Three-Toed Sloth

Tenure-Track Opening in Computational Social Science at CMU (a.k.a. Call to Pittsburgh, 2024 edition)

Attention conservation notice: Advertising an academic position in fields you don't work in, in a place you don't want to live, paying much less than the required skills can get from private industry.

We have a tenure-track opening at the intersection of statistics and complex social systems, a.k.a. computational social science:

The Department of Statistics and Data Science at Carnegie Mellon University invites applications for a tenure track position in Computational Social Science at the rank of Assistant Professor starting in Fall 2025. This position will be affiliated with the Institute for Complex Social Dynamics.

The Department seeks candidates in the areas of social science statistics and data science, as well as related interdisciplinary fields. Potential areas of interest include network science, social simulation, data science for social good, simulation-based inference, cultural evolution, using large text and image corpora as data, and data privacy. Candidates with other research interests related to the work of both the Department and the Institute are also highly encouraged to apply.

The Institute for Complex Social Dynamics brings together scholars at Carnegie Mellon University who develop and apply mathematical and computational models to study large-scale complex social phenomena. The core members of the Institute are based in the Departments of Statistics and Data Science, Social and Decision Sciences, and Philosophy. Interests of the Institute include studies of the emergence of social behavior, the spread of misinformation, social inequality, and societal resilience.

As tenure-track faculty, the successful candidate will be expected to develop an independent research agenda, leading to publications in leading journals in both statistics and in suitable social-scientific venues; to teach courses in the department at both the undergraduate and graduate level; to supervise Ph.D. dissertations; to obtain grants; and in general to build a national reputation for their scholarship. The candidate will join the ICSD as a Core Member, and help shape the future of the Institute.

CMU's statistics department is unusually welcoming to those without traditional disciplinary backgrounds in statistics (after all, I'm here!), and that goes double for this position. If this sounds interesting, then apply by December 15th. (I'm late in posting this.) If this sounds like it would be interesting to your doctoral students / post-docs / other proteges, then encourage them to apply.

(If you'd like to join the statistics department, but are not interested in complex social dynamics what's wrong with you? we have another tenure track opening, where I'm not on the hiring committee.)

Kith and Kin; Enigmas of Chance

Come Post-Doc with Me!

Attention conservation notice: Soliciting applications for a limited-time research job in an arcane field you neither understand nor care about, which will at once require specialized skills and pay much less than those skills command in industry.

I am, for the first time, hiring a post-doc:

The Department of Statistics and Data Science at Carnegie Mellon University invites applicants for a two-year post-doctoral fellowship in simulation-based inference. The fellow will work with Prof. Cosma Shalizi of the department on developing theory, algorithms and applications of random feature methods in simulation-based inference, with a particular emphasis on social-scientific problems connected to the work of CMU's Institute for Complex Social Dynamics. Apart from by the supervisor, the fellow will also be mentored by other faculty in the department and the ICSD, depending on their interests and secondary projects, and will get individualized training in both technical and non-technical professional skills.

Successful applicants will have completed a Ph.D. in Statistics, or a related quantitative discipline, by September 2025, and ideally have a strong background in non-convex and stochastic optimization and/or Monte Carlo methods, and good programming and communication skills. Prior familiarity with simulation-based inference, social network models and agent-based modeling will be helpful, but not necessary.

Basically, I need someone who is much better than I am at stochastic optimization to help out with the matching-random-features idea. But I hope my post-doc will come up with other things to do, unrelated to their ostensible project (God knows I did), and I promise not to put my name on anything unless I actually contribute. If you don't have a conventional background in statistics, well, I'm open to that, for obvious reasons.

Beyond that, the stats. department is a genuinely great and supportive place to work, I hope for fabulous things from ICSD, and CMU has a whole has both a remarkable number of people doing interesting work and remarkably low barriers between departments; Pittsburgh is a nice and still-affordable place to live. Apply, by 15 December!

--- If I have sold you on being a post-doc here, but not on my project or on me, may I interest you in working on social networks dynamics with my esteemed colleague Nynke Niezink?

(The post-doc ad is official, but this blog post is just me, etc., etc.)

