Cheap Document Camera

I built this using a $35 1080p web camera and a $20 LED lamp I already had in my office. 

I've tested on Google Meet and Zoom, it allows me to display equations and quick sketches to collaborators and students. There are much fancier purpose-built document cameras with similar functionality, but these are mostly sold out on Amazon, due to the increase in remote work and online teaching. The particular web cam I am using (see link above) has a manual focus, in case the software auto-focus is unsatisfactory.

Both Google Meet and Zoom allow to switch from the default internal camera on my laptop to the external web cam (a simple toggle in Settings). The web cam is auto-detected on both Mac OS and Chrome -- I did not have to install any drivers.

The plastic clip I used to attach the web cam to the lamp is from the kitchen (Bed Bath and Beyond). Tape would also work as the web cam is very light.

Here is the rig in action, using Google Meet. I write on the pad and my colleague can see it very clearly.

Magical Mystery MOOCs

I predicted in my earlier post Whither higher education? that the main beneficiaries of MOOCs would be people who are well above average in intelligence and/or drive to learn.

NYTimes: Year of the MOOC

... Some students are also ill prepared for the university-level work. And few stick with it. “Signing up for a class is a lightweight process,” says Dr. Ng. It might take just five minutes, assuming you spend two devising a stylish user name. Only 46,000 attempted the first assignment in Dr. Ng’s course on machine learning last fall. In the end, he says, 13,000 completed the class and earned a certificate — from him, not Stanford.

That’s still a lot of students. The shimmery hope is that free courses can bring the best education in the world to the most remote corners of the planet, help people in their careers, and expand intellectual and personal networks. Three-quarters of those who took Dr. Patterson’s “Software as a Service” last winter on Coursera (it’s now on edX) were from outside the United States, though the opposite was true of a course on circuits and electronics piloted last spring by Dr. Agarwal. But both attracted highly educated students and both reported that over 70 percent had degrees (more than a third had graduate degrees). And in a vote of confidence in the form, students in both overwhelmingly endorsed the quality of the course: 63 percent who completed Dr. Agarwal’s course as well as a similar one on campus found the MOOC better; 36 percent found it comparable; 1 percent, worse.

Ray Schroeder, director of the Center for Online Learning, Research and Service at the University of Illinois, Springfield, says three things matter most in online learning: quality of material covered, engagement of the teacher and interaction among students. The first doesn’t seem to be an issue — most professors come from elite campuses, and so far most MOOCs are in technical subjects like computer science and math, with straightforward content. But providing instructor connection and feedback, including student interactions, is trickier. ...

From Whither higher education?

1. Internet technology can enhance learning. However, I think the largest impact will be on cognitively gifted or very motivated individuals who will be able to accelerate their education (see, e.g., Khan Academy). For average students, the main barriers to learning have to do with self-motivation and I am not sure that streaming video of lectures, or even a virtual classroom environment which allows rich interaction, will provide better stimulus than the traditional lecture. It seems to me that my intro students have trouble paying attention even when I am literally dancing around at the front of the class, telling jokes and working through elaborate physics demonstrations (which often include explosions or bouncing balls or colorful animations). Moving the lectures online will be cheaper, but not necessarily better -- a win for efficiency, perhaps, but no solution for the difficulty that the average individual has in mastering challenging material.

Ask yourself what the ideal learning environment would be for your child if cost were no object. I think it might be the Oxbridge tutorial system, where a real expert devotes their full attention to training a small number of students (perhaps even a single individual) in great depth. Almost as good would be training in an environment where the student to faculty ratio is low, and the faculty are very focused on pedagogy. Interactions with peers of similar (or superior) ability are as important as those with the tutor/instructor. This ideal limit is quite far from the online systems currently envisaged. Is America too poor to provide this old-fashioned but superior education to (say) the top 10 percent of students? I doubt it.

At the highest quality levels, educational productivity has increased little in the last 100 years. We might improve things around the edges by, say, having lectures from the top scholars available online, along with tools enabling students from different universities to exchange ideas and answer each other's questions. But I don't think we'll see substantive productivity improvement here until we -- gulp -- solve the AI problem and create robot genius professors. Only a small number of students could crowd around Feynman at Caltech's Physics X to hear him explain the EPR paradox. I don't expect that to change anytime soon. (You can record Feynman's comments about EPR; you can't allow thousands of students around the world to interact with him one on one.)


