Two Methods for Converting Celsius to Fahrenheit in Your Head

Celsius to Fahrenheit conversion is one of the few remaining areas where doing the math in your head can be simpler and faster than looking it up. I’ve spent some time in Celsius countries and I still have no intuitive feel for how warm or cold any Celsius temperature is. With Fahrenheit it’s easy to think in 10 degree increments – temperatures in the 70s are easy to distinguish from temperatures in the 60s, etc. This method is not very useful for Celsius temperatures since a Celsius reading in the 20s can be anything from 68 to 84 Fahrenheit.

I spent some time thinking of a mental algorithm that could be relatively precise yet simpler than the math for the official converison (Fahrenheit = 9/5 * Celsius + 32). Fractions are difficult so I never find it worthwhile to attempt to do the conversion in my head. The simplest algorithm I’ve used is to double the Celsius temperature value and add 30. That is a decent fall back, but at warmer termperatures it will yield a too high Fahrenheit temperature, and at cooler temperatures the Fahrenheit value will be too low.

An ideal algorithm would be simple to calculate and easy to remember. The first method I thought of is this: Take 2 times the temperature Celsius, subtract the first digit of the result, and add 32 to get the degrees Fahrenheit. It works in reverse too, though I almost never need to do that calculation: take Fahrenheit and subtract 32, then subtract the first digit from the result, then divide by 2 to get the degrees Celsius.

For example, for 22 degrees Celsius: 22 Celsius * 2 = 44. Subtract 4 (the leading digit) and add 32 = 72 Fahrenheit. This method is consistently accurate. The problem with this method is remembering to only subract the tens digit and not subtracting anything if the Celsius temperature is 0-4 (the doubled value would be only one digit, so the first digit in that case could be thought of as 0). Also adding 32 is more difficult than adding 30, though it is easy to remember, since almost everyone knows that 0 degrees Celsius = 32 degrees Fahrenheit. Also, for negative Celsius temperatures you would subtract a negative, so you would have to add the tens digit. So -22 degrees Celsis would be: -22 * 2 = -44. Subtract -4 (so add +4) to get -40. Add 32 to -40 = -8 degrees Fahrenheit.

I was disappointed to learn that others had thought of this algorithm before me. I found it interesting that in this thread someone claimed that this method was just as complicated as the correct conversion equation. I disagree with that, and I think many people prefer to avoid fractions. However, it seems that this method could be a bit hard to remember, and it is not that much simpler.

I tried to think of a simpler method that would involve adding only thirty to simplify the addition part. The best I could come up with is similar to the first method, but less precise. Take the degrees celsius and subract the tens digit (if it is negative, add the tens digit). Double that number, then add 30.

For example: 22 degrees Celsius: take 22 degrees and subtract the first digit (2) to get 20. Double that to get 40, then add 30, to get 70. It’s reasonably close to the correct value.

This method would actually be more precise for positive Celsius values by adding 31. It works okay for adding 32, and is more precise for negative Celsius values than adding either 32 or 31. Ultimately I think the second method is less useful than the first method since it is probably harder to remember, even though computationally it might be a little easier.

Thoughts on the Apple Event and Apple’s Pricing

I viewed the event partly as a consumer and partly out of interest in Apple’s stock. Apple stock gained steadily during the Watch Series 3 presentation, but fell immediately after the FaceID demo failure. I expected the stock to rebound relatively quickly after it was determined the failure ocurred because handlers had caused FaceID to lock after a set number of failures, which then required a passcode to re-enable FaceID. This makes the failure seem less as bad since it wasn’t a technical problem with FaceID. However Apple’s stock price has not fully recovered so far today (two days after the event). Coverage has not focussed much on the FaceID fail, since it was not really the “fault” of FaceID. However, it now seems likely that the stock has not rebounded because Apple’s iPhone strategy is underwhelming. I see problems ahead with either higher entry level iPhone prices or a bifurcated lineup.

The Apple Watch Series 3 seems like a good product and received lots of positive attention so far. The Watch is an interesting maximalist product in that it attempts to do many things for many people, though it is still fitness focussed (it even has a barometric sensor). The features do seem like a strong point for the watch, though it still lacks an always on display which is common in competitors. The pricing seems reasonable, especially considering the large markup for cellular iPads. Apple has priced the Watch more aggressively than Macs or iPhones, and consistently lowered prices with the introduction of new models. The Series 3 without cellular is cheaper than the Series 2 when it was introduced, and it has a better processor and connectivity chip. Stylistically the Series 3 compares very favorably with the LG Watch Sport, an example of the large and clunky Android Wear watch with cellular connectivity. However Samsung’s watches are more attractive than most Android Wear watches, and the cellular model looks like it will be a viable alternative to the Series 3. It seems that even for Apple and Samsung (and definitely for LG) the current state of technology can’t deliver a thin form factor with all of the sensors the market expects and adequate battery life. It is unfortunate that the Series 3 is thicker than the Series 2 since this product out of all Apple products could most use some thinning, though Apple’s bands do a good job of making the watch appear thinner than it is. The bands also provide some nice variety, though Android Wear watches come in all different shapes and sizes. It is interesting that the Series 1 is still around – it is actually the thinnest version of the Watch. The price for the Series 1 is okay, it received a $20 price drop, but it will probably receive discounts around the holiday season and periodically from retailers.

