Links for July 2016

  • Strokes to the language-processing parts of the brain often manifest as expressive aphasia or fluent aphasia. Both are very grave disabilities, but can be fascinating. The latter looks like this:
  • Good HN discussion surround a Nature article on how bicycles are steered.
  • The results from the arXiv survey are in. Nature characterizes them as very conservative, but I as shocked to find that ~58% of responses thought “Allow readers to comment on papers” was very important, important, or somewhat important. From Andrej Karpathy:

    I developed and maintain Arxiv Sanity Preserver (http://www.arxiv-sanity.com/), one of the Arxiv overlays the article mentions. I built it to try address some of the pains that the “raw” arXiv introduces, such as being flooded by paper submissions without any support or tools for sifting through them.

    I’m torn on how Arxiv should proceed in becoming more complex. I support what seems to be the cited poll consensus (“The message was more or less ‘stay focused on the basic dissemination task, and don’t get distracted by getting overextended or going commercial’”) and I think the simplicity/rawness of arXiv was partly what made it succeed, but there is also a clear value proposition offered by more advanced search/filter/recommendation tools like Arxiv Sanity Preserver. It’s not clear to me to what extent arXiv should strive to develop these kinds of features internally.

    Whether they go a simple or more complex route, I really hope that they keep their API open and allow 3rd party developers such as myself to explore new ways of making the arXiv repository useful to researchers. Somewhat disappointedly, the arXiv poll they ran did not include any mentions of their API, which in my opinion are a critical, overlooked and somehow undervalued.

[continue reading]

Links for June 2016

  • Another transmissible cancer found, this time in mollusks.
  • The “modern” pentathalon is bizarre:

    The modern pentathlon is an Olympic sport that comprises five very different events: fencing, 200 m freestyle swimming, show jumping, and a final combined event of pistol shooting, and a 3200 m cross-country run. The sport has been a core sport of the Olympic Games since 1912 despite dispute…

    The addition of modern to the name distinguished it from the original pentathlon of the ancient Olympic Games, which consisted of the stadion foot race, wrestling, long jump, javelin, and discus. As the events of the ancient pentathlon were modeled after the skills of the ideal soldier of that time, Coubertin created the contest to simulate the experience of a 19th-century cavalry soldier behind enemy lines: he must ride an unfamiliar horse, fight enemies with pistol and sword, swim, and run to return to his own soldiers.

  • Sketches of the flying car design being funded by Larry Page. (H/t Scott Alexander.)
  • Why keep making new car commercials when you can just make one with a dummy car and digitally add in the car after the fact?
  • Everyone should know Moore’s here-is-one-hand argument:

    In his 1925 essay A Defence of Common Sense, Moore argues against idealism and skepticism toward the external world on the grounds that skeptics could not give reasons to accept their metaphysical premises that were more plausible to him than the reasons he had to accept the common sense claims about our knowledge of the world that skeptics and idealists must deny. In other words, he is more willing to believe that he has a hand than to believe the premises of what he deems “a strange argument in a university classroom.” “I do not think it is rational to be as certain of any one of these … propositions”….

[continue reading]

Links for May 2016

  • The Peacock Spider (Maratus speciosus):

    If you haven’t long ago seen the BBC Earth bit on the birds of paradise, check it out.
  • If you use Zotero and iOS, then check out PaperShip. I have two or three minor complaints, but on the whole it is very high quality.
  • The New Mexico whiptail is like a mule in that it’s a hybrid of two species, but unlike the mule it can reproduce semi-cloning:

    The New Mexico whiptail (Cnemidophorus neomexicanus) is a female species of lizard found in the southern United States in New Mexico and Arizona, and in northern Mexico in Chihuahua. It is the official state reptile of New Mexico. It is one of many lizard species known to be parthenogenic. Individuals of the species can be created either through the hybridization of the little striped whiptail (C. inornatus) and the western whiptail (C. tigris), or through the parthenogenic reproduction of an adult New Mexico whiptail.

    The hybridization of these species prevents healthy males from forming whereas males do exist in both parent species (see Sexual differentiation). Parthenogenesis allows the resulting all-female population to reproduce and thus evolve into a unique species capable of reproduction. This combination of interspecific hybridization and parthenogenesis exists as a reproductive strategy in several species of whiptail lizard within the Cnemidophorus genus to which the New Mexico whiptail belongs.

    And in the extremely unlikely event that you don’t already know what parthenogenesis is…

    Parthenogenesis… is a natural form of asexual reproduction in which growth and development of embryos occur without fertilization. In animals, parthenogenesis means development of an embryo from an unfertilized egg cell and is a component process of apomixis.

