Links for May 2017

Claims about what makes Amazon’s vertical integration different:

I remember reading about the common pitfalls of vertically integrated companies when I was in school. While there are usually some compelling cost savings to be had from vertical integration (either through insourcing services or acquiring suppliers/customers), the increased margins typically evaporate over time as the “supplier” gets complacent with a captive, internal “customer.”

There are great examples of this in the automotive industry, where automakers have gone through alternating periods of supplier acquisitions and subsequent divestitures as component costs skyrocketed. Divisions get fat and inefficient without external competition. Attempts to mitigate this through competitive/external bid comparison, detailed cost accountings and quotas usually just lead to increased bureaucracy with little effect on actual cost structure.

The most obvious example of Amazon’s SOA structure is Amazon Web Services (Steve Yegge wrote a great rant about the beginnings of this back in 2011). Because of the timing of Amazon’s unparalleled scaling — hypergrowth in the early 2000s, before enterprise-class SaaS was widely available — Amazon had to build their own technology infrastructure. The financial genius of turning this infrastructure into an external product (AWS) has been well-covered — the windfalls have been enormous, to the tune of a $14 billion annual run rate. But the revenue bonanza is a footnote compared to the overlooked organizational insight that Amazon discovered: By carving out an operational piece of the company as a platform, they could future-proof the company against inefficiency and technological stagnation.

In the 10+ years since AWS’s debut, Amazon has been systematically rebuilding each of its internal tools as an externally consumable service. A recent example is AWS’s Amazon Connect — a self-service, cloud-based contact center platform that is based on the same technology used in Amazon’s own call centers. Again, the “extra revenue” here is great — but the real value is in honing Amazon’s internal tools…

The key advantage that Amazon has over any other enterprise service provider — from UPS and FedEx to Rackspace — is that they are forced to use their own services. UPS is a step removed from backlash due to lost/destroyed packages, shipping delays, terrible software and poor holiday capacity planning. Angry customers blame the retailer, and the retailer screams at UPS in turn. When Amazon is the service provider, they’re permanently dogfooding. There is nowhere for poor performance to hide. Amazon has built a feedback loop as a moat, and it is incredible to watch the flywheel start to pick up speed.

So is it the end of an era? We may still use MP3s, but when the people who spent the better part of a decade creating it say the jig is up, we should probably start paying attention. AAC is indeed much better — it’s the default setting for bringing CDs into iTunes now — and other formats are even better than it, though they also take up mountains of space on our hard drives.

(That story also, incredibly, fails to mention that the patent expired.) The broader point, of course, is that the news is filled with these sorts of submarine stories, just generally a bit more concealed.

In mathematics, the Mertens conjecture is the false statement that the Mertens functionM(n) is bounded by √n, which implies the Riemann hypothesis.

It was later shown that the first counterexample appears below () (Pintz 1987) but above (100 trillion).

For more than 7,000 hours over the past five years, Mitchell has slowly exposed the fossil’s skin and bone. The painstaking process is like freeing compressed talcum powder from concrete. “You almost have to fight for every millimeter,” he says.

Recently unveiled in an Alberta museum. Michael Greshko:

Hey, I’m the journalist who wrote the story. A few things: – The organic-rich film preserving the outlines of scales (so, yes, fossilized skin) is only a few millimeters thick. So Mitchell had to prepare the fossil extremely slowly in order to follow the film through the matrix. – The half-life of DNA is ~521 years at 13.1°C, as found in this 2012 paper: http://rspb.royalsocietypublishing.org/content/279/1748/4724 The team’s model predicts higher half-lifes at truly freezing temperatures, but even at the extreme end, there’s no way DNA would survive 110 million years. – The dating on the site is well constrained to ~110 million years old. The fossil was found in the Wabiskaw Member of the Clearwater Formation, a well-dated rock formation in Alberta. The underlying oil sands have been radiometrically dated to 112±5.3 million years old. (http://science.sciencemag.org/content/308/5726/1293)

In one example of what can happen when prices are widely known, Germany required all gas stations to provide live fuel prices that it shared with consumer price-comparison apps. The effort appears to have boosted prices between 1.2 to 3.3 euro cents per liter, or about 5 to 13 U.S. cents per gallon, according to a discussion paper published in 2016 by the Düsseldorf Institute for Competition Economics.