Self-Centered

30 Years of Notebooks

Attention conservation notice: Middle-aged dad has doubts about how he's spent his time.

In September 1994, I wanted to write a program which would filter the Usenet newsgroups I followed for the posts of most interest to me, which led me to writing out keywords describing what I was interested in. I don't remember why I started to elaborate the keywords into little essays and reading lists (perhaps self-clarification?), but I did, and then, because I'd just learned HTML and was playing around with hypertext, I put the document online. (My records say this was 3 October 1994, though that may have been fixing on a plausible date retroactively.) I've been updating those notebooks ever since, recording things-to-read as they crossed my path, recording my reading, and some thoughts. The biggest change in organization came pretty early: the few people who read it all urged me to split it from one giant file into many topical files, so I did, on 13 March 1995, ordered by last update, a format I've stuck to ever since (*).

This was not, of course, what I was supposed to be doing as a twenty-year-old physics graduate student. (Most of the notebook entries weren't even about physics.) Unlike a lot of ideas I had at that age, though, I stuck with it --- have stuck with it. Over the last thirty years, I've spent a substantial chunk of my waking hours recording references, consolidating what I understand by trying to explain it, and working out what I think by seeing what I write, by using Emacs to edit a directory of very basic HTML files. (I learned Emacs Lisp to write functions to do things like add links to arxiv.)

Was any of this a good use of my time? I couldn't begin to say. Long, long ago it became clear to me that I was never going to read more than a small fraction of the items I recorded as "To read:". I sometimes tell myself that it's a way of satiating my hoarding tendencies without actually filling my house with junk, but of course it's possible it's just feeding those tendencies. I do use the notebooks, though honestly the have-read portions are the most useful ones to me. Some of the notebooks have grown into papers, though many more which were intended to be seeds of papers have never sprouted. I know that some other people, from time to time, say they find them useful, which is nice. (Though I presume most people's reactions would range from bafflement to "wow, pretentious much?") Whether this justifies all those hours not writing papers / finishing any of my projected books / gardening / hanging out with friends / being with my family / playing with my cat (RIP) / drinking beer / riding my bike / writing letters / writing al-Biruni fanfic / actually reading, well...

The core of the matter, I suspect, is that if anyone does anything for a decade or three consistently, it becomes a very hard habit to break. By this point, the notebooks are so integrated into the way I work that it would take lots of my time and will-power to stop updating them, as long as I keep anything like my current job. So I will keep at it, and hope that it is, at worst, a cheap and harmless vice.

I never did write that Usenet filter.

*: A decade later, I started using blosxom, rather than completely hand-written HTML, and Danny Yee wrote me a cascading style sheet. I also was happy to use first HTMX, and then MathJax, to render math, rather than trying to put equations into HTML. ^.

The Professoriate Considered as a Super-Critical Branching Process

Attention conservation notice: Academic navel-gazing, in the form of basic arithmetic with unpleasant consequences that I leave partially implicit.

A professor at a top-tier research university who graduates only six doctoral students over a thirty year career is likely regarded by their colleagues as a bit of a slacker when it comes to advising work; it's easy to produce many more new Ph.D.s. (Here is a more representative case of some personal relevance.) That slacking professor has nonetheless reproduced their own doctorate six-fold, which works out to $\frac{\log{6}}{30} \approx$ 6% per year growth in the number of Ph.D. holders. Put this as a lower bound --- a very cautious lower bound --- on how quickly the number of doctorates could grow, if all those doctorate-holders became professors themselves. Unless faculty jobs also grow at 6% per year, which ultimately means student enrollment growing at 6% per year, something has to give. Student enrollment does not grow at 6% per year indefinitely (and it cannot, even if you think everyone should go to college); something gives. What gives is that most Ph.D.s will not be employed in the kind of faculty position where they train doctoral students. The jobs they find might be good, and even make essential use of skills which we only know how to transmit through that kind of acculturation and apprenticeship, but they simply cannot be jobs whose holders spawn more Ph.D.s.

The professoriate is a super-critical branching process, and we know how those end. (I am a neutron that didn't get absorbed by a moderator; that makes me luckier than those that did get absorbed, not better.) In the sustainable steady state, the average professor at a Ph.D.-granting institution should expect to have one student who also goes on to be such a professor in their entire career.

