Wed Feb 28 22:29:34 EST 2018

Items of Interest

Various web links I found to be of interest recently.

  • How democracies die, explained

    The problems in American democracy run far deeper than Trump.

    In most modern cases, "democracies erode slowly, in barely visible steps." They rot from the inside, poisoned by leaders who "subvert the very process that brought them to power." They are hollowed out, the trappings of democracy present long after the soul of the system is snuffed out.
    ...
    "How Democracies Die" is being read as a commentary on Donald Trump, but the analysis of Trump is the book's least interesting, and least important, contribution. Trump is a symptom, not the cause, of the problems bedeviling American democracy.
    ...
    Demagogues and authoritarians do not destroy democracies. It's established political parties, and the choices they make when faced with demagogues and authoritarians, that decide whether democracies survive.

    Related Ezra Klein Show podcast.
  • Why American doctors keep doing expensive procedures that don't work

    The proportion of medical procedures unsupported by evidence may be nearly half.

    In 2002, The New England Journal of Medicine published a study demonstrating that a common knee operation, performed on millions of Americans who have osteoarthritis - an operation in which the surgeon removes damaged cartilage or bone ("arthroscopic debridement") and then washes out any debris ("arthroscopic lavage") - worked no better at relieving pain or improving function than a sham procedure. Those operations can go for $5,000 a shot.
    Many orthopedic surgeons and medical societies disputed the study and pressed insurance companies to maintain coverage of the procedure. Subsequent research on a related procedure cast further doubt on the value of knee surgeries for many patients with arthritis or meniscal tears, yet the procedures remain in wide use.

    ...
    The knowledge gap is especially large for medical procedures, as opposed to drugs, since there is no FDA for surgery. Doctors learn about new procedures from colleagues, specialty society meetings, and information provided by medical device companies - a potentially arbitrary and unscientific process.

  • Missing data hinder replication of artificial intelligence studies

    The booming field of artificial intelligence (AI) is grappling with a replication crisis, much like the ones that have afflicted psychology, medicine, and other fields over the past decade. AI researchers have found it difficult to reproduce many key results, and that is leading to a new conscientiousness about research methods and publication protocols.
    ...
    The most basic problem is that researchers often don't share their source code. At the AAAI meeting, Odd Erik Gundersen, a computer scientist at the Norwegian University of Science and Technology in Trondheim, reported the results of a survey of 400 algorithms presented in papers at two top AI conferences in the past few years. He found that only 6% of the presenters shared the algorithm's code. Only a third shared the data they tested their algorithms on, and just half shared "pseudocode"-a limited summary of an algorithm. (In many cases, code is also absent from AI papers published in journals, including Science and Nature.)

  • The Downsides to Deep Learning

    We've been promised a revolution in how and why nearly everything happens. But the limits of modern artificial intelligence are closer than we think.

    But there are many things that people can do quickly that smart machines cannot. Natural language is beyond deep learning; new situations baffle artificial intelligences, like cows brought up short at a cattle grid. None of these shortcomings is likely to be solved soon. Once you've seen you've seen it, you can't un-see it: deep learning, now the dominant technique in artificial intelligence, will not lead to an AI that abstractly reasons and generalizes about the world. By itself, it is unlikely to automate ordinary human activities.
    ...
    According to skeptics like Marcus, deep learning is greedy, brittle, opaque, and shallow. The systems are greedy because they demand huge sets of training data. Brittle because when a neural net is given a "transfer test"-confronted with scenarios that differ from the examples used in training-it cannot contextualize the situation and frequently breaks. They are opaque because, unlike traditional programs with their formal, debuggable code, the parameters of neural networks can only be interpreted in terms of their weights within a mathematical geography. Consequently, they are black boxes, whose outputs cannot be explained, raising doubts about their reliability and biases. Finally, they are shallow because they are programmed with little innate knowledge and possess no common sense about the world or human psychology.

  • Moralism and the Arts

    In other words, bad behavior, or even alleged bad behavior, can taint an artistic work, because the artist cannot be separated from his art. This is at least a more interesting proposition than the notion that art should be disqualified just because we don't like the way the artist behaved in private. But is it right?
    ...
    It is also true that art can transcend the private behavior of the artist. A writer, filmmaker, or painter who behaves badly toward wives or lovers can produce art that is deeply sympathetic to women. By the same token, perfectly behaved people can break all kinds of social taboos in their art. To judge the moral component of artistic expression, then, we must look not at the person who made it but at the work itself.

  • The Great Crime Decline

    Drawing the right lessons from the fall in urban violence. Book review.

    In the United States over the past three decades, while people argue about tax cuts and terrorism, the wave of social change that has most altered the shape of American life, as much as the new embankments of the Thames changed life then, has been what the N.Y.U. sociologist Patrick Sharkey calls "the great crime decline." The term, which seems to have originated with the influential Berkeley criminologist Franklin E. Zimring, refers to the still puzzling disappearance from our big-city streets of violent crime, so long the warping force of American life"driving white flight to the suburbs and fuelling the rise of Richard Nixon and Ronald Reagan, not to mention the career of Martin Scorsese. ("Taxi Driver" is the great poem of New York around the height of high crime, with steam coming out of the hellish manholes and violence recumbent in the back seat.) No one saw it coming, and the still odder thing is that, once it came, no one seemed adequately equipped to praise it.

  • Princeton Web Transparency & Accountability Project

    Measure Threats; Create Change; Inform the Public

    We monitor websites and services to find out what user data companies collect, how they collect it, and what they do with it. With our measurement platform, we study privacy, security, and ethics of consumer data usage.   Book.


Posted by mjm | Permanent link | Comments
comments powered by Disqus