Plausibly, the key fact is that the systems — “AI”s, although not really — gather lots of historical information about how their competitors react to price changes, opening a potential line of communication between the AIs. This makes it feasible, and perhaps likely, that the AI’s may settle on a game-theoretic equilibrium where they both “agree” to keep prices high because some earlier tit-for-tat behavior. It would be great to see a toy model of this explicitly. Here was another neat anecdote:

One client called to complain the software was malfunctioning. A competitor across the street had slashed prices in a promotion, but the algorithm responded by raising prices. There wasn’t a bug. Instead, the software was monitoring the real-time data and saw an influx of customers, presumably because of the long wait across the street.

“It could tell that no matter how it increased prices, people kept coming in,” said Mr. Derakhshan.

The things to note are (1) Self-sufficient colonies in Antarctica or under the ocean would be much easier than Mars. (2) Many of the key technologies of our civilization only make sense at scale (100’s of millions of humans) and with access to cheap resources (e.g., water). The economy is highly interdependent so that each resource you take away forces you to solve many surprising new problems. (Most of these things are heavy and cannot be continuously shipped from Earth to Mars.) (3) Humans do not go places because it’s a great story, they largely go there for economic rewards, and in particular, they rarely leave places they are wealthy to go to places where they become dirt poor.

The Vikings had the transportation technology to get to the New World many centuries before there were persistent European colonies, and this was in a time when the economy was much less dependent on the cooperation of large numbers of humans and many kinds of resources from diverse places. Our current era is much more analogous to 900 AD than 1492.

In 2005, Wikipedia co-founder and Wikimedia Foundation [WMF] founder Jimmy Wales told a TED audience:

So, we’re doing around 1.4 billion page views monthly…. And everything is managed by the volunteers and the total monthly cost for our bandwidth is about US$5,000, and that’s essentially our main cost. We could actually do without the employee … We actually hired Brion because he was working part-time for two years and full-time at Wikipedia so we actually hired him so he could get a life and go to the movies sometimes.

According to the WMF, Wikipedia (in all language editions) now receives 16 billion page views per month. The WMF spends roughly US$2 million a year on Internet hosting and employs some 300 staff. The modern Wikipedia hosts 11–12 times as many pages as it did in 2005, but the WMF is spending 33 times as much on hosting, has about 300 times as many employees, and is spending 1,250 times as much overall. WMF’s spending has gone up by 85% over the past three years.

Sounds a lot like cancer, doesn’t it? For those readers who were around three years ago, did you notice at the time any unmet needs that would have caused you to conclude that the WMF needed to increase spending by $30 million dollars? I certainly didn’t.

The large majority of funds donated to WMF are not spent on hosting, overhead, or other things necessary for the normal function of the site. See the HN comments for a rebuttals to the author’s claim that WMF’s software development team lacks a roadmap and goals/specifications (although pretty much everyone seems to agree the software output has been subpar).

As of 15 August 2016, the CAPS [Cirrus Airframe Parachute System] has been activated 83 times, 69 of which saw successful parachute deployment. In those successful deployments, there were 142 survivors and 1 fatality. No fatalities, unsuccessful deployments, or anomalies (with the exception of one that is still under investigation) have occurred when the parachute was deployed within the certified speed and altitude parameters. Some additional deployments have been reported by accident, as caused by ground impact or post-impact fires, and 14 of the aircraft involved in CAPS deployments have been repaired and put back into service.

Qualifies as a great landing.