Anyone who takes this as a defense of under-funding public universities, of adjunctification, or even of our society having more non-academic use for quantitative skills than for humanistic learning, has trouble with reading comprehension. Also, of course this is Malthusian reasoning; what made Malthus wrong was not anticipating that what he called "vice" could become universal the demographic transition. Let the reader understand.

Learned Folly

Books to Read While the Algae Grow in Your Fur, September 2024

Attention conservation notice: I have no taste, and no qualifications to opine on world history, or even on random matrix theory. Also, most of my reading this month was done at odd hours and/or while chasing after a toddler, so I'm less reliable and more cranky than usual.

Marc Potters and Jean-Philippe Bouchaud, A First Course in Random Matrix Theory: for Physicists, Engineers and Data Scientists, doi:10.1017/9781108768900

I learned of random matrix theory in graduate school; because of my weird path, it was from May's Stability and Complexity in Model Ecosystems, which I read in 1995--1996. (I never studied nuclear physics and so didn't encounter Wigner's ideas about random Hamiltonians.) In the ensuing nearly-thirty-years, I've been more or less aware that it exists as a subject, providing opaquely-named results about the distributions of eigenvectors of matrices randomly sampled from various distributions. It has, however, become clear to me that it's relevant to multiple projects I want to pursue, and since I don't have one student working on all of them, I decided to buckle down and learn some math. Fortunately, nowadays this means downloading a pile of textbooks; this is the first of my pile which I've finished.

The thing I feel most confident in saying about the book, given my confessed newbie-ness, is that Potters and Bouchaud are not kidding about their subtitle. This is very, very much physicists' math, which is to say the kind of thing mathematicians call "heuristic" when they're feeling magnanimous *. I am still OK with this, despite years of using and teaching probability theory at a rather different level of rigor/finickiness, but I can imagine heads exploding if those with the wrong background tried to learn from this book. (To be clear, I think more larval statisticians should learn to do physicists' math, because it is really good heuristically.)

To say just a little about the content, the main tool in here is the "Stieljtes transform", which for an $N\times N$ matrix $\mathbf{A}$ with eigenvalues $\lambda_1, \ldots \lambda_N$ is a complex-valued function of a complex argument $z$, \[ g^{\mathbf{A}}_N(z) = \frac{1}{N}\sum_{i=1}^{N}{\frac{1}{z-\lambda_i}} \] This can actually be seen as a moment-generating function, where the $k^{\mathrm{th}}$ "moment" is the normalized trace of $\mathbf{A^k}$, i.e., $N^{-1} \mathrm{tr}{\mathbf{A}^k}$. (Somewhat unusually for a moment generating function, the dummy variable is $1/z$, not $z$, and one takes the limit of $|z| \rightarrow \infty$ instead of $\rightarrow 0$.)

The hopes are that (i) $g_N$ will converge to a limiting function as $N\rightarrow\infty$, \[ g(z) = \int{\frac{\rho(d\lambda)}{z-\lambda}} \] and (ii) the limiting distribution $\rho$ of eigenvalues can be extracted from $g(z)$. The second hope is actually less problematic mathematically **. Hope (i), the existence of a limiting function, is just assumed here. At a very high level, Potters and Bouchaud's mode of approach is to derive an expression for $g_N(z)$ in terms of $g_{N-1}(z)$, and then invoke the assumption (i), to get a single self-consistent equation for the limiting $g(z)$. There are typically multiple solutions to these equations, but also usually only one that makes sense, so the others are ignored ***.

At this very high level, Potters and Bouchaud derive limiting distributions of eigenvalues, and in some cases eigenvectors, for a lot of distributions of matrices with random entries: symmetric matrices with IID Gaussian entries, Hermitian matrices with complex Gaussian entries, sample covariance matrices, etc. They also develop results for deterministic matrices perturbed by random noise, and a whole alternate set of derivations based on the replica trick from spin glass theory, which I do not feel up to explaining. These are then carefully applied to topics in estimating sample covariance matrices, especially in the high-dimensional limit where the number of variables grows with the number of observations. This in turn feeds in to a final chapter on designing optimal portfolios when covariances have to be estimated by mortals, rather than being revealed by the Oracle.