I wish Apple would allow the option for the watch to display time constantly. I think Apple’s reasoning to not allow this option is that with the cellular version, battery life would suffer too much. This doesn’t seem like a good enough reason for Apple to not allow time display when cellular is disabled, or on the non-cellular watches. One day battery life would be sufficient for many people, though Apple is making a push in to sleep tracking and probably hoping people will wear their watches while sleeping.

I was not thrilled about the different features between the iPhone 8 and iPhone X. It would have been nice for the iPhone 8 to have FaceID and Animoji (in addition to TouchID on the 8). Animoji seem targeted at teens, and the high entry price for the iPhone X will obviously be a barrier in that market segment. It is strange that the phone with no bezels has the fancy added sensors for FaceID, while the phone with the large bezels, which should have more space for FaceID tech, does not have it. Perhaps Apple did not want to have FaceID and TouchID on the 8 and just FaceID on the X. It also would have been nice for the dual camera system to come to the iPhone 8 in addition to the 8 Plus. I’m not sure if the dual cameras are important for AR Apps, so I don’t know if the lack of the dual cameras will affect AR on the iPhone 8 compared to the 8 Plus. I hope that Apple put a lot of effort in to improving the camera and camera software on all the phones, Google has caught up quickly with the Pixel and Samsung has consistently had great camera performance. Obviously camera performance would be an important selling point for these phones.

I don’t like that the base model iPhone 8 increased by $50 and by $30 for the 8 Plus. It is bad marketing and PR since the 256GB iPhone 8 model is actually the same price as the iPhone 7 256GB when introduced, and $20 less for the Plus. It costs $850 for the 8 and $950 for the 8 Plus. Instead of the traditional 3 tiers with $100 price difference, Apple is going with two tiers only with a $150 price difference. However, even though the 256GB version either stayed the same or decreased slightly, the headlines will focus on Apple’s price increases. Also, removing the 128GB option forces consumers to the more expensive version. Apple also picked a bad time to increase the prices of the iPad Pros, as that could play in to the narrative of Apple raising prices. The Apple TV is also not priced very competitively, and will mostly appeal to those who are locked in to iTunes because of previous purchases.

It’s interesting that the iPhone 6S is still around, I wonder if it is popular partly because it still has a headphone jack. It is priced attractively since it received a $100 discount for both the 32GB and 128GB versions. The iPhone SE received a $50 price cut for both storage tiers.

Although the discounts for the 6S and SE were welcome, and the Apple Watch and AirPods have been priced reasonably, I don’t like Apple’s pricing because it points towards a consumer unfriendly future for the iPhone. If the iPhone X is the future as Apple emphasizes, will Apple stop making phones like the iPhone 8 (will there be an 8S or 9?) Apple will either have to continue with a bifurcated line, which has obvious disadvantages for marketing and complexity, drop prices on the iPhone X and its successors signficiantly, or as is most likely, transition to the iPhone X as the base model iPhone at a new higher price point. Apple traditionally lowers prices by $100 one year after a phones introduction, so the iPhone X would still be $900 next year. Further, the X Plus, which is surely on the way, would probably sell at a premium to the X. Assuming the usual $150 markup for the Plus models, the X Plus would start at $1150. I’m not sure that this pricing is sustainable. The Samsung S8 starts at $725 with similar design to the X. The Samsung Note 8, which would be comparable to the X Plus, is around $950 dollars.

Apple must feel confident that its smartphones will offer compelling features if it plans to sell future iPhones at prices similar to the iPhone X. A lot of commentary has focussed on Apple’s strength in chip design, and the tentative benchmarks for the new A11 bionic show significant improvements. However, it is worth noting that Qualcomm had surpassed the Apple A10 in multicore CPU and GPU performance with the Snapdragon 835, so it would not be surprising for Qualcomm to catch up again some time during the A11’s life. Further, Google will also begin designing custom silicon, though that might still be farther away.