Conventional geothermal is about 4-5 cents/kWh, on par with natural gas in the US. The dominant capital cost is drilling a large well (you need volume), and so geothermal plants are generally only built in areas that require shallow (1km) wells.

Well costs are ~quadratic in depth. Given how much money has already been spent optimizing drilling for the oil&gas industry, along with how cutthroat that market is, I don’t see the cost coming down significantly. As a result, deep geothermal will likely be limited to niche regions like Iceland. And you need deep geothermal to scale it past the existing locations.

I would love to be wrong, since geothermal checks all the boxes for renewables and is also suitable for base load power, but I don’t see an obvious path forward short of a drilling tech miracle.

“I think it’s fair to say…that our understanding of the worm has not been materially enhanced by having that connectome available to us. We don’t have a comprehensive model of how the worm’s nervous system actually produces the behaviors. What we have is a sort of a bed on which we can build experiments—and many people have built many elegant experiments on that bed. But that connectome by itself has not explained anything.”

It is, however, a very useful starting point

In the 1980s, as a postdoctoral student in Brenner’s lab, Martin Chalfie—now at Columbia University—used the C. elegans wiring diagram to explain one of the worm’s behaviors: He identified the specific neural circuits responsible for the worm’s tendency to wriggle backward when poked on the head and to squirm forward when touched on the tail. “The connectome was absolutely critical,” Chalfie says. “Without it, we simply would not have known which cells were connected to which.” By combining the wiring diagram with evidence from previous research, Chalfie predicted that a particular set of interneurons mediated forward movement and that another was involved in backward movement. Annihilating those neurons with lasers confirmed his predictions.

Check out the worm atlas.

As part of our research on the history of philanthropy, I recently investigated several case studies of early field growth, especially those in which philanthropists purposely tried to grow the size and impact of a (typically) young and small field of research or advocacy.

The full report includes brief case studies of bioethics, cryonics, molecular nanotechnology, neoliberalism, the conservative legal movement, American geriatrics, American environmentalism, and animal advocacy. My key takeaways are:

• Most of the “obvious” methods for building up a young field have been tried, and those methods often work. For example, when trying to build up a young field of academic research, it often works to fund workshops, conferences, fellowships, courses, professorships, centers, requests for proposals, etc. Or when trying to build up a new advocacy community, it often works to fund student clubs, local gatherings, popular media, etc.
• Fields vary hugely along several dimensions, including (1) primary sources of funding (e.g. large philanthropists, many small donors, governments, companies), (2) whether engaged philanthropists were “active” or “passive” in their funding strategy, and (3) how much the growth of the field can be attributed to endogenous factors (e.g. explicit movement-building work) vs. exogenous factors (e.g. changing geopolitical conditions).

1. A major institution (e.g. Vanguard) holds index funds, but the shareholder voting for each company in those funds is done by a different in-house team whose compensation is tied to the performance of that particular company. (More realistically, you’d just need a different team for each company in a given industry, but one team could vote for multiple companies that don’t compete with each other.) It’s “passive investing with active shareholder governance” in the sense that only voting, not dollars, are actively directed.
2. Accelerate the trend of issuing two classes of stock, with and without voting rights. Only active investors and corporate raiders bother to pay a premium for influence.

Existing gasoline-powered vehicles may be converted to run on CNG [compressed natural gas] or LNG [liquid natural gas], and can be dedicated (running only on natural gas) or bi-fuel (running on either gasoline or natural gas). Diesel engines for heavy trucks and busses can also be converted and can be dedicated with the addition of new heads containing spark ignition systems, or can be run on a blend of diesel and natural gas, with the primary fuel being natural gas and a small amount of diesel fuel being used as an ignition source.

The share of hired crop farmworkers who were not legally authorized to work in the U.S. grew from roughly 15 percent in 1989-91 to almost 55 percent in 1999-2001. Since then it has fluctuated around 50 percent. Since 2001, the share who are citizens has increased from about 21 percent to about 33 percent, while the share who hold green cards or other forms of work authorization has fallen from about 25 percent to about 19 percent.

adulthood is emailing “sorry for the delayed response!” back and forth until one of you dies’

(H/t Rob Wiblin.)