2. Credentialing is complex and even the system we have had in place for several generations is not well understood either by students or by employers. What are the key factors that employers need to determine about an applicant? Intelligence (reasonably well measured by simple tests; but even this is not widely acknowledged in broader society), Conscientiousness (difficult to measure without actually putting someone through a challenging program over a period of years), Ambition/Drive (similar to Conscientiousness), and finally: Creativity, Adaptability and Interpersonal Skills -- all extremely difficult to measure.

I am not sure that Internet technologies will really improve our credentialing capabilities. We already have testing centers, GRE subject exams, Actuarial exams, narrow skill certifications like Microsoft MCSE, etc. It's more a matter of cultural attitudes than anything else -- when will employers start accepting a high SAT score and some narrow skill certification in place of, say, an engineering degree from a well-known university? Has anyone done systematic research on the relative validities (predictive power) of different kinds of certification for a wide variety of employment settings? I only know of results for general cognitive ability (g).

The burden of students

I always enjoyed interacting with Sidney Coleman (sadly, now deceased) when I was a postdoc. I was quite pleased to find this interview, part of the AIP Oral History project.

His views about working with students are not surprising to me, despite the high quality of Harvard PhD students. The gap in brainpower between Sidney and even an exceptional graduate student might be vast. It's worth noting that Sidney had a large number of PhD students who became prominent theorists.

I often make the analogy between teaching (or training PhD students) and pushups or running. Perhaps unpleasant while you are doing it, but (hopefully) it makes you stronger. Certainly I learn a lot from teaching, if only from reviewing the material in preparation for lectures. If the students are exceptionally good, I might even learn things from questions asked in class.

But you do enjoy working with students, or do you?

Coleman: No. I hate it. You do it as part of the job. Well, that's of course false...or maybe more true than false when I say I hate it. Occasionally there's a student who is a joy to work with. But I certainly would be just as happy if I had no graduate students. There are plenty of colleagues around here whom I can work with. There are plenty of research fellows; junior faculty. This is true all through the Cambridge area. There's not only Harvard, there are people to work with at MIT, at Brandeis, and there are some good people at places like Northeastern... places loaded with physicists to collaborate with, to talk about physics ideas with, who are ready and KNOW basically how to do research. You know who's good and who's bad. It's not a question of their being embryonically possibly good or possibly rotten. So certainly if I want physicists to collaborate with I don't have to have graduate students. Occasionally there is a graduate student who is a joy to collaborate with. Both David (Politzer) and Eric (Weinberg) were of this kind, but they were essentially almost mature physicists. They were very bright by the time they came to me. In general, working with a graduate student is like teaching a course. It's tedious, unpleasant work. A pain in the neck. You do it because you're paid to do it. If I weren't paid to do it I certainly would never do it.

Interview with Dr. Sidney Coleman by Katherine Sopka at Harvard Physics Department, Cambridge, Massachusetts January 18, 1977.

Teaching effectiveness

The two figures below (click for larger versions) are taken from the Brookings report by Gordon, Kane and Staiger: Identifying Effective Teachers Using Performance on the Job. The report has received a lot of attention recently thanks to Malcolm Gladwell's New Yorker article. Both are worth a look if you are interested in education. The top figure shows that certification has no impact on teaching effectiveness. The second shows that effectiveness measured in the years 1 and 2 is predictive of effectiveness in the subsequent year. In this case effectiveness is defined by the average change in percentile ranking of students in the teacher's class. Good teachers help their students to improve their mastery, hence percentile ranking, relative to the average student studying the same material.






It's obvious to me that there is gigantic variation in effectiveness among teachers. Gladwell emphasizes how difficult it is to evaluate teaching capability in initial hiring, and how the single most important impact on overall school effectiveness is due to individual teachers (he also makes the analogy to scouting college QBs for pro football -- it's very hard to predict NFL performance based on college performance). The Brookings paper has many policy suggestions, but the basic idea is that if we were disciplined and data-driven we could easily determine which teachers are good and which ones are not.

New Yorker: ...One of the most important tools in contemporary educational research is “value added” analysis. It uses standardized test scores to look at how much the academic performance of students in a given teacher’s classroom changes between the beginning and the end of the school year. Suppose that Mrs. Brown and Mr. Smith both teach a classroom of third graders who score at the fiftieth percentile on math and reading tests on the first day of school, in September. When the students are retested, in June, Mrs. Brown’s class scores at the seventieth percentile, while Mr. Smith’s students have fallen to the fortieth percentile. That change in the students’ rankings, value-added theory says, is a meaningful indicator of how much more effective Mrs. Brown is as a teacher than Mr. Smith.