Obviously it seems that Apple thinks this feature is AR. However, Apple’s lead in AR may not be that strong, and Android phones with AR capabilities sell at similar prices to “normal” Android phones. Perhaps Apple thinks it has some killer feature coming that will justify higher prices, such as AR glasses. However, at the moment it is not obvious to me that Apple has a large lead in AR or any other area to justify the high prices.

The Crisis in Venezuela – Some Good Explanatory Articles

I have been reading about the crisis in Venezuela for some time and following events there with interest. Although I have read a lot of good work that brings to life the tragic situation, I have seen relatively few in depth articles about the causes of the crisis. Venezuela seems like a case where economic policy has had powerful impact, and the crisis should be good case study to examine the causes and consequences of various policies. I think there are relatively few articles on this topic because an in depth discussion of the policies requires discussion of relatively difficult economic.

My current favorite article from last year on Quartz These are the brutal emergency measures it would take to pull Venezuela back from total collapse right now. The article lays out three basic reforms and one additional recommendation. Each of the reforms points to a cause of the crisis.

Reform one is to get rid of price controls. The full impact of price controls is difficult to estimate, however they are undoubtedly extremely harmful and counterproductive in fighting inflation. Price controls have harmed the domestic economy in Venezuela. Local producers face bureaucracy and uncertainty.

Price controls als inevitably lead to a thriving black market where basic goods are sold for many times their official price. These leads to wasteful activity like arbitrage by “Bachaqueros.” This Bloomberg article includes anecdotes like teachers skipping work to stand in line to resell goods. This reform seems relatively straightforward, and could lead to Venezuelan producers slowly growing and replacing unproductive activity such as selling items on the black market. However, for local business to truly grow again, many other reforms to ameliorate the business climate would also have to occur, see this Atlantic piece for more sad details.

Reform number two is to let the exchange rate float. Currently Venezuela has multiple fixed exchange rates that are far below the black market exchange rate. There are complicated capital controls which favor insiders, as only those with connections can actually import goods at the official rate. This reform seems much more difficult and probably painful than getting rid of price controls, and in order to be successful would require other reforms, such as increasing the attractiveness of the business climate for foreign investment and local business. This would require stable property rights and no price controls, among many other things.

The artificially low fixed official rate is essentially a subsidy keeping imported food cheaper. Imports are necessary in part because of price controls and the hostile business climate, since such policies greatly diminished the local production of food and goods. However, with a greater reliance on imports and the government printing more money, the Venezuelan Bolivar has decreased greatly in value and probably would continue decreasing in value with a floating exchange rate.

Letting the exchange rate float would increase the cost of imports greatly, possibly making food unaffordable for many. However, the current multiple exchange rate structure offers great incentives for corruption and arbitrage – insiders connected to the government can import goods and exchange dollars at favorable exchange rates, and sell the goods or exchange the dollars on the black market at much higher rates. The large difference in rates provides huge profit opportunities for corrupt insiders. Further, foreign investment could actually increase from a devalued Bolivar because of how cheap everything in Venezuela would be, but only if exchange rates remained relatively stable for the medium to long term.

The third reform is directly related to the second reform. It is to eliminate unproductive subsidies. This section should have also included more generally reforming the environment for doing business. The unproductive subsidies include the forced nationalization of many industries, resulting in collapsing production, and the subsidies for electricity and gasoline. The lack of productivity of the domestic economy is in part a cause of the weakness of the Venezuelan Bolivar. Private business has been harmed by regulations and price controls, and nationalized companies have floundered due to corruption, patronage, and waste. Once again, nationalizations will be difficult to undue because of corruption. Insiders will not want to give up their benefits.

The fourth suggestion is for Venezuela to go begging, possibly to the IMF or to other lenders. This has become more fraught recently, with U.S. sanctions and the possibility of default looming. Venezuela has not defaulted because creditors could seize its oil assets, possibly including Citgo assets or oil tankers. Without oil revenues, which are 90% or its dollar earnings, imports would collapse.

The Financial Times Alphachat podcast has had several good podcasts about Venezuela. One podcast was with Ricardo Hausmann, who has written informatively about the crisis. The article for the podcast has several other links, including a very interesting survey of business leaders which shows just how bad the business environment in Venezuela is. The other FT Alphachat podcast about Venezuela is a rather technical exploration of Venezuela’s options for restructuring its debt, see also this article on new bonds from Venezuela. Any debt restructuring would be extremely messy, yet investors are holding on to Venezuelan bonds because the government has continued to pay – so far. A default could lead to a collapse of the government, which would remove the government’s hold on opportunities for patronage and corruption.

Don’t Short Tesla

I am not convinced by recent arguments for shorting Tesla. I am not long Tesla, as I view it as an extremely risky stock. However, there are good reasons for its high valuation that shorts are overlooking – principally the high valuations of other companies in the lead in the self driving race. Also, the possibility of an acquisition will probably keep a floor under Tesla’s stock price. 

