Links for June-July 2017

Elephants are secretly wearing high heels.
Cost per unit hard-drive space is flattening (for consumer models).
On the historical visibility of the Southern Cross:
Crux was known to the Ancient Greeks due to the fact that it can be seen from southern Egypt; Ptolemy regarded it as part of the constellation Centaurus. It was entirely visible as far north as Britain in the fourth millennium BC. However, the precession of the equinoxes gradually lowered its stars below the European horizon, and they were eventually forgotten by the inhabitants of northern latitudes. By AD 400, most of the constellation never rose above the horizon for Athenians.
Zotero 5.0 has significant changes and is out now.
“Mainland China has 36 nuclear power reactors in operation, 21 under construction, and more about to start construction.” See also Wikipedia and this long piece on Chinese investment in Namibia. In comparison, the US gets essentially all nuclear power from reactors built at least 30 years ago, and has just 4 new reactors under construction.
Sentience Institute: “In discussions of effective animal advocacy (EAA) — the field of study for how we can most effectively help animals, also known as effective altruism for animals — there are several important, challenging, and sometimes controversial foundational questions that come up over and over. This post attempts to summarize and catalog the key evidence cited by EAA supporters on each side of these debates for easy reference.”
Third black hole merger detected by LIGO. No neutron stars yet. Binary BH distribution might be more massive and have more misaligned spins than popular models. Nothing revelatory.
Vulcan aerospace unveils airplane with world’s largest wingspan (by far) as part of air-launch orbital rocket service.
GiveWell continues to expand from charity evaluator to charity creator though their incubation grant program. Naively, I’m pessimistic on the ability for the most existing philanthropic institutions to internalize EA philosophy, and I think they will just have to be out-competed by new charities. So I strongly support this.
Server room with seismic isolation floor in Japan earthquake:
From Chris Blattman: Paul Romer on clear writing and randomized controlled trial against sweatshops.
It’s not commonly known that antibiotic use on farms is primarily to accelerate weight gain rather than to treat (observable) infections. No one fully understands why this works, but the suppression of sub-symptomatic infections is one possibility. Interestingly, this has been considered as a cheap and scalable way to ameliorate childhood malnutrition in the developing world. GiveWell briefly investigated it as a cause area, but concluded that there was “Not enough known to make an informed guess”. There is conflicting evidence at to whether this is effective. (I presume there is less risk of this contributing to antibiotic resistance than when it is deployed intensively on farms, and the remaining risk is outweighed by the much greater ethical imperative to reduce childhood morbidity. )
Relatedly: How to invest in early-stage philanthropic organizations like they were for-profit start-ups? What principles carry over and which don’t? This essay has issues through. For instance, VC firms and Hedge funds are almost risk neutral, non-profit investing is actually competitive in many (though weirder) ways, and getting data in non-profits is usually harder (not easier) than for-profits.
The classic phrase “England expects that every man will do his duty”
…was a signal sent by Vice Admiral Horatio Nelson, 1st Viscount Nelson, from his flagship HMS Victory as the Battle of Trafalgar was about to commence on 21 October 1805. Trafalgar was a decisive naval engagement of the Napoleonic Wars. It gave the United Kingdom control of the seas, removing all possibility of a French invasion and conquest of Britain.
It originally had “confides” (i.e. is confident) in place of “expects”, but the latter was used because it was part of the signal-flag vocabulary and therefore would not need to be spelled out letter by letter.
In principle, organ transplantation would allow individual organs to live longer than any particular human. It appears the oldest transplanted organ is currently 101 years old, residing in a 68 year-end woman, and so much younger than the (organs inside the) oldest human, at 122 years old. (Tissue like corneas have occasionally come from centenarian donors, but not a full organ like a liver or kidney, to my knowledge.) The woman would need to live to 90 without kidney failure, quite a feat, to open up this new category of longevity.
Relatedly, here is Ezra Klein interview of Cal Newport on social media and the cognitive effects of distraction. Newport’s way of thinking about this is the one that most accords with my own.
Google Scholar releases (evil?) list of metrics for top journals. Interestingly, the #14 highest impact journal is Angewandte Chemie, the German journal of chemistry, although this is related to the fact that is publishes reviews.