Delope…is the practice of throwing away one’s first fire in a pistol duel, in an attempt to abort the conflict. According to most traditions, the deloper must first allow his opponent the opportunity to fire after the command (“present”) is issued by the second, without hinting at his intentions…

The delope could be attempted for practical reasons, such as if one duelist thought their opponent was superior in skill, so as not to provoke a fatal return shot. Deloping could also be done for moral reasons if the duelist had objections to attempting to kill his opponent or if he were so skilled a marksman as to make the exchange unfair. Deloping in a duel, for whatever reason, could be a risky strategy whether or not the delope was obvious to all present. Deloping with a near miss, in order to save one’s honor without killing, could backfire if the opponent believed the effort to be genuine and responded with a fatal shot. Also, regardless of whether the delope was near or wide, the opponent might infer that he was being insulted as “not worth shooting” (an unworthy opponent) and either take care to aim his own shot to kill or insist on a second exchange.

However, for the opponent to insist upon a second shot after a delope was considered bloodthirsty and unbecoming. Often, it would fall to the seconds to end the duel immediately after a delope had been observed.

The term delope is specific to the use of firearms in a duel which, historically speaking, were typically flintlock pistols. These pistols were notorious for their lack of accuracy at long distances and a particularly skilled marksman might attempt to delope unnoticed with a well-placed “near-miss.” The distance between the two combatants had to be great enough that all others present would assume that any miss was due to this inherent inaccuracy and not intentional. This way the shooter could avoid killing his opponent and, if accused of deloping, claim he had made a genuine effort. Also, the opponent might recognize the “near-miss” as a delope but understand that it was meant for the benefit of any witnesses present and, if the opponent was not insulted, also delope. Both parties could then claim they had each tried to shoot the other and the duel would end without any fatalities.

And this is just one aspect of the actual duel itself. Just imagine the amount of additional social structure that would arise in societies with dueling.

I think my favorite book on management is “High Output Management” by Andy Grove, which is kind of the Bible of how do you scale an organization. I have a very long list of recommendations. “The Effective Executive” by Peter Drucker is really good. But basically what you’re talking about is, you need to have good judgment about a lot of different things. What that really is saying is, how do you gain wisdom really quickly? Last book recommendation, “Poor Charlie’s Almanack” by Charlie Munger. This is probably the best book on that that I’ve read, which is really about how do you build a bunch of mental models to break down the complexity of the world and know when to use each.

There is a ton of room for innovation around tunnel boring machines built by the major manufacturers (primarily Herrenknecht and Robbins), around reliability, durability, ease of serviceability etc. I cannot emphasize enough how modern TBM’s require an unbelievable amount of engineering attention, repair labor, spare parts infrastructure etc, similar to many super-early-stage fragile prototype technologies. Unfortunately TBM’s are no longer early stage, but for some reason the technology is frozen at just-good-enough-to-barely-work.”

However I don’t know if the economics will work out to fix any of these. Here are a couple of the big problems:

• Most TBM’s are semi- or fully-customized for a single job. This raises machine costs. It’d be better if there were only a small range of small-medium-large TBM’s that work ~everywhere.
• Most TBM’s are fully assembled in the factory, smoketested aka turned on to make sure they work, disassembled and shipped to the job, then reassembled and used. This is not efficient, surely we can figure out a better way.
• Most TBM’s are entombed aka thrown away at the end of the job, because getting them out of the hole is expensive and difficult.
• Changing the cutterheads is labor intensive and dangerous and requires highly trained very expensive humans, and it’s slow. While you change the cutterheads, your billion dollar toy is sitting there doing nothing.
• TBM architecture is highly dependent on geology. A slurry faced TBM that works in mixed soils is a totally different beast from a hard-rock TBM. It would be cool to have one machine that works in many geologies, perhaps with minimal or automated modifications.
• TBM’s require lots of care and feeding from a small army of humans. This raises job costs.
• Topside support infrastructure such as slurry plants and ground freezing machinery comes from different vendors, often even from different countries. E.g. it’s common to buy your topside slurry plant from the MS company, in France, while your ground freeze vendor might be Tachibana from Japan. Often each subsystem’s engineers on site literally don’t even speak a common language. Hilarity predictably ensues. Vertical integration would pay huge dividends here.