If you are shorting Tesla, you should consider Nvidia’s valuation, the price Intel paid for Mobileye, Google’s valuation and Uber’s valuation.The high valuations of these companies show that the market is bullish on autonomous driving. Once Waymo is spun out of Alphabet we will have further data. from Waymo’s valuation. Companies perceived as in the lead in autonomous driving technology have huge P/E multiples. Compare Nvidia’s P/E of 47.77 and market cap around $98.87 billion to AMD (market cap of $12.19 billion) and Intel (P/E of 13.71, market cap of $168.8 billion). Nvidia is clearly in the lead in autonomous driving, with Tesla and Audi both using Nvidia computers. Nvidia’s lead in graphics chips and cryptocurrency is not large enough to justify its large P/E multiple compared to AMD or IntelThe market thinks Nvidia’s position in the self driving race justifies a huge premium above its competitors.

Google, with its P/E of 30.73, has been a leader in self-driving cars for some time while Apple, with a P/E of 18.34, has just recently entered the competition and lacks the big data experience probably necessary to succeed in this area. The different valuation multiples for Google and Apple obviously reflect more than their different likelihoods of success in the self driving car race, but it shows investor value big data and machine learning expertise. Uber’s current business model doesn’t actually make money, and yet it is valued around $50 billion. Uber, along with Google and Tesla, is a leader in autonomous driving and investing heavily in it. The text messages that recently surfaced between Levandowski and Kalanick show those two were closely following Tesla and Google. This implies Tesla is viewed as a leader and significant competitor in self driving field.  Intel paid an extremely high price of $15 billion for Mobileye, even though Mobileye had recently faced struggles (Tesla dropped Mobileye but still uses Nvidia). The price Intel paid for Mobileye works out to a P/E ratio around 130. Qualcomm’s acquisition of NXP was also likely related to gaining self driving expertise.

Tesla has a strong brand, and is currently a leader in the luxury car market in the U.S. It is considered one of the most prestigious places to work in Silicon Valley, and it will continue to attract top talent. With some of the best engineers, Tesla will likely maintain its position in the self driving race. Tesla’s market value slid significantly after news of a crash while Autopilot was engaged (the crash turned out to not be related to Autopilot. The large stock slide following this incident implies significant portion of Tesla’s valuation is based on Autopilot software. 

Comparisons of Tesla to other car companies’ market capitalization, which are popular in the press, are not informative. Since Tesla has not taken out much debt until recently, Tesla’s market capitalization should be compared to the enterprise values of Ford (EV around 150 billion) and GM (EV around 115 billion) Ford and GM do have a lot of debt. The enterprise values show that Tesla’s valuation is still significantly below both companies. However, if Tesla were acquired, the acquiring company would have to pay a premium over its current market capitalization. Tesla would make an attractive target for Apple or for the handful of other companies who could afford it. Its brand and workforce would provide considerable value even if Tesla were not making any money. Levandowski received a $120 million bonus from Google, so Tesla’s valuation is only 500 times greater than one self-driving expert. Considering Tesla’s brand, workforce, and prominent position in the self driving race, even if Tesla faced a cash crunch, it would make an attractive acquisition target.


I like this piece on patents by Noah Smith on Bloomberg. It’s a topic he has written about before. I don’t really agree with much of his writing, but his piece is a promising sign that patent policy is something economists from either side could agree on. I’m glad that the issue of patents is continually in the news, Alex Tabarrok has pushed the topic since his book.

It seems many programmers are extremely opposed to patents, which gives me hope for the future (or lack thereof) for software patents. I wish companies such as Apple paid a larger reputational cost for pursuing and enforcing [very stupid] patents, though probably most large companies are approximately equal offenders. The only “true believers’ in the patent system in my experience are lawyers and judges, who obviously benefit greatly from a strong patent system. 

In law school I wrote a piece for my journal (never published) arguing that patents should be eliminated, and that the pharmaceutical industry should receive industry specific patent-type protection for new drugs. My editor was not convinced by my arguments that the patent system was a tax on innovation in the economy and could increase transaction costs. Unfortunately, people who believe in patents are convinced by arguments based on the pharmaceutical industry, when in reality the patent system doesn’t even work that well for that industry. A non patent-based intellectual property system for pharma wouldn’t actually be that radical, as new drugs currently receive “data exclusivity” after approval for five years, and drugs developed to treat orphan diseases receive “market exclusivity” for eight years. Data exclusivity, which means that a generic cannot use the FDA trial data for approval, works out to be market exclusivity in practice, since the cost of trials is prohibitive. After five years, a generic can use the data from the pioneer drug FDA trials in an ANDA, though in practice most drugs have patent protection that is longer than the data exclusivity period. In my article I questioned why drugs to treat rare diseases received larger incentives, and argued that “market exclusivity” could be extended to all drugs and replace patent protection. 