[continue reading]

Links for April 2016

  • Paul Christiano has bet me $500 at even odds that a self-driving car can be reliably hailed by a member of the general public in at least 10 North American cities by July 2023.

    Details: At least 8 cities must be outside San Fransisco Bay Area. The car must available on at least 50% of days, i.e., not confined to very narrow weather or traffic situations. The car must be self-delivering, in the sense that it drives itself to the user, but not necessarily fully self-driving, in the sense that the user might need to drive it to the destination. (It’s easy to imagine tech and regulatory scenarios where self-driven cars are limited to speeds that are unacceptably slow during transportation of passengers, like the ~20 mph that Google’s car usually does, but are sufficient for getting to the hailing passenger if the density is high enough.) Carl Shulman will adjudicate any edge cases.

    I ascribe a 45% chance that a self-delivering car reaches this threshold, and 38% chance that a fully self-driving car does.

    Here’s a list of optimistic predictions for self-driving car timelines, which notably doesn’t mention the recent Google pessimism.

  • People I know build great stuff!

    My brother Will is an electrical engineer at Apple. He has been heavily involved in improving Apple display technology for the past two years, especially the True Tone feature and especially with the iPad Pro 9. Well, the reviews from the experts are in:

    The Absolute Color Accuracy of the iPad Pro 9.7 is Truly Impressive as shown in these Figures. It is the most color accurate display that we have ever measured. It is visually indistinguishable from perfect, and is very likely considerably better than any mobile display, monitor, TV or UHD TV that you have.

[continue reading]

Links for March 2016

  • With AlphaGo’s victory, Carl Shulman won his $100 bet with me (announced before the match here). For hindsight, here is a bit more evidence that AlphaGo’s win isn’t that shocking — perhaps even right on schedule — and therefore shouldn’t cause you to update much on overall AI progress:

    Comment from mjn:

    Fwiw, the point where the Go curve massively changes slope is when Monte-Carlo Tree Search (MCTS) began to be used in its modern form. I think that’s been an underreported part of AlphaGo’s success: deep networks get the lion’s share of the press, but AlphaGo is a hybrid deep-learning / MCTS system, and MCTS is arguably the most important of the algorithmic breakthroughs that led to computer Go being able to reach expert human level strength.

    (HN discussion.) John Langford concurs on the importance of MCTS.

  • Also: Ken Jennings welcomes Lee Sedol to the Human Loser Club. And: Do the Go prodigies of Asia have a future? (H/t Tyler Cowen.) These articles basically write themselves.
  • Also from Tyler: It was only a matter of time before Facebook began to hire reporters. And: “Will all of economic growth be absorbed into life extension?“:

    Some technologies save lives—new vaccines, new surgical techniques, safer highways. Others threaten lives—pollution, nuclear accidents, global warming, and the rapid global transmission of disease. How is growth theory altered when technologies involve life and death instead of just higher consumption? This paper shows that taking life into account has first-order consequences. Under standard preferences, the value of life may rise faster than consumption, leading society to value safety over consumption growth. As a result, the optimal rate of consumption growth may be substantially lower than what is feasible, in some cases falling all the way to zero.

[continue reading]

Links for February 2016

Just in the nick of time…

  • Eliezer Yudkowsky has a large Facebook thread resulting in many public bets on the Lee Sedol vs DeepMind’s AlphaGo match.

    In particular, I have bet Carl Shulman $100 at even odd that Sedol will win. (For the record, my confidence is low, and if I win it will be mostly luck.) The match, taking place March 9-15, will be streamed live on YouTube.

    Relatedly, here is excellent (if slightly long winded) discussion of why the apparent jump in AI Go ability may be partially attributable to a purposeful application of additional computing power and researcher GO-specific expertise, rather than purely a large jump in domain-general AI power.

  • SciHub has been in the news recently, and I guess they decided to upgrade their appearance.
  • Victorian Humor.
  • Want a postdoc doing theoretical physics, machine learning, and genomics? You’re in luck.
  • Luke Muehlhauser has good quote from Bill Gates on AI timelines.
  • Assortative Mating—A Missing Piece in the Jigsaw of Psychiatric Genetics“.

    Why are psychiatric disorders so highly heritable when they are associated with reduced fecundity? Why are some psychiatric disorders so much more highly heritable than others? Why is there so much genetic comorbidity across psychiatric disorders?

    Although you can see assortative mating for physical traits, like height and weight, with your own eyes, the correlation between spouses is only approximately 0.20 for these traits. For personality, assortative mating is even lower at approximately 0.10. In contrast, Nordsletten and colleagues1 find an amazing amount of assortative mating within psychiatric disorders. Spouse tetrachoric correlations are greater than 0.40 for attention-deficit/hyperactivity disorder (ADHD), autism spectrum disorder (ASD), and schizophrenia.