Ideally Elon can mass-produce TBM’s that just work out of the box for most jobs, and that are easier to work on. Then we can laugh him out of town for his stupid “put cars in tunnels” ideas and use his miracle machines to build sensible train tunnels.

Lots of parallels with the rocket industry.

My suggestion, on which I have much to write in the future, is that theorists should get their house in order: distill what is known in a single authoritative place. It should incorporate the best mechanisms from the Review of Particle Physics, Reviews of Modern Physics, the Stanford Encyclopedia of Philosophy, Wikipedia, and GitHub, and would require developing much more powerful tools for collaboration and dispute resolution than we have now, eventually with great value to other fields.

…how good is the social science in this area? I would say “not so great.” Try looking for good public choice treatments of how climate intentions end up translated into climate policy. That is a remarkably important question, and yet it is understood poorly.

Or “how many of the people who make proclamations in this area have a decent understanding of Chinese energy and climate policy?”, and the answer is hardly any, even though that may be the most important topic in the area. And I ask that question not only of the casual tweeters but also of the academics who work on climate change.

(Pilfered from here)

The difference from suspension bridges:

In a suspension structure such as the Golden Gate Bridge primary cables are strung from tower to tower and secondary cables drop down from those to hold the roadbed in place. Cable-stayed bridges, by contrast, have cables that run directly from the tower to the road. They essentially eliminate the cables between towers.

But the longest spans still go for suspension:

Cable-stay is inadvisable for bridges with a main span longer than 915 meters because the towers would have to soar twice as high as the towers of a suspension bridge of the same length to string enough cables to hold the road deck in place. For very long bridges, traditional suspension wins out.

(H/t ghaff.)

A position of Huygens’ landing site on Titan was found with precision (within one km – one km on Titan measures 1.3′ latitude and longitude at the equator) using the Doppler data at a distance from Earth of about 1.2 billion kilometers.

When people lament the demise of communities or multi-generation family units in the United States, this is the kind of mutual support they’re thinking of. The extent to which America was once comprised of warm, child-raising villages in its real-life past is, of course, greatly exaggerated, and we certainly shouldn’t romanticize local communities per se: they always have the capacity to be meddling, oppressive, and exclusionary. But all communities don’t have to be like that, and instead of abdicating community ideals as outdated, we could be working to realize them better in the particular places we live. As American lifestyles become increasingly mobile and rootless, close involvement in a community may not be foremost on people’s minds; to the extent that people these days talk about “settling down” somewhere, they usually seem to be thinking in terms of sending their kids to a local school, patronizing nearby restaurants, and attending summer concerts in the park, not trundling around to people’s homes and asking what they can do for them.

But even if we aren’t planning to live in the same town for the entire rest of our lives, we mustn’t allow ourselves to use this as a convenient excuse to distance ourselves from local problems we may have the power to ameliorate. People who come to the U.S. from other parts of the world often find our way of living perverse, in ways we simply take for granted as facts of human nature, rather than peculiar societal failings. I was recently talking to a Haitian-born U.S. citizen who works long hours as a nurse’s aid, and then comes home each night to care for her mentally disabled teenage son. She told me that if it were possible, she would go back to Haiti in a heartbeat. She was desperately poor in Haiti, but there, she said, her neighbors would have helped her: they would have invited her over for dinner, they would have offered to look after the children. “Here,” she said, “nobody helps you.” That’s one of the worst condemnations of American civil society I’ve heard in a while.