The patent system actually creates perverse incentives by discouraging research into those drugs that are not patentable (which may be the most promising compounds). Drugs that have previously been investigated or discussed would probably be more likely to be effective but least likely to be patentable. It’s silly to deny protection for a treatment breakthrough just because the compound or method is not patentable, and incentives should be tailored to encourage research in to every possible avenue. This could and probably would result in some company bringing Vitamin C to market as a treatment for X and charging $10,000.00 a month for it, but even in that case we would have learned through the clinical trials that Vitamin C was an effective treatment for X (and in practice market exclusivity would be hard to enforce for such a common compound). 

The most absurd thing about patents is that they are for the same amount of time (20 years), no matter the industry or type of invention (there are some tweaks in patent-life for drugs to make up for the time spent in clinical trials). I don’t think patents or patent like protection should exist for any industry outside of pharma, but even if patents are sometimes helpful, it is impossible that the incentive of 20 years exclusivity is exactly appropriate for every single industry.

I think patents are unnecessary outside of the pharmaceutical industry. First mover advantages are real. My hope is that some open source software program becomes extremely widespread and includes in its license a requirement that all users agree not to enforce their patent rights. This would eliminate patents through contract, sort of like how arbitration has gotten around some of the least efficient parts of the legal system. Current open source licences already have provisions about patents, usually related to patent claims on the software itself.

Using iPhone Models to Estimate (Dis)inflation Part 2

In this follow up blog post I will discuss my iPhone price measurements in greater detail, and discuss possible issues with my numbers. My first iPhone post is a summary and I meant for it to be able to stand alone. However some readers would like to know more about the BLS’s CPI methodology and my assumptions. That’s what this post is for.

My deflation rate roughly matches what Byrne & Corrado calculated for cell phones between 2005 and 2009. Byrne & Corrado measured the prices of used phones, and even using a large discount for used phones vs new phones, arrived at the high figure of -15.5% yearly depreciation rate. This is reassuring given the large difference between my deflation rate and the CPI rate. The CPI consists of a market basket that is based on the Consumer Expenditure Survey. The entire value of the market basket is set at 100, and items have an index value that reflects the proportion of consumer spending on that item. 

The telephone hardware, calculators, and other consumer information items is.091 of the September 2016 index, which implies that annual consumer spending on smartphones is .091% of total spending. Assuming per capita expenditure of $40,865 it implies per capita spending of around $37 per year. These numbers are not exact since my methodology is much simpler that the BLS’s. Note that the BLS state per capita spending on smartphones and wireless plans was $51.72 and $1022.85, respectively. The index value for wireless plans is a seemingly much more reasonable value of 1.740, for $711 per capita yearly spending. A brief search shows that the smartphone market in the U.S. is about $55 billion yearly, and therefore should be around .45 (or .45% of approximately $12 trillion annual consumer spending). This would get us a more realistic $183 / year per capita spending. Assuming people replace smartphones every couple of years, and some people still don’t have smartphones, this number seems reasonable. 

I emailed the BLS about this discrepancy, and they admitted the number in the CPI from the Consumer Survey seemed low and it was possible smartphones were included as part of the wireless plans item. My Verizon plan includes a monthly device payment, and the device payment with the data charges might be reported to the BLS as the cost of the wireless plan. This seems likely but is not very reassuring, since it implies that the CPI is also mis-measuring the cost of wireless plans. The BLS representative stated that the Consumer Expenditure Survey results yielded $51.72 per capita spending on smartphones vs $1022.85 on wireless plans.

One problem I faced was picking start and end dates for comparison with the CPI. By ending at September 2016, I am incorporating that price change in my measurements, without the following flat price (as of July 2017). However, I am biasing my measurement by not incorporating the price drop before September 2012, from the iPhone 4S to the iPhone 5. I struggled with whether to include the iPhone 4S to 5 transition, but did not include the 4S because non-subsidized contracts were not common before the iPhone 5, and therefore measurement errors would be more likely. See this Matt Yglesias article on the issue. 

For that matter, the iPhone 4 was the first iPhone available on Verizon and not limited to AT&T. I asked the BLS how they dealt with the issue of subsidized phone plans and smartphone prices. As the previous paragraph shows, the BLS does not seem to be currently handling this issue correctly, since it is drastically underestimating consumer spending on smartphones. However, according to the BLS steps were taken to avoid counting the value of smartphones in subsidized wireless plans. If smartphones were measured at their subsidized prices, the yearly price declines of $100 would be a much larger percentage of the total price, and this would bias CPI measures towards too high deflation. However, prior to my measurement period, iPhone price declines were generally smaller and not as easily comparable across model changes as after 2012.