It’s only a crude measure, of course. A teacher is not solely responsible for how much is learned in a classroom, and not everything of value that a teacher imparts to his or her students can be captured on a standardized test.

Nonetheless, if you follow Brown and Smith for three or four years, their effect on their students’ test scores starts to become predictable: with enough data, it is possible to identify who the very good teachers are and who the very poor teachers are. What’s more—and this is the finding that has galvanized the educational world—the difference between good teachers and poor teachers turns out to be vast.

Eric Hanushek, an economist at Stanford, estimates that the students of a very bad teacher will learn, on average, half a year’s worth of material in one school year. The students in the class of a very good teacher will learn a year and a half’s worth of material. That difference amounts to a year’s worth of learning in a single year. Teacher effects dwarf school effects: your child is actually better off in a “bad” school with an excellent teacher than in an excellent school with a bad teacher. Teacher effects are also much stronger than class-size effects. You’d have to cut the average class almost in half to get the same boost that you’d get if you switched from an average teacher to a teacher in the eighty-fifth percentile. And remember that a good teacher costs as much as an average one, whereas halving class size would require that you build twice as many classrooms and hire twice as many teachers.

Hanushek recently did a back-of-the-envelope calculation about what even a rudimentary focus on teacher quality could mean for the United States. If you rank the countries of the world in terms of the academic performance of their schoolchildren, the U.S. is just below average, half a standard deviation below a clump of relatively high-performing countries like Canada and Belgium. According to Hanushek, the U.S. could close that gap simply by replacing the bottom six per cent to ten per cent of public-school teachers with teachers of average quality. After years of worrying about issues like school funding levels, class size, and curriculum design, many reformers have come to the conclusion that nothing matters more than finding people with the potential to be great teachers. But there’s a hitch: no one knows what a person with the potential to be a great teacher looks like. The school system has a quarterback problem.

In my experience as a university professor I find that most colleagues think of themselves as above-average teachers, even when they are not. Essentially no "value-added" analysis is ever done, so people can have a 30 year teaching career without ever realizing that they aren't effective in the classroom. I've done many dozens of business presentations, to venture capitalists, technology partners, customers, analysts and even potential M&A acquirers, which has helped me improve my own teaching and communication skills. Despite the business setting such meetings are 90 percent teaching -- trying to convey key points to the audience in a limited time. I'm usually there with a team and my team isn't shy about telling me afterwards what worked and what didn't work, so I've had a lot of honest feedback that most professors never get.



The New Yorker cartoon and article capture some essential aspects of teaching and communication that are not widely understood. The teacher has to be simultaneously on top of the material itself and aware of what the class is doing / thinking / confused about. The big neglected factors in teaching are the ability to be a kind of air traffic controller (or symphony conductor) for the class, and the ability to empathize with (read the mind of) an individual student, to see what, exactly, is confusing them.

Bell and GHZ: spooky action at a distance

I think it is safe to say that no one understands quantum mechanics. -- Richard Feynman

Recently I've been lecturing on quantum weirdness (in Einstein's terminology, "spooky action at a distance") in my graduate quantum mechanics class. The main result is Bell's theorem:

No physical theory of local hidden variables can ever reproduce all of the predictions of quantum mechanics.

Usually this result is proved using the Bell inequalities, which have been tested experimentally. The problem with the Bell inequalities is that they are statistical in nature. I prefer to discuss the so-called GHZ states:

| GHZ > = | 000 > - | 111 >

(after Greenberger, Horne and Zeilinger), with which one can demonstrate a much sharper disagreement between local reality and quantum mechanics.

It's interesting that so much time elapsed between Einstein's 1935 paper with Podolsky and Rosen (EPR) that first discussed spooky action at a distance, and Bell's theorem in 1964. Bell was a particle theorist working at CERN who only did foundations of quantum mechanics on the side (he's also the Bell in the Adler-Bell-Jackiw anomaly in quantum field theory). The GHZ paper didn't appear until 1989. For a long time foundations of quantum mechanics was dismissed by physicists as a fringe activity, suitable only for fuzzy headed philosophers. It's only recently, with the explosion of work in quantum information, that there has been renewed interest in the subject.

