This is from an article in the THE. Catherine Heymans is a physicist at the University of Edinburgh, who works on “dark energy”. She is planning to leave the UK to work in Germany (yes, Brexit). But what caught my eye was this quote describing one of those lightbulb moments (pun intended)
Question: As a physics undergraduate, how did you feel when the theory of dark energy first emerged?
Heymans: ‘It was 9am, and I was sat in a lecture theatre waiting for our lecturer to turn up – he was late. Eventually he ran into the room and said: “We’re not going to be studying high-energy astrophysics today, because the most amazing paper has just been published – you have to see this stuff.” It was new data that showed that the expansion of the universe was getting faster and faster, which could only be explained by extra, unseen “dark energy” in the universe.’
It is an interesting test for whether you believe in the ‘research led teaching’ trope. Or is it: will this be in the exam?
In addition to its vulnerability to spoofing, for example, there is its gross inefficiency. “For a child to learn to recognize a cow,” says Hinton, “it’s not like their mother needs to say ‘cow’ 10,000 times”—a number that’s often required for deep-learning systems. Humans generally learn new concepts from just one or two examples.
There is a nice review on Deep Learning in PNAS. The spoofing referred to, is an ‘adversarial patch’ — a patch comprising an image of something else. In the example here, a mini-image of a toaster confuses the AI such that a very large banana is seen as a toaster (the paper is here on arXiv — an image is worth more than a thousand of my words).
Hinton, one of the giants of this field, is of course referring to Plato’s problem: how can we know so much given so little (input). From the dermatology perspective, the humans may still be smarter than the current machines in the real world, but pace Hinton our training sets need not be so large. But they do need to be a lot larger than n=2. The great achievement of the 19th century clinician masters was to be able to create concepts that gathered together disparate appearances, under one ‘concept’. Remember the mantra: there is no one-to-one correspondence between diagnosis and appearance. The second problem with humans is that they need continued (and structured) practice: the natural state of clinical skills is to get worse in the absence of continued reinforcement. Entropy rules.
Will things change? Yes, but radiology will fall first, then ‘lesions’ (tumours), and then rashes — the latter I suspect after entropy has had its way with me.
I noted that he seems to be one of the leading thinkers in the push to rebrand STEM as STEAMED (Science, Technology, Engineering, Arts, Math, and Everything Delightful).
Alan Kay: The Computer Revolution Hasn’t Happened Yet, OOPSLA 1997
Of course, children can learn many things without special mentoring just by experimentation, and by sharing knowledge amongst themselves. But we don’t know of any examples where this includes the great inventions of humanity such as deductive mathematics and mathematically based empirical sciences. To use an analogy: what if we were to make an inexpensive piano and put it in every classroom? The children would certainly learn to do something with it by themselves – it could be fun, it could have really expressive elements, it would certainly be a kind of music. But it would quite miss what has been invented in music over centuries by great musicians. This would be a shame with regard to music – but for science and mathematics it would be a disaster. The special processes and outlook in the latter (particularly in science) are so critical and so hidden that it is crippling not to be taught them as “skills which allow the art”. As Ed Wilson has pointed out, our genetic makeup for social interests, motivations, communication, and invention, is essentially what humans were in the Pleistocene. Much of what we call modern civilization is made from inventions such as agriculture, writing and reading, math and science, governance based on equal rights, etc. These were hard to invent, and are best learned via guides.
Any real education is incapable of robust widely accepted psychometric assessment that will satisfy a professional regulator.
There is one of those beautifully written pieces on Medium, written by the ex-editor of Nature, Philip Ball [link]. It speaks of something particular, and also in the round.
My handwriting has been terrible as long as I can remember. I have tried on various occasions to improve it, but these attempts seldom last as long as the end of the day. In truth, it is not just others who find my writing hard to decipher — after a few minutes I seldom can make much sense of it. And much as though I would like to blame being a doctor for my troubles, I suspect this is just wishful thinking.
Ball’s article is about the fixation on cursive versus, rather than manuscript writing. He writes:
Something like modern cursive emerged from Renaissance Italy, perhaps partly because lifting a delicate quill off and on the paper was apt to damage it and spatter ink. By the 19th century cursive handwriting was considered a mark of good education and character.
I just smile when I read titbits like this. One of those fascinating explanations for something I had never considered or thought about. He argues that the usual arguments for cursive writing — that it is faster, or that it helps with spelling, or that it is useful for those with dyslexia are not well founded. So why does it persist?
