteenth-Century Bank of England. This book is a remarkable achievement, as it truly does recreate the rhythms of a typical day in the life of the Old Lady of Threadneedle Street:
Early afternoon was viewed by the Inspectors as one of the times during which the Bank was at its most vulnerable. This was because most of the senior men would leave for the day at around three o’clock, some to pursue second jobs and others to take advantage of the leisure ‘earned’ by their seniority. It was an action that the Inspectors regarded as ‘very extraordinary’ and certainly not what was expected from men in positions of responsibility.
How many other works have sentences like this?:
There is no record of the types of locks generally used at the Bank, but it is clear that the Inspectors were unconvinced of their quality in at least some cases.
This book is not for all readers (not everyone will care about the particular topic), but it represents a very real advance in our understanding of how things actually work.
Catawba Digital Economic Zone Passes Banking Code
The Catawba Digital Economic Zone (CDEZ), a project of the Catawba Indian Nation (I am an advisor), has passed a banking and financial services regulatory code. The code allows financial services companies and banks to receive charters to operate a bank similar to that offered by US states.
…The goal of the code is to create a “best-of-all-worlds” set of laws that will provide the Nation with: 1) a comprehensive legal code for the regulation of traditional and emerging digital financial activities; 2) legal terms that are already recognized and accepted by the federal government for access to the U.S. and global financial systems; and 3) provisions that enhance the Nation’s sovereignty and create competitive advantages for the Nation’s economic development.
To achieve these objectives, the foundation of the code synthesizes terms from the existing financial codes of three states: South Dakota, North Dakota, and Wyoming. These state codes were selected as a starting point based on their commitment to regulating innovative financial activity to protect consumers and encourage responsible innovation.
AGI and the division of powers within government
I’m not sure the AGI concept is entirely well-defined, but let’s put aside the more dramatic scenarios and assume that AI can perform at least some of the following functions:
1. Evaluate many policies and regulations better than human analysts can.
2. Sometimes outperform and outguess asset price markets.
3. Formulate the most effective campaign strategies for politicians.
4. Understand and manage geopolitics better than humans can.
5. Write better Supreme Court opinions and, for a given ethical point of view, produce a better ruling.
You could add to that list, but you get the point. These are a big stretch beyond current models, but not on the super-brain level.
One option, of course, is simply that everyone can use this service, like the current GPT-4, and then few questions arise about differential political access. But what if the service is expensive, and/or access is restricted for reasons of law, regulation, and national security? Exactly who or what in government allocates use of the service within government?
Can any member of the House of Representatives pay the service a visit and ask away? Do incumbents then end up with a major new advantage over challengers?
How do you stop the nuttier Reps from giving away the information they can access, perhaps to unsalubrious parties or foreign powers? Don’t national security issues suddenly become much tougher, as if all Reps suddenly are on the Senate Intelligence Committee?
Surely the President can claim it is a weapon of sorts and access it at will? Can he or she veto the access of other individuals? Will the rival running for President, from the other party, have any access at all?
Can the national security establishment veto the access of individuals within the political establishment? If so, does the Executive Branch and national security establishment gain greatly in power?
Have we now created a kind of “fourth branch” of government?
Do we ask the AI who or what should get access?
Say the Republicans or Democrats win a trifecta? Do they now have a kind of monopoly access over the AI?
Can the technically non-governmental Fed access it? If so, just the chair, the whole FOMC, or the staff as well? If the staff cannot access it, what good are they?
We haven’t even talked about federalism yet — what if a governor has a pressing query? Will Texas build its own model?
Let’s say this is the UK — does the party in opposition have equal access to the AI? Exactly which legal entity with which governance mechanism counts as “the party in opposition”? Can you start a small party, opposing the national government, just to get access?
Say some Brits are in a coalition with one of those tiny parties from Northern Ireland. Can the coalition partner demand access on equal terms? (How about Sinn Fein?) How about in PR systems?
Doesn’t this make all political coalitions higher stakes, more fraught, and more fragile? And more suffused with security risks?
Inquiring minds wish to know.
*The AI Revolution in Medicine: GPT-4 and Beyond*
A new, forthcoming book by Peter Lee, Carey Goldberg, and Isaac Kohane, with Sebastian Bubeck. The researchers were given advance access to GPT-4 (with no editorial controls), and this book documents the power of the results, for instance:
In our testing, when given a full battery of USMLE [medical licensing exam] problems, GPT-4 answers them correctly more than 90 percent of the time.
And it can give very good explanations.
Due out May 13, this book is the documentation, definitely recommended, especially for the skeptics.