I find that the hardest thing about teaching this material in class is that, after half a year of training students' brains to think quantum mechanically, it is extremely difficult to get them to feel the weirdness of Bell's theorem and spooky action. It all seems quite normal to them in the context of the course -- they know how to calculate, and that's just how quantum mechanics works!

For my limited thoughts on quantum foundations (mostly about many worlds or "no collapse" formulations), see this talk (PDF) I gave at the Institute for Quantum Information at Caltech, and these blog posts.

Amazingly, I found almost all the reference links above (to GHZ, Bell's theorem, Bell inequality, EPR) on Wikipedia!

Note added: See Dave Bacon on ScienceBlogs for more discussion and some comments. It appears many younger physicists claim to not find QM weird. However, there may be some selection bias towards researchers in quantum information, who generally work in a non-relativistic setting, and may not have thought as much about causality, the light cone, the intricate spacetime structure of quantum field theory, etc. (i.e., unlike Einstein). Or, it could really be a generational change :-)

The curve of binding energy

I'm teaching about fission, fusion, nuclear power and bombs in class this week. I always search for the simplest way to organize and present complex material (students may disagree ;-). For this set of topics, I am struck by the elegance of the curve below.



It reminded me of The Curve of Binding Energy by New Yorker writer John McPhee, which I read many years ago. In it, he profiles Theodore Taylor, a leading bomb designer at Los Alamos who eventually became an anti-nuclear activist.

Theodore Brewster Taylor was born on July 11, 1925, in Mexico City. His grandparents had been missionaries, and his father was general secretary of the Y.M.C.A. in Mexico. A brilliant boy (he completed sixth grade the same year he started fourth), Ted was enthralled by his chemistry set, or, more precisely, its explosive possibilities.

"He enjoyed putting potassium chlorate and sulfur under Mexico City streetcars," Mr. McPhee wrote. "There was a flash, and a terrific bang."

Dr. Taylor received a bachelor's degree from the California Institute of Technology in 1945 and pursued a doctorate in physics at the University of California. But he failed his oral examinations - he lacked the capacity to focus on things that did not interest him - and he left the department in 1949. (He would eventually earn a Ph.D. from Cornell in 1954.)

He found a job at Los Alamos. "Within a week, I was deeply immersed in nuclear weaponry," Dr. Taylor wrote in a 1996 article in Bulletin of the Atomic Scientists. "I was fascinated by every bit of information I was given during those first few days."

Preternaturally inept at ordinary tasks (parking a car defeated him), he became an artist of the fission bomb, taking the massive nuclear weapons developed for the Manhattan Project and making them smaller and lighter without sacrificing explosive power. Over the next seven years, he designed a series of ever-smaller bombs, whose cunning names - Scorpion, Wasp, Bee, Hornet - captured both their size and their sting.

Dr. Taylor would develop the smallest fission bomb of its time, Davy Crockett, which weighed less than 50 pounds. (By contrast, Little Boy, dropped on Hiroshima, weighed almost 9,000 pounds.) At the other extreme, he designed Super Oralloy, which was at the time, Mr. McPhee wrote, "the largest-yield pure-fission bomb ever constructed in the world."

Viewed as a theoretical abstraction, Dr. Taylor's work had a cool, compelling elegance. Exploded in the Nevada desert, it made a satisfying flash and bang. The weapons, he often reminded himself, were meant to deter nuclear war, and if the United States did not develop them, the Soviets soon would.

In his 1996 article, he recalled how he spent Nov. 15, 1950, the day his daughter Katherine was born:

"Instead of being with my wife, Caro, I had spent the day at a military intelligence office, poring over aerial photographs of Moscow, placing the sharp point of a compass in Red Square and drawing circles corresponding to distances at which moderate and severe damage would result from the explosion at different heights of a 500-kiloton made-in-America bomb. I remember feeling disappointed because none of the circles included all of Moscow."

Nuclear Weapons Responsibility

Presentation by Theodore B. Taylor, PhD, 20 April 1998, at Mickleton Monthly Meeting, Religious Society of Friends (Quakers)

...It's a long and dreary story, those twenty years or so of working on nuclear weapons. How that happened to take place after my writing home and saying I'm never going to work on these things was, I think, the kind of rationalization that anybody goes through when they are facing an addiction of some kind. That is, you have to make excuses for why you're doing this.