He fixes (rightly) on the strange set of beliefs that constitute considered thought in education. You know, the sorts of things that are not far away from the “ I went to school, so I understand education” trope. (The medicine version is of course: ‘I know how to treat people, so I know how to teach other people to treat people’).
It suggests that what teachers “know” about how children learn is sometimes more a product of the culture in which they’re immersed than a result of research and data. It seems unlikely, in this regard, that teaching cursive is unique in educational practice. Which forces us to wonder: What happened to evidence?
This must surely lead us to wonder how much else in education is determined by a belief in what is “right,” unsupported by evidence. Education and learning are difficult to pin down by research. Teaching practices vary, it’s often impossible to identify control groups, and socioeconomic factors play a role. But it’s often the case that the very lack of hard, objective evidence about an issue, especially in the social sciences, encourages a reliance on dogma instead. The danger is greater in education, which, like any issue connected to child rearing and development, is prone to emotive views.
This all bothers me. Not that I think he is wrong, but rather, I find it hard to conceptualise what character of enquiry is both robust and useful in this domain. One look at medical education, and you realise, we are nowhere close to being there.
A few words from Melvyn Bragg about his radio programme ‘In our time’
He insisted that the programme should be “never knowingly relevant” and jumped wildly from the gin craze of the 18th century to the Palaeocene-Eocene thermal maximum. He expected to be out of a job in six months
In times like ours, not a bad motto to live by.
I do not have a coherent overview of many of the traditional professions, but I wonder if people will soon say similar things about doctors.[Link]
“The big issue that concerns me at the moment in the English education system is the supply of high-quality teachers. We’ve seen quality issues in recruitment to teaching and our schools are getting increasingly desperate to find decent teachers. The whole workload issue has come to a big head again in England with teachers having very big workloads and their conditions of service is deteriorating a lot recently. We’re seeing a big exodus in teaching and so of course, we need a bigger inflow to maintain the balance.”
Wiliam is talking about schooling, but it is also true of medical education.[Link].
“For me, I think the issue in the United States in particular is how we improve education at scale. I argue there are two things that have particularly powerful impact. One is a knowledge-based curriculum, recognizing that the purpose of curriculum is to build long-term memory into our students and what distinguishes novices from experts is knowledge not skills. And the second one is creating a culture where every teacher accepts the need to improve, not because they’re not good enough, but because they can be even better.”
I am pleased with the comment about long-term memory: intellect’s ballast. Knowing things matters.
“I have been seriously attempting to raise money to carry out this science education effort ever since the Nobel Prize (in 2001),” Wieman said. “While on sabbatical last year I prepared about 34 proposals for support directed to private individuals and foundations, mostly in Colorado, and to state and federal funding agencies,” he said. None of the proposals were awarded.
The title above and quotes below are from this article by Lincoln Allison. To create teaching machines, you need to make teaching so bad that even the machines can do it. We are almost there.
The most particular annoyance for me was the doubling of seminar size from nine to 18 – allegedly to free up time for research. As if anyone is going to develop the capacity for original thought because they have two or three more hours available in the week! To some of my colleagues, this was merely a technical change, but to me it was the abolition of the real seminar, the thing we should have been most proud of in the English university system.
It was part of a general deprioritising of teaching. I remember a colleague looking at her extremely poor ratings on student “feedback” and remarking gaily: “I’m really not very good at this, am I?” She had just had a book published that was extremely well received, and she couldn’t care less that she was failing in her core duties to communicate her ideas within an academic community. Her remark stiffened my resolve to leave – especially once students picked up the vibe about the level of staff interest in teaching and became less challenging and more instrumental.
Much of what I have seen and heard of UK universities in the 14 years since I retired seems to relate to what I would consider proper university teaching about as much as “value” tinned food relates to fresh food. And I think that just as there are people who have never tasted fresh food, there are people who have not experienced real lectures and seminars.
Education is probably the field in which we deceive ourselves the most, because the damage only appears decades later. We pretend that all children learn at the same rate and in the same way. Every teacher and parent knows this to be untrue, and to deny it is folly. But deny it we do.
Algeria Shut Down the Internet to Prevent Students from Cheating on Exams
Via Bruce Schneier. The solution in New South Wales, Australia was to ban smartphones.
One selling point of MOOCs (massive online open courses) has been that students can access courses from the world’s most famous universities. The assumption—especially in the marketing messages from major providers like Coursera and edX—is that the winners of traditional higher education will also end up the winners in the world of online courses.
But that isn’t always happening.