Programming-by-example in Microsoft Excel:

It’s funny that this is targeted at “the 99% of people who don’t know how to program”; I know the basics of how to program, but learning the syntax of the Excel macro language would take too long to save time on net with that sort of automated solution. (Even if I am put in this situation multiple times over the years, it’s still not worth investing in learning the syntax because there’s enough time between uses that I’ll forget and have to re-learn. Despite having written several macros in the past, the only function I can remember is “SUM”.)
Ironically, one way I might solve this problem in the past is by “coding by example”, i.e., grabbing code snippets or boilerplate off the internet and learning enough syntax by viewing usages (rather than having anything defined).
CEA took grant applications for a broad range of projects including people who want to do community organizing full time. I believe they will be doing this again in the future.
William Hamilton’s original two papers where he laid down the foundation of Hamiltonian mechanics, typeset cleanly in PDF from modern, hand-corrected TeX files. See also a longer list of his publications. (H/t Godfrey Miller.)
Also from Godfrey: Colorado finally implements an idea whose praises I’ve been singing for years: avoid perverse incentives by sending government fines to be donated to charity rather than funding the police who issue the fines. It’s an exemplar of how broken our policy methods are that such a blindingly obvious fix with no equilibrium downsides is not even discussed, much less implemented. In this sense it is similar to the fact that schools do not use spaced repetition for rote memorization. The silver lining is that this ought to be a ratchet; what police department could ask that it gets to keep money earmarked for charity?
Laydown delivery
…is a mode of attack using a freefall nuclear weapon: the bomb’s descent to the target is slowed by ribbon parachute so that it actually lands on the ground before detonating.[1] Laydown delivery requires the weapon’s case to be reinforced so that it can survive the force of impact and generally involves a time-delay fuze to trigger detonation e.g. 45 seconds after hitting the ground. Laydown mode can be used to increase the effect of the weapon’s blast on built-up targets such as submarine pens or to transmit a shock wave through the ground to attack deeply-buried targets.
Jeff Kaufman is looking into AI risk as a cause area for individual technical EAs. See his posts since July 3rd.
Also from Jeff is this explanation of why sniping makes sense in auctions with proxy bidding like eBay. He doesn’t explicitly explain why people feel bad after they are sniped, but it follows directly from the main idea: the snipe bid is giving you additional information about how valuable the item is, which you didn’t have before; your previous max bid may no longer reflect your new assessment, but it’s too late to act.
Short post by Scott Alexander on plausible deniability in social interactions. I’m not (diagnosed) autistic, but I constantly make the sort of mistakes ascribed here to autistics. (Of course, the lead example is not nearly so clear cut; see comments.) In my experience, I have been able to improve somewhat through conscious study, but this only works well if I get a lot of practice or if there’s a simple theory/rule I can follow and build on (e.g., “ask people about themselves, and foster in yourself a genuine interest about them by finding the most interesting thing they can teach you” a la Dale Carnegie). Most of the time it doesn’t become habit, and I find myself slipping up if I haven’t seen the situation in a while, only it’s even more painful because I can recognize my mistakes in retrospect.
The United States intelligence community has its own set of three classified wikis corresponding different levels of classification, which can only be accessed with the proper credentials. It also has an interestingly different philosophy in that non-neutral-point-of-view opinions are encouraged.
This hierarchy of energy sources as a framework for understanding the evolution of life was linked by several people, and is tempting to physicists since it draws on something (energy conservation, negentropy) they have a gut instinct for, but it didn’t make sense to me in parts. Why is “fire” so important if cooking food probably increases calories absorption less (or much less) than 50%. Farming seems more important than fire, and presumably metabolic pathways and other detailed chemistry constraints are very often more important than energy constraints (though not always).
Never gets old:
On Gough Island, “Giant carnivorous mice threaten world’s greatest seabird colony“. Gruesome. There are plans to eradicate the mice by 2021.