After some student activism at the University of California at Berkeley, in which three of us got very intense about calling for a general strike of all nuclear physicists worldwide, until the bombs were gone, we presented that to [J. Robert] Oppenheimer, who said, "Take it, burn it, forget you ever had anything to do with it, because you're going to be labeled as Communists the rest of your lives if you don't do what I say." Well, we didn't burn it, we didn't forget it, but we didn't pursue it.

Not long after that, I found myself very interested in the work I was doing, which at that time wasn't on bombs, it was high-energy physics at the University of California laboratory. In that situation, I did very well at the laboratory, but I did very poorly preparing for my oral exams on various subjects. I wasn't interested in those subjects. To make a somewhat long story short, I flunked out of graduate school. Although I was reinstated later if I wanted to, my boss at the Radiation Laboratory in Berkeley, Robert Serber, calmed me down. He was very happy with some work I had been doing for him and with other theoretical physicists, and he said, "Don't worry; I'll get you a job at Los Alamos." And so, he called a person who, slightly later, became my boss, Carson Mark, and said, "There's this fellow, here, who's very good at what he's interested and very bad at what he's not interested in. Why don't you hire him? I'll bet he'll do something very helpful to the laboratory."

So Caro - my wife - and I and a four-month-old baby arrived at Los Alamos, November 1949. I suddenly just got so high within a week on what I was doing - finding out there were some real secrets about how these things work, things I had never imagined - but more important to me, as it turned out later, was there were a lot of things not yet followed through. My job was to look for extremes, things that people hadn't really tried before, to answer the question, can you make a bomb that can be fired out of a cannon, can you make a bomb that can be fired out of something more like a rifle, how big can you make a bomb, can you make a bomb that would destroy all of Moscow - which the bomb that destroyed Hiroshima would not do, by a long shot. So I got caught up in extremes. That went on for almost 20 years, not all of it at Los Alamos. I then changed jobs, because I wanted to try my hand at designing nuclear power systems, for peaceful purposes. ...

In reading about Taylor, I couldn't help but notice strange parallels with the life of another cold war Theodore -- Ted Kaczynski, the unabomber.

Two slit experiment

I've been teaching about the two slit experiment in my introductory physics class. The course is for non-science majors, so I'm not allowed to use much math. It's a great challenge to convey the concepts of modern physics under this constraint. The two slit experiment, as emphasized by Feynman, exhibits the central mystery of quantum mechanics. While it's not surprising that waves diffracted through slits can create a static interference pattern on a distant screen, the fact that individual particles like photons or electrons can be made to interfere with themselves, is indeed shocking and mysterious. From a colleague's web page: The dark and light regions are called interference fringes, the constructive and destructive interference of waves.

 
If we lower the intensity of light (or the flux of electrons), we should be able to see each photon (electron) strike the screen. Each photon (electron) makes a dot on the screen, but where is the interference pattern?

 
The interference pattern is still there, it simply takes some time for enough photons, or electrons, to strike the screen to build up a recognizable pattern. Interference, a wave phenomenon, is still occurring even if we only let the photons, or electrons, through one at a time. So what are the individual particles interfering with? Apparently, themselves!

 
One can now demonstrate this phenomenon in an introductory class.

With Two-Slit Interference, One Photon at a Time, TeachSpin has built an apparatus that allows students to encounter wave-particle duality with photons, the quanta of light. With this instrument, students perform the seminal two-slit interference experiment with light, even at the limit of light intensities so low that they can record the arrival of individual photons at the detector. That raises the apparent paradox which has motivated the concept of duality: in the very interference experiment which makes possible the measurement of wavelength, one observes the arrival of light energy in particle-like quanta, individual photon events. How is it possible for light to propagate as if it were a wave and yet to be detected as if it were a particle? This paradox is the central theme in Richard Feynman's introduction to the fundamentals of quantum mechanics: "We choose to examine a phenomenon which is impossible, absolutely impossible, to explain in any classical way, and which has in it the heart of quantum mechanics. In reality, it contains the only mystery. We cannot make the mystery go away by explaining how it works . . . In telling you how it works we will have told you about the basic peculiarities of all quantum mechanics." It is the purpose of this apparatus to make the phenomenon of light interference as concrete as possible, and to give students the hands-on familiarity which will allow them to confront duality in precise and operational terms. When they have finished, students might not fully understand the mechanism of duality - Feynman asserts that nobody really does - but they will certainly have had direct experience of the phenomenon itself.