[continue reading]

Links for January 2016

  • Mechanistic insight into schizophrenia?
  • Wide-ranging (and starry-eyed) discussion on HackerNews about what startup can do to make the world a better place.
  • All six naked-eye-visible planets in one wide-angle image.

    (Source.) You can see the current configuration of the solar system here.
  • Holden Karnofsky argues persuasively that selection bias implies that we should have fewer and more high-quality studies than we would in a hypothetical world with ideal, unbiased researchers.

    Chris Blattman worries that there is too much of a tendency toward large, expensive, perfectionist studies, writing:

     

    …each study is like a lamp post. We might want to have a few smaller lamp posts illuminating our path, rather than the world’s largest and most awesome lamp post illuminating just one spot. I worried that our striving for perfect, overachieving studies could make our world darker on average.

    My feeling – shared by most of the staff I’ve discussed this with – is that the trend toward “perfect, overachieving studies” is a good thing…

    Bottom line. Under the status quo, I get very little value out of literatures that have large numbers of flawed studies – because I tend to suspect the flaws of running in the same direction. On a given research question, I tend to base my view on the very best, most expensive, most “perfectionist” studies, because I expect these studies to be the most fair and the most scrutinized, and I think focusing on them leaves me in better position than trying to understand all the subtleties of a large number of flawed studies.

    If there were more diversity of research methods, I’d worry less about pervasive and correlated selection bias.

[continue reading]

Links for December 2015

[continue reading]

Links for November 2015

[continue reading]

Links for October 2015

  • More well-deserved praise for the Stanford Encyclopedia of Philosophy. Lots to be learned from how the SEP was created. A key chicken-or-egg problem:

    …several SEP authors and editors…said that the encyclopedia is used frequently both as a reference and as a teaching tool. This means that philosophers are some of the SEP’s core readers, and they can alert authors or subject editors to incorrect or insufficient entries.

    Stanford does pay most of the operating costs. But the SEP has a paid staff of only three—Zalta, Nodelman, and Allen—plus five other Stanford employees who spend 20% of their time on technical support. Neither the authors, nor the dozens of subject editors, get so much as a dime for their troubles.

    To pay running expenses not covered by Stanford, the team obtained nearly $2 million in grants over the first 15 years. But they wanted something more sustainable… The SEP asks academic libraries to make a one-time contribution [that now provides around a third of the budget]. That doesn’t get them access to the SEP, since it’s already freely accessible, but they enjoy some extra “member benefits,” like the ability to use their own branding on a version of the encyclopedia, and to save the full archives.

    Moreover, their money goes into an SEP endowment, managed by the same company that takes care of Stanford University’s endowment of over $20 billion. If the SEP ever shuts down, Stanford promises to give the libraries that contributed to SEP all their money back, with interest. “It became a no-risk investment for the libraries, and it’s a way for them to invest in open access,” says Zalta.

    Libraries were enthusiastic. The SEP was able to raise over $2 million from the long list of contributors, and Stanford added $1 million to the library endowment.

[continue reading]

Links for September 2015

  • Chris Blattman on the Center for Global Development’s endorsement of cash transfers. (Report.)
  • Here’s to several decades of grinding out a couple of decimal places to parameterize a charged Higgs:

    Three years ago the BaBar collaboration at SLAC measured the branching ratios for B-meson decay to produce either a muon or a tau. For two slightly different decays, they found 2σ or greater deviations from the democratic standard-model expectation. Now the LHCb collaboration at CERN has confirmed the BaBar result for one of the decays. In a preprint, the Belle group at KEK in Japan has also announced results that show a similar though less strong deviation from the standard model. The figure below (from the Heavy Flavor Averaging Group) shows the branching ratios (R) measured by the groups for the two decays, denoted D and D*, along with the standard-model prediction. Taken together, the groups’ measurements have struck a 3.9-σ blow to the principle of lepton democracy. If they hold up, the standard model will have to be modified—perhaps by the addition of a new charged Higgs boson, whose interactions would depend on mass.

    Importantly, this is a combination of several experiments rather than easily attributable to a systematic mistake in one.

  • Advanced LIGO turns on after completing upgrade. From now on, LIGO will be able to notify any number of 75 astronomical observatories around the world who have agreed to, at a moment’s notice, point their telescopes to the sky in search of light signals corresponding to possible gravitational wave detections.
  • New data on great filter from density of habitable planets.

    these new results offer little support for the scenario where we have a good chance of growing out into the universe and meeting other aliens before a billion of years have passed.