My main dis-satisfaction with the book is that I left it without any real feeling for why the eigenvalue density of symmetric Gaussian matrices with standard deviation $\sigma$ approaches $\rho(x) = \frac{\sqrt{4\sigma^2 - x^2}}{2\pi \sigma^2}$, but other ensembles have different limiting distributions. (E.g., why is the limiting distribution only supported on $[-2\sigma, 2\sigma]$, rather than having, say, unbounded support with sub-exponential tails?) That is, for all the physicists' tricks used to get solution, I feel a certain lack of "physical insight" into the forms of the solutions. Whether any further study will make me happier on this score, I couldn't say. In the meanwhile, I'm glad I read this, and I feel more prepared to tackle the more mathematically rigorous books in my stack, and even to make some headway on my projects. §

*: As an early example, a key step in deriving a key result (pp. 21--23) is to get the asymptotic expected value of such-and-such a random variable. Using a clever trick for computing the elements of an inverse matrix in terms of sub-matrices, they get a formula for the expected value of the reciprocal of that variable. They then say (eq. 2.33 on p. 22) that this is clearly the reciprocal of the desired limiting expected value, because after all fluctuations must be vanishing. ^

**: We consider $z$ approaching the real axis from below, say $z=x-i\eta$ for small $\eta$. Some algebraic manipulation then makes the imaginary part of $g(x-i\eta)$ look like the convolution of the eigenvalue density $\rho$ with a Cauchy kernel of bandwidth $\eta$. A deconvolution argument then gives $\lim_{\eta \downarrow 0}{\mathrm{Im}(gx-i\eta)} = \pi \rho(x)$. This can be approximated with a finite value of $N$ and $\eta$ (p. 26 discusses the numerical error). ^

***: There is an interesting question about physicists' math here, actually. Sometimes we pick and choose among options that, as sheer mathematics, seem equally good, we "discard unphysical solutions". But sometimes we insist that counter-intuitive or even bizarre possibilities which are licensed by the math have to be taken seriously, physically (not quite "shut up and calculate" in its original intention, but close). I suspect that knowing when to do one rather than the other is part of the art of being a good theoretical physicist... ^

Fernand Braudel, The Perspective of the World, volume 3 of Civilization and Capitalism, 15th--18th Century

This is the concluding volume of Braudel's trilogy, where he tries (as the English title indicates) to give a picture of how the world-as-a-whole worked during this period. It's definitely the volume I find least satisfying. Braudel organizes everything around a notion of "world economies" borrowed from Immanuel Wallerstein (an unfortunate choice of guide), postulating that these are always centered on a single dominating city, and spends a lot of his time tracking the shifts of what he says is the dominating city of the European world economy. But by his own definition of world economy, I don't see how there was more than one during his period, because all his other "world economies (East Asia, India, sub-Saharan Africa, the Americas, etc., etc.) were all tied in to the same economic system as western-and-central Europe. In fact, Braudel goes on at great length about these ties! (At most, Australia and Oceania might have been outside the world economy during this period.) This is also the volume where the, let us say, eccentricity of Braudel's economic thought began to press on me *. It was his discussions of cycles, "the conjuncture" and time-series decomposition which however truly irritated me. Or, rather, it made me want to sit him down and give him a lecture on the Yule-Slutsky effect, because I am quite certain he was smart enough to grasp it **. --- All these remarks are, of course, the height of presumption on my part. §

Previously.