In fact, three of the 10 most popular courses on Coursera aren’t produced by a college or university at all, but by a company. That company—called Deeplearning.ai—is a unique provider of higher education. It is essentially built on the reputation of its founder, Andrew Ng, who teaches all five of the courses it offers so far. Link
The MOOC story is like so much of tech — or drug discovery for that matter. Finding a use for a drug invented for another reason often offers the biggest payback. This story has barely begun.
This is a tweet from Dylan Wiliam — who knows more about education than…..well I am too polite to go there.
“goes straight to the top of my list of studies that I trust but wish were not true. I think it is the most important book on education I have ever read.”
He is referring to Bryan Caplan’s disturbing and excellent book. (The case against education). One comment of mine: not in all possible worlds.
And no, I wouldn’t have thought the effect was measurable. Wrong again.
From the results presented here it is clear that there has been a slow but steady decline in the frequency of certain variants in the Icelandic gene pool that are associated with educational attainment. It is also clear that education attained does not explain all of the effect. Hence, it seems that the effect is caused by a certain capacity to acquire education that is not always realized.
Maybe it is just me, but I find many of the graphics in the BMJ hard to follow. The image below is from a clinical update on “Depression and anxiety in patients with cancer” (BMJ 28 April 2018, p116-120). It occupies two whole pages. I am not certain what problem the graphic is trying to solve. For me, it just induces a sense of incomprehension. Or nausea.
In dermatology, there was a famous US academic known for producing slides with numerous arrows, many involving feedback. It was professional cargo-cult science (as the BMJ is cargo-cult education). Sam Shuster always cautioned: more than 3 or 4 arrows per slide, usually means bullshit.
That which is simple is wrong; that which is complicated is useless (Paul Valery).
(Isaiah)Berlin had learned that if you studied them with philosophical intent, certain second-rate minds grappling with first-rate problems could teach you more than first-rate minds lost in the shrubbery. (Another reason, perhaps, that he abandoned analytic philosophy.).
Which for some reason reminds me of a quote from the Economist:
Professors fixated on crawling alone the frontiers of knowledge with a magnifying glass.
Training gets a bad rap for a reason – it’s all a bit, well, dull and inflexible. At one point in my life I point blank refused to be in a room with round tables, a flipchart, coloured pens and a bowl of mints for inspiration.
Donald Clark Link. And please no breakout sessions.
This article (‘Humans may not always grasp why AIs act’) in the Economist gets to the right answer, but by way of a silly example involving brain scanning. The issue is that people are alarmed that that it may not be possible to understand how AI might come to a certain decision. The article rightly points out that we have the same problem with humans. This issue looms large in medicine where many clinicians believe they can always explain to students how they come to the correct answer. The following is one of my favourite Geoff Norman quotes:
Furthermore, diagnostic success may be a result of processes that can never be described by the clinician. If the right diagnosis arises from pattern recognition, clinicians are unlikely to be able to tell you why they thought the patient had gout, any more than we can say how we recognize that the person on the street corner is our son. Bowen claims that “strong diagnosticians can generally readily expand on their thinking”; I believe, instead, that strong diagnosticians can tell a credible story about how they might have been thinking, but no one, themselves included, can really be sure that it is an accurate depiction.
We are Strangers to Ourselves, as Timothy Wilson put it.
The article is about Germany, but I just wonder how much the rite of passage of moving out of the family home is relevant.
Second, apprentices in less prestigious positions are paid very poorly, she said. A trainee hairdresser might receive just €350-€400 (£311-£356) a month, not enough to allow them to move out of their parents’ house, Professor Solga explained, and sectors with shortages such as hotel work or food processing often involve shift and evening work. “For young people, they are not the best working conditions,” she said. THE
The result, he says, is not only a meek student population but also “the biggest Ponzi scheme in British history” – a comparison famously made by Theresa May’s former adviser Nick Timothy.
“The great thing about a Ponzi scheme”, Professor Sutherland continued, “is that you can keep expanding it.”
(All the way to jail, some might say).
Young people, both rich and poor, are ill-served by the arms race in academic qualifications, in which each must study longer because that is what all the rest are doing. It is time to disarm.
I guess we need a version of CND fit for out time. Economist.
I love statistics, but I am just not very good at it, and find much of it extremely counter intuitive (which is why it is ‘fun’). The Monty Hall problem floored me, but then Paul Erdos got it wrong too (I am told), so I am in good — and numerate — company. During my intercalated degree in addition to a research methods tutorials (class size, n=2), we had one three hour stats practical each week (class size, n=10). We each used a Texas calculator, and working out a SD demanded concentration. Never mind, that during the rest of the week we were learning how to use FORTRAN and SPSS on a mainframe, ‘slowing’ down the process was useful.