Behavioural individuality in clonal fish arises despite near-identical rearing conditions
David Bierbach, Kate L. Laskowski, and Max Wolf
Behavioural individuality is thought to be caused by differences in genes and/or environmental conditions. Therefore, if these sources of variation are removed, individuals are predicted to develop similar phenotypes lacking repeatable individual variation. Moreover, even among genetically identical individuals, direct social interactions are predicted to be a powerful factor shaping the development of individuality. We use tightly controlled ontogenetic experiments with clonal fish, the Amazon molly (Poecilia formosa), to test whether near-identical rearing conditions and lack of social contact dampen individuality. In sharp contrast to our predictions, we find that (i) substantial individual variation in behaviour emerges among genetically identical individuals isolated directly after birth into highly standardized environments and (ii) increasing levels of social experience during ontogeny do not affect levels of individual behavioural variation. In contrast to the current research paradigm, which focuses on genes and/or environmental drivers, our findings suggest that individuality might be an inevitable and potentially unpredictable outcome of development.
(H/t Kaj Sotala.)
A nice narrative about the historical discovery that air is not weightless and vacuums do not pull.
Wikipedia: “The following 1953 proof by Dov Jarden has been widely used as an example of a non-constructive proof since at least 1970”:
A Simple Proof That a Power of an Irrational Number to an Irrational Exponent May Be Rational.
${\sqrt {2}}^{\sqrt {2}}$ is either rational or irrational. If it is rational, our statement is proved. If it is irrational, $\left({\sqrt {2}}^{\sqrt {2}}\right)^{\sqrt {2}}=2$ proves our statement.
Sarah Constantin has preliminary thoughts on medical regulatory arbitrage.
This article on decoherence, Darwinism, and Wojciech Zurek was sent to me because I am mentioned briefly. It’s a surprisingly faithful popular description of why we think these topics are important for understanding why the macroscopic world appears as it does. (Of course, there are professionals who disagree, and this article can’t equip an outsider to form an informed opinion on the dispute.) The only major disagreements with the article are those I have with Wojciech, who I’m sure was the primary source.
Package sorting facility in China. It’s hard for me to guess the requirements and constraints that produce a solution like this.
Thermodynamical timescales actually gives you a reason to use a double log plot.
The restrictions on ivory has driven a market in fossilized tusks of woolly mammoths and horns of woolly rhinos, extinct and dug out of Siberia.
Good HN discussion of Etherium.
Roomba creator is kickstarting a robot garden weeder. (H/t Keller Scholl.)
Crash-testing for pet car restraints:

(H/t The Wirecutter.)
Although the craters and peaks of the moon are only about ~1% of its radius, the umbra of the shadow cast by the moon during a solar eclipse greatly magnifies the distortions so that the area of totality deviates from an ellipse by ~20%.
Geometric explanation for the origin of a sequence of sum relations between integer squares, with the pythagorean triple $3^2 + 4^2 = 5^2$ as a special case.
Alex Tabarrok describes his old but new-to-me idea of dominant assurance contracts, a proposed method for solving public good problems using some (fragile) game theory rather than by law. But here are some excellent worries from Anonymous:
After Alex’s last post about the topic, I became very enthusiastic about dominant assurance contracts. However, as long as the idea is not empirically tested, I can’t put too much trust in them. Would people really contribute to DACs either as entrepreneurs implementing the contracts or “customers” (or whatever they would be called) paying for the project? They might think that there is something fishy in the contract (“why would someone give me free money?”); they might think that it’s a problem that some people can free-ride (even if it’s not a problem from the economical point of view) and not contribute because of that (people can sometimes reject favorable propositions if they think that someone else benefits “unfairly” from it).
One more practical obstacle that I thought about is this: suppose a DAC is implemented in some platform that is perhaps similar to Kickstarter. If it looks like the contract is not going to be accepted by enough people (but is somewhat close to it), the entrepreneur can himself contribute the rest of the money in order to avoid paying the failure fee. The escrow service might try to prevent this by not accepting money from the entrepreneur himself, but it wouldn’t be too difficult find friend, family, etc. to do this for the entrepreneur. If the public knows that the entrepreneur always has the option of contributing the rest of the money himself and thus avoid failure, a DAC becomes just a normal assurance contract.
GiveWell and Open Philanthropy Project complete their organizational separation. Note that OpenPhil has chosen to become an LLC rather than a non-profit. This doesn’t mean they won’t investigate and recommend non-profits, nor that donations to their recommended non-profits will fail to enjoy the normal tax deduction. It just means that money funding the investigative work will not be tax advantaged. This was done to allow them to also investigate for-profit organizations as giving opportunities.