Harry Belafonte, RIP
Tuesday assorted links
1. New research on the Terra Luna crash.
2. Is the UK accelerationist? And Perry Metzger on FOOM.
3. Chess.com and the chess boom (NYT). And Harry Lorayne, RIP (NYT, he was a memory wizard).
4. The deep roots of gun deaths.
5. Filipino virtual assistants.
6. The Eric Schmidt theory of Powerpoint?
7. Sloth with child.
8. Zvi on AI tax policy (note there is too much “huff and puff,” I don’t per se endorse emails that I reproduce on MR).
John Lennon and Yoko on Over Population
He was right.
Hat tip: Steven Hamilton who now likes the Beatles.
A few random Tucker Carlson thoughts
A few times I was invited to be on the show, typically after I would write something on immigration. I always refused, figuring I wouldn’t get fair treatment and would only feed a very unfair method of conducting the discourse.
I never have been a regular watcher, not of any news show, but I have seen him a number of times, often in other venues or homes, or I might pause when moving through the channels. It struck me each time how remarkably talented and smart and energetic he was, while what he said was very often not smart at all. (I’ve heard the same about his core smarts from a few different people who knew him when he was younger.) It also struck me how fluently he could coin an attack phrase, maybe better than anyone else but DT?
It is now increasingly debated how much he meant the different things he was saying, for instance about the last presidential election. And was it smart to put criticisms of Fox management into texts?
The biggest lessons here are cautionary ones. First, the biggest stars can do some very unwise things, and eventually so many of them do.
Second, “the right wing” should pay heed to the reality that many of its most talented representatives go down such dark paths. Since most (non-right wing) people are reluctant to admit Carlson’s extreme talent, this point does not always come up. Carlson, of course, was doing very well with his chosen path, and he still has the option of doing very well again.
But in which directions will the constellations of the audience and of fame point him? Which exactly was the guiding star he lacked?
Generative AI at Work
We study the staggered introduction of a generative AI-based conversational assistant using data from 5,179 customer support agents. Access to the tool increases productivity, as measured by issues resolved per hour, by 14 percent on average, with the greatest impact on novice and low-skilled workers, and minimal impact on experienced and highly skilled workers. We provide suggestive evidence that the AI model disseminates the potentially tacit knowledge of more able workers and helps newer workers move down the experience curve. In addition, we show that AI assistance improves customer sentiment, reduces requests for managerial intervention, and improves employee retention.
That is from a new NBER working paper by Erik Brynjolfsson, Danielle Li, and Lindsey R. Raymond.
The One-Child Policy and Intergenerational Mobility in China
We examine whether and how the world’s largest population planning program, the One-Child Policy, has shaped intergenerational mobility in China. Using a dataset with 2,096,798 childparent(s) pairs combined from various rounds of ten separate national household surveys, we leverage exogenous variation in fine rates imposed for One-Child Policy violations across provinces to study causal impacts of the One-Child Policy on intergenerational persistence. Using a continuous difference-in-differences approach, we find that for cohorts born between 1980 and 1996, the One-Child Policy reduced persistence in intergenerational income, education, and social class, comparing to those born prior to 1979. We estimate that the overall effect of the One-Child Policy fines was to reduce persistence in intergenerational income, education, and social class by 28.1%, 48.7%, and 24.8%, respectively. Analyzing mechanisms, we find that the One-Child Policy boosted China’s intergenerational mobility by diminishing elite family heirship, concentrating resources for lower-income families, and decreasing returns to education. The One-Child Policy has brought about a significant socioeconomic reshuffle that has reshaped the role of China’s longstanding class solidification.
That is from a recent paper by Shanthi Manian, Qi Zhang, and Bin Zhao. Via Linghui Han. Might some similar results be true for any other low-fertility societies? Or are the environments too disparate?
Monday assorted links
2. Why you should teach at a community college.
3. Fewer Jews in the Ivy League, though up at Brown.
4. La Nacion interview with me (in Spanish). And new re-review of Notorious Byrd Brothers.
5. Chris Hughes (earlier of Facebook) is now an academic Arthur Burns revisionist? (NYT)
7. Dating apps with AI? And recurrent memory transformers?
8. “So experts doubt that the measures will be tightly enforced.” (The Economist, on Chinese AI restrictions)
Tenured professor has salary reduced to zero
Math professor Harold Donnelly is still listed on Purdue University’s website—but fellow math professors have shared something odd about his salary.
It’s $0, after the university began reducing it in fall 2021, according to emails these professors provided.
“I have carefully considered your career of contributions to mathematics and to our department, college and university,” says a July 2021 email from a math department head, whose name is redacted, about the first salary slash, a 20 percent cut. Irena Swanson, who was the department head then and now, didn’t return requests for comment Friday.
Here is the full story. File under “What, no clawback!?”