*: After quoting a detailed passage from Ricardo about how both Portugal and England are better off if the former grows wine and the latter grows wheat and they exchange, Braudel spends many pages going over how Portuguese wine-growers came to rely on credit from English merchants. Stipulating that this is all true, and even stipulating that in some sense those English merchants dominated the Portuguese vintners, it does not refute Ricardo! (The cooperative socialist commonwealth will care very much about comparative advantage.) Or, again, Braudel repeatedly talks about how certain cities or countries were dis-advantaged by their high wages, without ever considering that some employers there must have felt those wages were worth paying. Indeed many employers there must have, or those would not have been the prevailing wages. --- In general, I sympathize with wanting to rescue older perspectives, here those of the mercantilists, from the condescension of posterity, but I think Braudel takes that too far, to the detriment of his understanding of his material. ^

**: To be fair, there are some hints in those passages that Braudel might have been happy to accept Slutsky's perspective on the effect. Namely: the appearance of low-frequency cycles is the natural consequence of high-frequency noise (what Braudel would call "events") whose effects just take time to work their way through the economic system. (This reminds me that I need to actually read Barnett's biography of Slutsky one of these years.) ^

Tamim Ansary, The Invention of Yesterday: A 50,000-Year History of Human Culture, Conflict, and Connection

It's not quite true to say that this is an attempt to write Marshall Hodgson's never-completed world history as a volume of pop history. This is not quite true because it is also, and equally, inspired by McNeill and McNeill's The Human Web. The result is extremely engaging, and while I didn't particularly learn from it, I daresay most of the prospective audience will not, in fact, have read as widely in Ansary's sources as I happen to have done. §

Errata: When describing Mesopotamian civilizations, Ansary repeatedly refers to Sumerian as a Semitic language, which is wrong. This is not particularly consequential, and I didn't notice any other errors of fact.

Disclaimer: My grandfather and Ansary's father were friends, so he's a family connection.

Books to Read While the Algae Grow in Your Fur; Writing for Antiquity; The Great Transformation; Mathematics; Enigmas of Chance; The Dismal Science

Books to Read While the Algae Grow in Your Fur, August 2024

Attention conservation notice: I have no taste, and no qualifications to opine on world history. Also, most of my reading this month was done at odd hours and/or while chasing after a toddler, so I'm less reliable and more cranky than usual.

Fernand Braudel, The Wheels of Commerce, volume 2 of Civilization and Capitalism, 15th--18th Century

On re-reading, I am much more struck by Braudel's constant, but un-supported, assertions about who controlled various trades. "Power" and "control" are unavoidably causal notions, which means that at the very least they involve counterfactuals, the sort of thing B. would normally say he avoids. A bit more concretely: say Braudel is right that in the late 1500s and earlier 1600s, the European international currency market was funneled through a small number of mostly Genovese merchants/bankers operating at periodic fairs. (Subsequent scholarship seems to agree.) I'm sure those traders made money. In fact, I'm sure that they exploited network externalities to make supra-normal profits. But asserting that they controlled those currency markets implies that they could have imposed different outcomes --- on exchange rates? on discount rates? Braudel never bothers to say --- if they had wanted to, as opposed to having their actions more or less dictated by the real economic activity which generated bills of exchange, demand for currency exchange, etc. Braudel does not provide evidence for control, and it doesn't even arise as a problem within his horizon.

Wishing Braudel had collaborated with someone like (impossibly) Paul Krugman, or (more plausibly, but still impossibly) Charles Kindleberger, is both idle and impertinent, even philistine, but I can't help it.

Sequel.

Glen Cook, The Black Company, Shadows Linger, The White Rose (also an omnibus, Chronicles of the Black Company, not seen by me)

Mind candy military fantasy, in which Our Protagonists find themselves working as mercenaries for the Dark Lord's only-barely-less-evil ex-wife, on whom the narrator develops an unhealthy crush. (These are not spoilers.) There are a lot of these, and I might pick up more later, but this first trilogy comes to a satisfying ending point.

(I'd be very surprised if these weren't an influence on Graydon Saunders, though the prose style is rather different.)

Megan E. O'Keefe, The Blighted Stars

Mind candy: a very angsty romance wrapped in a shell of space opera. Both aspects of the story are left in media res, and I look forward to reading the sequels.

Phillip Kennedy Johnson et al., The Last God

Mind candy, comic book fantasy heavily influenced by D & D.

Wen Spencer, Tinker

Mind candy fantasy, in which Pittsburgh is transported to Elfland for most of each month. I read it for the local interest (admission to CMU plays a small part of the plot, and bad things happen in Turtle Creek [which I can only too easily believe]), and it was OK, but not good enough to make me pick up any of the many sequels.