Mark Bils discusses CPI measurement and the difference between forced substitutions and scheduled substitutions. “Forced substitutions” occur when a product is discontinued and the BLS must substitute a new model in its survey. Bils reviews how the BLS treats these products, and finds that for computer equipment, price increases were generally treated as quality improvements. For consumer electronics, part of the price changes were attributed to quality improvement, while for other goods including cars, only a small part of the price increase was attributed to quality improvement. 

I also looked at the 1997 BLS handbook to try to confirm which types of products received quality adjustments during forced substitutions, see page 184. The handbook states that at least in some cases, price increases in new models are implicitly assumed to be quality improvements. 

I asked BLS how and whether smartphone quality adjustments occur, and how the BLS treats forced substitutions. The BLS stated no quality adjustments are made, which matches a Heritage paper (where the author corresponded with the same BLS representative as I did, and apparently received the same answer). The BLS representative stated that smartphone prices are measured until the smartphone model leaves the market, at which point a replacement is chosen. I assumed that this process would implicitly allocate price increases to quality improvement. For example, the iPhone 6 16 GB, introduced in September 2014 at $650 and discounted in September 2015 to $550, would remain in the CPI until it was discontinued in September 2016. At this point, probably the higher priced iPhone 7 32 GB ($650) would replace the iPhone 6. I suspected that its $100 price increase would not count toward inflation, and instead would implicitly count as quality improvement. The BLS stated that the $100 price increase would not be treated as quality improvement. The BLS would adjust the price on imputation, but the methodology for this is unclear.

If, contrary to their response but in line with the BLS handbook and Bils’s findings for some subset of goods, that the BLS does assume a price increase in a new model is due to quality improvements, then chaining together the iPhone models after each one is discontinued gets us results relatively close to the CPI.

The Rise and Fall of the Third Reich Review

I just finished listening to the audiobook The Rise and Fall of the Third Reich by William Shirer. I’m not able to competently evaluate all of its factual claims, but I thought the book was terrific. The narration by Grover Gardner was excellent, and the extremely long audio book kept me interested for the large majority of its 57 hours and 13 minutes run time. It took me over a month of listening to finish but it was worth it. 

As long as the book is, I think almost every period was covered with the right amount of detail for a general overview. Shirer mentions some minor characters repeatedly thanks to the depth of detail, and this helped me with retaining the information and following the narrative. If most of the characters had only showed up once when they were at the center of action, I would have gotten lost and probably forgotten more. Shirer does a great job weaving in relatively minor players who will become important later. Shirer has funny and memorable adjectives for many of the characters – the “wily” Papen, “corpulent” Reich Marshall (Goering), Streicher the pornographer, Himmler the former chicken farmer, Rohm the homosexual etc.

In order to better understand the context and characters, I tried to integrate my listening with other WWII material to better understand the history. I listened to the audiobook In the Garden of Beasts right before this, and that provided good background for the Night of Long Knives. That book deals with the French and English ambassadors, Neurath, and Putzi Hanfstaengl who are also in this book. Hjallmar Schacht gets top billing in Lords of Finance, though I didn’t remember all the details from that book, which I read a while ago. I watched the movie Dunkirk right around when I got to that portion of the narrative, which was exciting, though the movie didn’t really add much information. I also watched Der Untergang at about the time I started this book and I very highly recommend the movie. The book provides good background and detail to the scenes in the bunker from the movie, Shirer mentions Hitler’s ever more violent outbursts, and even provides some detail on the infamous endlessly parodied “Der Angriff Steiners” scene from der Untergang. It’s also nice to be able to put the faces of the actors to the lesser known characters like Keitel, Jodl, Bormann etc. I thought about also integrating Saving Private Ryan, Valkyrie and Stalingrad (the German version), but it didn’t work out for me timing-wise. I still want to (re)watch these movies, I think I will be a much greater appreciation for the setting and the stakes. My old WWII history professor inspired me, he taught us the material through movies and his list included The Train, Schindler’s List, Das Boot, and Der Untergang among others.

The narrative gives somewhat short shrift to D-day, though in fairness the Battle of the Bulge receives more attention probably because it was more important from the German point of view and Hitler’s involvment. The July 20th plot is recounted in detail, though Shirer makes clear while discussing the various plots of the generals (aside from the July 20th plot) that much of the details are unknown and the generals involved had reason to exaggerate their plots. In contrast, Shirer takes Speer’s claims after the war at face value, I’m far from an expert on this but I think Shirer could have expressed more skepticism towards Speer (Speer meant to kill Hitler but was thwarted by a tall chimney?) I also couldn’t help but wonder if some of the other characters mentioned in the book were actually plotting against the regime and their actions are lost to us forever (Mother Night has this type of story line). 