This is all sensible, although I was disappointed that they confirmed they may indeed be recommending the support of political candidates.
OpenPhil gave $20M to develop and prepare for Malaria control through gene drives.
It’s sort of crazy that the esophagus is behind the trachea since the nose is, topologically, behind the mouth.

Getting food from the mouth to the stomach without any getting into your lungs is an amazing feat. Here are videos of a person swallowing in MRI and X-ray.
Stench gas:
Stench gas warning systems give miners the signal to head to shelter during serious emergencies.
Deep underground in a noisy mining environment, it can be very difficult to quickly raise an emergency alarm. It is often very noisy. It may be very dark or there may be many bright lights, or both in turn. There may be no source of electric power. Workers may be isolated from each other, spread over large distances, and separated by thick rock. But there will always be air (mines work very hard to make sure of this) and that air allows mines to use a very interesting way to communicate: Stench gas.
Ethyl Mercaptan has a very distinct, highly unpleasant odor that humans can easily detect and recognize, even when it is present in small amounts. Mine alarm systems (remote or manual) can release the gas, and a non-flammable propellant, into the ventilation system. It spreads rapidly through a large volume of air. When the smell is detected, miners immediately report to designated shelter areas. They remain there until an all-clear is given, which may include a distinct all-clear scent such as wintergreen.
Tubes vs pyramids as a cartoon for professional filtering hierarchies.
Topoisomera is an enzyme that breaks and re-bonds DNA strands to remove knots and excess torsion.
There is some evidence that cystic fibrosis is a genetic adaptation to tuberculosis in Northern Europeans. Similarly to sickle-cell anemia as an adaptation to Malaria, cystic fibrosis is a potentially deadly disease when both copies of a mutated gene are present in an individual but may be fitness enhancing in individuals with a single copy.
When do start-up stock options as compensation make sense?
Some evidence that desirability bias trumps confirmation bias, at least with regard to political polls. (H/t Giego Caleiro.)
HD GoPro video with a 3rd-party view of stage separation for a sounding rocket:

(H/t Will Riedel.)
Frequent batch auctions are an alternative mechanism for stock exchanges. See popular and academic discussion (H/t Matt Levine, who recently did an AMA.) The basic idea is to hold regular (or stochastic) auctions on a timescale that is at least as fast as human decision makers would ever need price information but slow enough that electronic switching and the laws of physics are not relevant factors in market making.
Eric Posner: Twenty theses about Twitter (also from Levine). If I’m asked “Jess, why do you hate people?” over beers, part of the answer goes like this: In the ancestral environment human minds were devoted to solving lots of diverse problems, but in modern rich countries they have essentially all material needs fulfilled. In this environment, the human mind focuses strongly on zero-sum competitions for social and romantic success; these fights are, at least for now, not dramatically reduced by our vast resources and technology.^a Occasionally these desires can be successfully steered toward valuable endeavors by linking positive-sum productivity to status — indeed, this is how civilization is built — but the institutions that do this are fragile and slow to adapt. Facebook and Twitter are sometimes said to be “corrupting” or “commandeering” our social impulses, but my Hansonian view is that they are basically just revealing our natural but previously hidden disposition.
Video of a mosquito’s proboscis searching within skin for a blood vessel.
From Sean Carroll: LISA will fly and the continuing evolution and supremacy of Google Maps.
The United States is so rich that it consistently produces more bread than can be eaten, even when given away for free to the needy. Here is a description from one food bank:
Depending on the size of their household people were given number of points each month that they could “spend” on food. Everything was given price in these points. For example a can of corn might be 1 point, whereas a gallon of milk might be 3. The only thing not given a points price was bread. Bread was free and you could always take as much bread as you wanted…We always had bread left over after we got all the people through the building.
Obviously bread alone cannot make a well-balanced or enjoyable diet, and food banks are valuable for more than just providing bare calories.
The motivation for the quirky rule in baseball about dropped third strikes is unclear, and may go back even as far as the 18th-century origins of the game.
Juno’s retreat from perijove. Stunning:
https://vimeo.com/219148572
Note that this is a partial rendering. This earlier version makes it more clear what’s going on. Every flicker is a new image. The trajectory is an interpolation of these photos taken at various points in the orbit. Not as honest as I was hoping, but still spectacular and reasonably grounded in true images. (H/t Jana Grcevich.)