Lilith Saintcrow, Moon's Knight

Mind candy portal fantasy, in which Our Protagonist's struggles to escape back to Mundania are rather complicated by her growing recognition that her life there sucked, actually, and maybe fighting strange beasts and stranger people isn't so bad in comparison...

Books to Read While the Algae Grow in Your Fur; Scientifiction and Fantastica; Writing for Antiquity; The Dismal Science; Heard About Pittsburgh PA

Books to Read While the Algae Grow in Your Fur, July 2024

Attention conservation notice: I have no taste. Also, most of my reading and viewing this month was done at odd hours and/or while chasing after a toddler, so I'm less reliable and more cranky than usual.

Elizabeth Bear, Ancestral Night

J. S. Dewes, The Last Watch and The Exiled Fleet

Space opera mind candy. The Bear is good, as usual, but not quite her best. Dewes is new to me, and I'd say not quite as good at either world-building or character development as Bear, but still enjoyable, and I'll get the third book in this trilogy when it comes out.

(Parenthetical with implicit spoilers for Ancestral Night: I can't decide if Bear's characters are merely fooling themselves when they assert that the Synarchy has progressed beyond money, or if Bear has not thought through what "providing more value than you use up" is going to entail, especially when one needs to balance, say, a recovered spacecraft hull against expended fuel. Since Bear's narrator is, demonstrably confused about a lot of important matters, I am inclined to think this is the character's mis-apprehension, perhaps encouraged by propaganda. [But then, I would.])

Prometheus

Spoilers for this movie from 2012 follow.

I enjoy a good re-telling of At the Mountains of Madness more than the next fan, and am pleased to learn that xenomorphs are, in fact, shoggothim. (That is: originally amorphous bits of protoplasm, built as weapons or tools, which learned to imitate their creators and then destroyed them.) But there's a huge part of the plot which makes no sense: if you thought humanity was the creation of beings who were merely an advanced alien species, merely engineers, why of why would you think they have any more of an answer to the riddles of existence than we do, or even a way of making human bodies last forever? The fact that the existentially-befuddled human beings in this scenario have created intelligent androids would seem to make this obvious. (In fact the android character basically says as much!)

Also: Am I right in thinking that this is the first time "being infected by the alien parasite is like being pregnant" has moved from subtext to text?

2001: A Space Odyssey

Re-watched for the nth time as a palette-cleanser after Prometheus. Icy perfection from start to finish. (The iciness is part of the perfection.) --- The influence of Stapledon's Last and First Men and Star-Maker on Clarke, and so on this movie, is, naturally, very patent to me on this re-watch.

The Hidden Fortress, Yojimbo, Sanjuro

And since I was staying up late watching beloved classics... I can't remember exactly how old I was when my father took my brother and me to see a revival of The Hidden Fortress at the old Biograph Theater in Georgetown, but we couldn't have been more than eleven, and we both imprinted. I am happy to say these movies, too, only improve with age and re-watching. §

Books to Read While Algae Grow in Your Fur; Scientifiction and Fantastica; Tales of Our Ancestors

The "Quality Control" Interview for Big Classes

Attention conservation notice: Advice on teaching, which I no longer follow myself.

I teach a lot of big classes --- the undergraduate advanced data analysis class passed 100 students many years ago, and this spring is over 230 --- which has some predictable consequences. I don't get to talk much to many of the students. They're mostly evaluated by how they do on weekly problem sets (a few of which, in some classes, I call "take-home exams"), and I don't even grade most of their homework, my teaching assistants do. While I try to craft problem sets which make sure the students practice the skills and material I want them to learn, and lead them to understand the ideas I want them to grasp, just looking at their scores doesn't give me a lot of information about how well the homework is actually working for those purposes. Even looking at a sample of what they turn in doesn't get me very far. If I talk to students, though, I can get a much better sense of what they do and do not understand fairly quickly. But there really isn't time to talk to 100 students, or 200.