Shirer is opinionated, which I like. His “German essentialist” thesis (that National Socialism was a logical continuation of German history) has been criticized by some historians, but it had absolutely no negative impact on my enjoyment. I generally reject the essentialist theory out of hand, though I can’t blame Shirer writing in 1957 not far removed from two World Wars, with personal experience in Germany, for advancing it. In his afterword written in 1990, Shirer admits that his theory was probably biased because of his personal experiences. If Shirer stated that the rise and fall was mostly a series of random crazy coincidences and immensely bad luck (which is what I think it was) it would anesthetize his personal experiences and visceral description of the horrors of the concentration camps and scientific experiments. I think of the thesis (which isn’t that heavily emphasized in the book anyways) on a sort of meta-level that some critics of Islam or Russia today should bear in mind. 

Despite Shirer’s own theory, the way the narrative unfolded emphasized the unfortunate contingencies of this story, and gave a lot of support to a “great man” theory of history. How many things had to go wrong for the Nazis to gain power, and stay in power. I lost count of the assassination attempts. The horrible events really were shaped by one individual, and this story is as much the rise and fall of Hitler as of the Third Reich. While it was helpful to learn that some of Hitler’s ideas were present in German society before his rise and therefore Hitler and National Socialism didn’t come from nowhere, I wasn’t convinced that racism and xenophobia in Germany during the 1920s and 1930s was much greater than elsewhere in Europe. It did seem that the Germans grievances to the WWI settlement had a strong negative effect on German politics and played in to the hand of nationalist politicians and parties like the Nazis. Hitler was apparently a true believer and absolutely dedicated to his crazy conspiracy theories. His “dedication” probably helped him rise to the top, he was not just a phony looking to enrich himself (like Goering), though many of his followers greatly enriched themselves through corruption etc. His constant harping on his obsessions and his outlook infected those below him who might have otherwise been relatively normal members of society, though he did also naturally attract virulent anti-Semites. I want to read Hannah Arendt’s Eichmann in Jerusalem to better understand her “banality of evil” thesis, based on the articles I have read I think it is compatible with an “incentive-structure” thesis, where leaders set incentives for obedience and belief in a particular world-view, and subordinates committed atrocities because of the terrible direction set from the top. 

When Shirer got in to the details of the holocaust and experiments it can be difficult to stomach, and I had to take frequent pauses during this part. It is incredibly difficult to understand how subordinates could not use the simplest heuristics (don’t slaughter women and children) to push back against the terrible Nazi ideology. That many didn’t shows the tremendous power that a leader has in setting the culture and actions of subordinates. 

Shirer’s opinion which I most disagreed with was his Hitler “genius or “evil genius” theory. Obviously it’s hard to think about this sort of thing objectively, but as with the narrative actually supporting a contingency explanation instead of an essentialist German thesis, Shirer’s own narrative gave plenty of examples where Hitler’s blundering refusal to listen to his generals accelerated his downfall. I think Hitler was not particularly smart, smart people don’t believe in idiotic conspiracies, smart people delegate authority when it is obvious they lack the necessary expertise, they don’t talk constantly for hours on unrelated topics in the midst of dire circumstance, and they aren’t fooled by their own propaganda, etc. etc. Hitler’s genius, such as it was, seems to be the lesson that if you disregard all morality, you can gain temporary advantages in negotiations especially if your negotiating partners are not aware of your character. Before the war, England and France seemed willing to grant Germany the claims that they viewed as plausibly-defensible, assuming they were dealing with a rational good faith negotiator. 

Maybe Hitler’s “genius” was also the inclination to take aggressive and risky actions which were viewed as unlikely to succeed by the status quo (the generals) which could in certain situations (probably when the opponent has low information and you have an advantage) yield surprising upside. However, as conditions changed, it seems that the payoff from aggressiveness and risk taking declined and actually turned negative. I’m thinking here of Hitler’s refusal to retreat from Stalingrad, the Battle of the Bulge, and refusal to delegate authority on D-day. Once the advantages had been gained, a more conventional strategy might have held the territory more effectively. 