City subway maps vs. their actual geography.
Katja Grace and collaborators have released their much-anticipated survey of expert predictions for times at which artificial intelligence systems will be able to supplant humans at various tasks.
When Will AI Exceed Human Performance? Evidence from AI Experts
Katja Grace, John Salvatier, Allan Dafoe, Baobao Zhang, Owain Evans
Advances in artificial intelligence (AI) will transform modern life by reshaping transportation, health, science, finance, and the military. To adapt public policy, we need to better anticipate these advances. Here we report the results from a large survey of machine learning researchers on their beliefs about progress in AI. Researchers predict AI will outperform humans in many activities in the next ten years, such as translating languages (by 2024), writing high-school essays (by 2026), driving a truck (by 2027), working in retail (by 2031), writing a bestselling book (by 2049), and working as a surgeon (by 2053). Researchers believe there is a 50% chance of AI outperforming humans in all tasks in 45 years and of automating all human jobs in 120 years, with Asian respondents expecting these dates much sooner than North Americans. These results will inform discussion amongst researchers and policymakers about anticipating and managing trends in AI.
Grace on the apparent discrepancy between predictions for human-level AI and AI automation of all jobs:
It is an obvious case of massive framing bias I think: the definitions mean that HLMI should be strictly later than labor being fully automatable. It might be less clear whether anyone in particular made errors, because the questions weren’t necessarily answered by the same people.
(H/t Alyssa Vance.)
Tolstoy:

The distinction that really matters is not between violence and non-violence, but between having and not having the appetite for power. There are people who are convinced of the wickedness both of armies and of police forces, but who are nevertheless much more intolerant and inquisitorial in outlook than the normal person who believes that it is necessary to use violence in certain circumstances. They will not say to somebody else, ‘Do this, that and the other or you will go to prison’, but they will, if they can, get inside his brain and dictate his thoughts for him in the minutest particulars. Creeds like pacifism and anarchism, which seem on the surface to imply a complete renunciation of power, rather encouraged this habit of mind. For if you have embraced a creed which appears to be free from the ordinary dirtiness of politics — a creed from which you yourself cannot expect to draw any material advantage — surely that proves that you are in the right? And the more you are in the right, the more natural that everyone else should be bullied into thinking likewise.
(H/t Paul Sutter, foxhedgehog, and hitekker.)
Photos from China’s lunar rover.
The OpenTimeStamping for the Internet Archive.
I disagree with the general thrust of Tyler Cowen responding to Noah Smith’s claim that the existence of “two good papers” from a field is a reasonable rule of thumb for deciding whether a vast literature should be expected reading. I’m sure you can pick two ethics papers that can convince a reasonable outsider that the field has something significant and convincing to say, and if you couldn’t then that would be damning. (I suspect Cowen, who is old and well-read in this field, just can’t remember the last time an ethics paper changed his mind.) I agree with Cowen on the value of review papers, but as the update on Smith’s post emphasizes this is not the point of the rule.
This is what the top of Everest looks like:
Surprising (to me) fact about the unusually high costs in the US of public works: land acquisition costs are relatively low in the US compared to some other developed countries like Japan. Costs in the US are driven primarily by construction (not, e.g., the somewhat higher population density in Manhattan), perhaps due to labor regulation. (H/t Alyssa Vance.)
Also from Vance: Gregory Clark on what caused the industrial revolution. (Hint: we don’t know.)

As I said, the mystery of the industrial revolution is that the majority of the innovation was made by people who had no formal connection to science and no formal scientific education, and who were just instead these kind of tinkerers and mechanics, and really had no access to that kind of high level scientific knowledge. Again, it is very hard to disprove that this wasn’t connected to high level scientific facts, but it is very hard to show that there really was any strong connection

How can it be that something this enormous taking place in relatively recent history is not understood at even a basic level? I wonder to what extent the curse of knowledge is at play, where people learn things but can’t even really notice it.
I didn’t realize that with all the advances in private launch services, it was only just recently that the first launch from a private launch pad took place. Gizmodo describes Rocket Lab‘s 56-foot, 150-kg-payload rocket as “adorable“.