Medicine has big problems with statistics although it is often not so much to do with ‘mathematical’ statistics but evidence in a broader sense. IMHO the biggest abusers are the epidemiologists and the EBM merchants with their clickbait NNT and the like. But I do think this whole field deserves much greater attention in undergraduate education, and cannot help but feel that you need much more small group teaching over a considerable period of time. Otherwise, it just degenerates into ‘What is this test for?’ exam fodder style of learning.
The problems we have within both medicine and medical research have been talked about for a long while. Perhaps things are improving, but it is only more recently that this topic has been acknowledged as a problem amongst practising scientists (rather than medics). This topic certainly resurfaces with increased frequency, and there have been letters on it in Nature recently. I like this one:
Too many practitioners who discuss the misuse of statistics in science propose technical remedies to a problem that is essentially social, cultural and ethical (see J. Leek et al. Nature 551, 557–559; 2017). In our view, technical fixes are doomed. As Steven Goodman writes in the article, there is nothing technically wrong with P values. But even when they are correct and appropriate, they can be misunderstood, misrepresented and misused — often in the haste to serve publication and career. P values should instead serve as a check on the quality of evidence.
I think you could argue with the final sentence of this (selected) quote, but they are right about the big picture: narrow technical solutions are not the problem here. Instead, we are looking at a predictable outcome of the corruption of what being a scientist means.
“What were your most memorable moments at university?”
“There was a man called Walter Ullmann who taught medieval critical philosophy at 10am – and there was standing room only. I went every week, regardless of how wasted I’d got the night before, because he was brilliant.” THE
Reminds me of people queueing to get into to listen to Isaiah Berlin. Some merit as a metric: standing room only. (Until H&S arrive)
PS. And, for another example, see this from a recent book review of a biography of Enrico Fermi (The Last Man Who Knew Everything: The Life and Times of Enrico Fermi, Father of the Nuclear Age. By David Schwartz).
[the author]..He interviewed many of Fermi’s students and colleagues, shedding light also on Fermi the educator (his lectures were so renowned that even notes taken by his assistants were a bestseller).
I am not a big fan of lectures. The single best piece of advice I received at medical school was not to attend. I therefore skipped lectures for three years (although I got the handouts). It is not that all lectures are bad, they are not. It is just that often they are used for ‘content delivery’, much as we think about delivery of a takeaway. They are ill suited to this role, now that we can write and distribute text cheaply. Good lectures serve a different purpose, but you don’t need too many of them and, in my experience of medicine, there are very few people who lecture well. Lecturing well means choosing those fragments of a domain that lend themselves to this media type. Lectures are (and should be) theatre, but the theatre of the mind needs more.
By chance, I came across the following thoughts from the preface to the Ascent of Man (the TV series and the book). Bronowski understood many things, and I still marvel at how prescient his ideas were.
If television is not used to make these thoughts concrete, it is wasted. The unravelling of ideas is, in any case, an intimate and personal endeavour, and here we come to the common ground between television and the printed book. Unlike a lecture or a cinema show, television is not directed to crowds. It is addressed to two or three people in a room, as a conversation face to face – a one-sided conversation for the most part, as the book is, but homely and Socratic nevertheless. To me, absorbed in the philosophic undercurrents of knowledge, this is the most attractive gift of television, by which it may yet become as persuasive an intellectual force as the book.
The printed book has one added freedom beyond this: it is not remorselessly bound to the forward direction of time, as any spoken discourse is. The reader can do what the viewer and the listener cannot, which is to pause and reflect, turn the pages back and the argument over, compare one fact with another and, in general, appreciate the detail of evidence without being distracted by it.
Then there was PowerPoint and lecture capture.
The endless concern about stamps of approval and achievement distorts education and can even rob an interesting career of its joys. A professor friend introducing students at an East Coast college to Beethoven was greeted with a dead-eyed question from the back of the class: ‘Excuse me professor, will this be in the test?
A new copy of Glenn Hubbard and Tony O’Brien’s widely used introductory economics textbook costs more than some smartphones. The phone can send you to any part of the web and holds access to the sum of human knowledge. The book is about 800 heavy pages of static text. Yet thousands of college students around the US are shelling out $250 for these books, each semester, wincing at the many hours ahead of trying to make sense of this attempt to distill the global economy into tiny widgets and graphs.
The easiest way to predict the future is to prevent it.
Original version is Alan Kay (the easiest way to predict the future is to invent it), and this permutation is his, too. As he says, very appropriate for education.