[continue reading]

Links for August 2015

Didn’t quite make it on time this month…

[continue reading]

Links for July 2015

  • Cochrane review finds that there is little evidence that water fluoridation has any dental benefits, although it does not dispute that fluoride in toothpaste reduces tooth decay:

    Studies that attest to the effectiveness of fluoridation were generally done before the widespread usage of fluoride-containing dental products like rinses and toothpastes in the 1970s and later, according to the recent Cochrane study. So while it may have once made sense to add fluoride to water, it no longer appears to be necessary or useful, Thiessen says.

    It has also become clear in the last 15 years that fluoride primarily acts topically, according to the CDC. It reacts with the surface of the tooth enamel, making it more resistant to acids excreted by bacteria. Thus, there’s no good reason to swallow fluoride and subject every tissue of your body to it, Thiessen says.

    Another 2009 review by the Cochrane group clearly shows that fluoride toothpaste prevents cavities.

  • Earth rise, as seen by a Japanese lunar orbiter
  • Dermatomes:

    A dermatome is an area of skin that is mainly supplied by a single spinal nerve.[1] There are 8 cervical nerves (C1 being an exception with no dermatome), 12 thoracic nerves, 5 lumbar nerves and 5 sacral nerves. Each of these nerves relays sensation (including pain) from a particular region of skin to the brain…

    Along the thorax and abdomen the dermatomes are like a stack of discs forming a human, each supplied by a different spinal nerve. Along the arms and the legs, the pattern is different: the dermatomes run longitudinally along the limbs. Although the general pattern is similar in all people, the precise areas of innervation are as unique to an individual as fingerprints.

[continue reading]

Links for June 2015

  • Arrow bending.
  • Casting a fireant colony in molten aluminum:
  • The Crossrail tunnel boring machine in London.
  • The best introduction for laymen to the how and why particle physics experiments are built the way that they are: the Particle Adventure.
  • I have been saying for years that the mostly-positive metric is hands-down the better convention. Peter Woit gathers reasons.
  • Korea’s Team KAIST Wins the 2015 Darpa Robotics Challenge.
  • I share Noah Smith’s enthusiasm for experiments in economic policy. I subscribe to the philosophy that (perhaps charitably) models folks’ political biases as strong Bayesian priors. Whenever we have weak evidence, we are pulled to favor the hypothesis that leans strongest toward our biases while still being minimally compatible with the evidence. Although folks’ beliefs would track truth better if they could reduce bias, and I am pessimistic about this improving, we can still pull different sides together by compiling better evidence.

    There are uncountable fierce political battles that are usually considered intractably ideological, but which I have been convinced by others actually would yield considerably under the introduction of robust, unambiguous evidence. It’s darkly satisfying to think “my opponent ignores reason, so it doesn’t matter what evidence we collect; I’ll just have to beat him with rhetoric”, but this is a mistake largely grounded in a lack of appreciation for how weak the evidence we typically collect really is. Most of the vast resources we individuals put into pitched verbal battles would be better spent on ambitious controlled experiments, e.g., Robin Hanson’s call for larger scale version of the Oregon Medicaid and RAND health insurance experiments.

  • Airplanes use air taken directly from the engine to pressurize the cabin, known as bleed air.
[continue reading]

Links for May 2015

  • Are those cocky physicists-turned-biologists living up to their reputation and embarrassing themselves by ignoring epistasis, epigenetics, and gene-gene interactions? Steve Hsu says no, citing the strong success of additive models in twin studies, and the theoretical explanation from stability for the same.
  • An excellent video (from here) depicting a 1956 vision of the future of autonomous driving.

    (H/t Tyler Cowen.)
  • I’m pretty skeptical of the feasibility Elon Musk’s Hyperloop concept, based on earlier criticism, but an unaffiliated company just signed a contract for the land to build a 5-mile test track, and Musk himself claims a different test is coming to Texas. Note however that the first company is long on hype, short on details.
  • All these amateurs and celebrities have been talking about the long risk of recursively self-improving artificial intelligence, but don’t the actual AI researchers think this is a bunch of hooey? Only if you cherry pick, notes Scott A. In particular, if you’re willing to trust the survey skills of people with skin in the game, then

    …a survey of AI researchers (Muller & Bostrom, 2014) finds that on average they expect a 50% chance of human-level AI by 2040 and 90% chance of human-level AI by 2075.

  • If you are a freshman or sophomore undergrad, GiveWell is hosting a one-week summer fellowship that may be of interest.
  • Rockets engine tanks are often designed with excess hydrogen relative to their oxidizer because the unburnt hydrogen has a higher exit velocity for a given temperature, increasing specific impulse.
  • I have some comments on this critique of the PhD Thesis that’s been making the rounds.

    He would much prefer to see theses’ introductory sections “written along the lines of a good review article, where the student does a critical appraisal of the state of the field”

    Yes!

[continue reading]