The Joeveo “Temperfect” mug has a sealed inner lining with a large thermal mass that melts at 140°F. When you pour in hot coffee (e.g., 170°F), it melts the liner and brings the temperature of the coffee down to the melting point. As the coffee leaks heat, the heat of fusion in the liner keeps the coffee temperature constant until the liner is completely solid again. The maker claims this plot is a real, empirical temperature trajectory:
(Source.)
No idea if it actually works, but the idea is just so elegant.
(It started as a kickstarter in 2013, is just now shipping mugs, and are still filling backorders.)
Some quadriplegics who require mechanical ventilation can, as an alternative, get diaphragm pacing, where their phrenic nerve is electrically stimulated directly to induce contraction of the diaphragm. Incidentally, iron lungs still exist in the form of negative pressure cuirasses.
valuearb: “5M years ago parts of the mediterranean were 6,000 to 15,000 feet below sea level.” (By comparison, the lowest points on Earth today is the shore of the Dead Sea, at 420 meters (1380 feet) below sea level.) This was the Messinian salinity crisis, when the Mediterranean sea almost completely evaporated. When the Atlantic re-flooded the Mediterranean, the local sea level may have risen 10 meters per day. The potential future reoccurrence of this flood is the subject of XKCD’s “Time” magnum opus.
I knew that the various stripes and swirls on Jupiter were temporary, and the appearance of the planet was unstable on long timescales. The Great Red Spot has changed shape and size over the 350 years it has been observed. However, I did not know that entire (Earth-width) stripes can disappear over just one year. Note though that Glenn Orton of NASA’s Jet Propulsion Lab “thinks the belt is not actually gone, but may be just hiding underneath some higher clouds.”
philipkglass on near-infrared vision in animals:
NIR vision appears to be entirely absent from vertebrates….See “The Verriest Lecture 2009: Recent progress in understanding mammalian color vision”
https://pdfs.semanticscholar.org/c9df/0b61e4a45e10577c001513a4fc1432696b3c.pdf…
…Vertebrate photopigment sensitivity drops off rapidly toward the near IR (~700 nm).
I found this fascinating related (but not peer reviewed) document on arXiv, “Did Evolution get it right? An evaluation of near-infrared imaging in semantic scene segmentation”
https://arxiv.org/pdf/1611.08815.pdf
The author performs semantic segmentation with convnets using conventional visual-spectrum images and compares performance with images spectrally extended into the near-infrared. He finds that adding the additional NIR band did not improve task performance — hence that evolution “got it right” by not expanding vision into a wavelength region that fails to improve semantic segmentation. I’m not sure how good this work is from a professional biologist’s perspective, but I found it damn interesting.
This psychological experiment didn’t get as much press as the Milgram study or the Stanford prison experiment, but I found its mistreatment of its subjects even more disturbing. Oh, and it may have played a role in creating the Unabomber. Possibly relevant to recent discussion of loosening of IRB rule.
How hens brood.
Jurisdiction shopping significantly curtailed by 8-0 US Supreme Court ruling. Prior to the ruling: “More than 40 percent of all patent lawsuits are filed in East Texas.” Amazing to me that the hugely impactful precedent was set in 1990, and then was unanimously overruled.
The forthcoming Boeing 777X will have twin engines, each with the cross-section of a 737, and folding wingtips, which add 7 additional meters of wingspan and fold in 20 seconds:
https://youtu.be/VSBct_G1iNo?t=38s
Therac-25 as a classic case study in software design safety. A race condition led to patients receiving fatal doses of radiation.
Von Braun’s Mars mission design from circa 1950.
GPS darts as alternative to high-speed pursuit.
How to design a rotating house:
“I’ve had the strangest questions about the waste water system,” he says. “Even things like: ‘do you have to wait for the house to spin so that the pipes line up underneath before you can flush the toilet!’ It’s funny how people think. All the plumbing services run under the flooring back to a 100mm swivel joint from a broad acre farm walking irrigation boom arm.
“This fitting houses all the waste water pipes and provides a water tight seal. Because the house rotates, it was important to find something that was designed to spin and not leak. This is perfect.
“It wasn’t that hard to find one, once I asked the right people what they thought would do the trick. Being in the bush you can often find people who are good at finding solutions.”
Uber improves price discrimination. Nothing wrong with this, but emphasizes how important it is to have a viable Uber competitor.