About ten years ago, now, I decided to apply some of the tools of my discipline to get out of this dilemma, by means of random sampling. Every week, I would randomly select a fixed number of students for interviews. These interviews took no more than 30 minutes each, usually more like 20, and were one-on-one meetings, distinct from regular open office hours. They always opened by me asking them to explain what they did in such-and-such a problem on last week's homework, and went on from there, either through the problem set, or on to other topics as those suggested themselves.

In every class I did this in, it gave me a much better sense of what was working in the problems I was assigning and what wasn't, which topics were actually getting through to students and which were going over their heads, or where they learned to repeat examples mechanically without grasping the principle. There were some things which made the interviews themselves work better:

• Reading each students' homework, before the meeting. (Obvious in retrospect!)
• Handing the student a copy of what they turned in the week before. (Though, as the years went on, many brought their laptops and preferred to bring up their copy of the document there.)
• Putting a firm promise in the syllabus that nothing students said in the interview would hurt their grade. (Too many students were very nervous about it otherwise.)
• Putting an equally firm promise in the syllabus that not coming in to the interview, or blowing it off / being uncooperative, would get them a zero on that homework. (Obvious in retrospect.)
• Offering snacks at the beginning of the interview.

Setting aside a fixed block of time for these interviews didn't actually help me, because students' schedules are too all-over-the-place for that to be useful. (This may differ at other schools.)

Choosing the number of students each week to interview has an obvious trade-off of instructor time vs. information. I used to adjust it so that each student could expect to be picked once per semester, but I always did sampling-with-replacement. In a 15-week semester with 100 students, that comes out to about 3.5 hours of interviews every week, which, back then, I thought well worthwhile.

I gave this up during the pandemic, because trying to do a good interview like this over Zoom is beyond my abilities. I haven't resumed it since we went back to in-person teaching, because I don't have the flexibility in my schedule in any more to make it work. But I think my teaching is worse for not doing this.

Corrupting the Young

The Presentation Exchange for Workshops and Classes

Attention conservation notice: Advice for running an academic workshop, which I've only followed myself a few times.

Some years ago, Henry Farrell and I ran a series of workshops about cooperative problem-solving and collective cognition where we wanted to get people with very different disciplinary backgrounds --- political theorists, computer scientists, physicists, statisticians, cognitive psychologists --- talking to each other productively. We hit upon an idea which worked much better than we had any right to hope. (Whether it's ultimately due to him, or me, or to one of us tossing it out as obviously dumb and the other saying "Actually...", neither of us can now recall.) We've both used it separately a few times in other settings, also with good results. Since we both found ourselves explaining it recently, I thought I'd describe it in a brief note.

1. Every participant in the workshop writes a brief presentation, with enough lead time for the organizers to read them all.
   In the context of an inter-disciplinary workshop, what often works best is to describe an outstanding problem in the field.
2. The workshop organizers semi-randomly assign each participant's presentation to someone else, with enough lead time that the assignee can study the presentation.
   Again, in the interdisciplinary context, the organizers try to make sure that there's some hope of comprehension.
   (While I called this the "presentation exchange", it needn't be a strict swap, where A gets assignd B's presentation and vice versa.)
3. Everyone gives the presentation they were assigned, followed by their own comments on what they found interesting / cool / provocative and what they found incomprehensible. No one gives the presentation they wrote.
   In some contexts, I have found it helpful to institute the rule that the author don't get to speak until after the presentation is finished...

Doing this at the beginning of the workshop helps make sure that everyone has some comprehension of what everyone else is talking about, or at least that mis-apprehensions or failures to communicate are laid bare. It can help break up the inevitable disciplinary/personal cliques. It can, and has, spark actual collaborations across disciplines. And, finally, many people report that knowing their presentation is going to be given by someone else forces them to write with unusual clarity and awareness of their own expert blind-spots.

As I said, Henry and I hit on this for interdisciplinary workshops, but I've also used it for disciplinary workshops --- because every discipline is a fractal (or lattice) of sub-sub-...-sub-disciplinary specialization. I've also used it for student project classes, at both the undergrad and graduate level. That requires more hand-holding and/or pastoral care on the part of the teacher than a research workshop, and I've never tried to make it the way I start a class.

Learned Folly; The Collective Use and Evolution of Concepts; Corrupting the Young