Ultimately, I have to trust Shirer’s recounting of the facts, and that is the one thing that gives me pause while reflecting on the book. I don’t know this topic well enough to know how correct Shirer’s summaries are. Shirer’s recounting of the sheer duplicity prior to WWII is shocking, and at times listening I wondered if Shirer could be biased against the Germans. It’s simply hard to comprehend the degree bad faith negotiation by the Nazis with no defectors on the German side. I know that given the realities of holocaust I am probably focusing too much on Nazi negotiation tactics, but as a lawyer I am used to seeing cases where looking at the evidence from each point of view, both sides have a plausible argument based on legitimated misunderstandings etc. Here, it is almost inconceivable how the Nazis could justify their actions, and all I could think of to explain it was that some conspiracy-theory prone paranoid people are intrinsically corrupt and willing to take moral shortcuts. One of the few satisfying aspects of the narrative was learning that Ribbentrop was executed at Nurnberg. 

Shirer ends his afterword with the traditional “Remembrance of the past helps us understand the present” cliché. And I think there really is a lot to learn from this history and a lot to think about. Not in any particular order, here are some things I thought about while listening to the narrative – these are meant go beyond the obvious stuff which hopefully everyone knows at this point. 

First lesson: we may be attracted to leaders who express certainty, confidence, and a clear world-view, and this can have bad results depending on what the specific ideology of the leader is. Second lesson: A leader has huge influence on the incentives for subordinates. A leader’s ideology can attract like-minded followers, and with a few true believers at the top, the incentives can lead to shockingly bad behavior from subordinates. Third: The power of incentives generally – I’m already on the economics band-wagon so I believe in the power of incentives. Here, the incentives were to follow orders and endorse Nazi ideology, and it’s crazy how few people defected, including organizations that were otherwise not aligned with the unhinged parts of Nazi ideology. Fourth: conspiracy theories and paranoia may be closely related to immoral behavior, and perhaps the most dangerous people are those who sincerely believe conspiracy theories and don’t merely espouse them for short term gains. Those who truly think their opponents are acting in bad faith or evil should be kept from power if at all possible, as paranoid conspiratorial thinking might be directly related to immoral behavior. Fifth lesson: state control of the media works, and this can be extremely dangerous because it will increase support for the government in power, perpetuate the length of the government’s rule, and hide inconvenient facts which many people will never discover on their own. Possibly those pulling the strings will begin to believe their own propaganda. Sixth and related to this – few people will make an independent effort to figure things out and spread the news. Figuring things out without guidance is difficult. Seventh: When war happens, resistance dwindles further. Criticizing the leadership in times of strain is extremely difficult, and war can unite an otherwise apathetic populace. Eight: we may be underestimating the risks of patriotism and nationalism, and militarism. National Socialism presents a tail risk of what can happen. National Socialism undoubtedly received great support from grievances based on WWI, and general German nationalism that supported some of the Nazi aims. What positive contributions have come from patriotism to possible offset the disaster of National Socialism? Should we try to influence people in countries where we see nationalism and nationalist grievances taking hold? Lastly: how bad would things have to get for me to dissent and act against an immoral organization. Would I be able to find out the truth by myself when others did not care and it was relatively difficul? Would I be able to ignore incentives if they were set up to reward transparently evil ends? How often should I consider what current incentive structures reward? And should we generally be thinking more about what structure incentives are creating?

I don’t think that we are generally in a situation at all like the Third Reich. But this story shows how hard it can be to be a good person, and how easy it is to fail at that.

Tesla Follow-up (Or What are the Pessimists Missing)

My iPhone pricing follow up post is on the way, but I find the Tesla commentary interesting and I have reactions to some of the recent pieces that I have read.

A few articles I’ve read recently at the Financial Times and WSJ been bearish on EV uptake and Tesla, even as they note Tesla’s recent positive stock price movements. The Financial Times has made an industry out of doubting the pace of electric vehicle adoption, while Bloomberg is far more bullish. The FT notes the slow growth of battery technology as a leading factor, while citing other important issues most of which are probably familiar to EV proponents. While the slow change in battery technology is disappointing, the 5% yearly improvement in quality or price that FT estimates is still fairly significant. Gasoline engines are also becoming more efficient; however gasoline infrastructure is well developed, while electric infrastructure has a lot of room to improve. The FT’s battery estimates are on the low end, and I see estimates from between 5% to 20% yearly price or quality improvement. Bloomberg notes that battery prices have dropped by 73% since 2010, I’m not sure how to square this with the FT’s battery claims. I’m also not sure about FT’s China argument – that China will take over the market for batteries as it did for solar panels, and this will depress prices and stifle innovation. The FT does not link to any sources for its claim, and lower battery prices from subsidies could only be a good thing. Further, contrary to the FT’s claims that car companies are skipping investments in battery tech, Toyota is investing heavily in it, and the European car companies seem strategically aligned with an EV future (but what’s taking them so long). Also, it doesn’t have to be car companies that invest in battery technology, and this would be outside the traditional competency of car companies. 

The pessimistic take on electric vehicles may be missing the quality difference between EVs of gasoline engine cars. These analyses look at the timeline for “price parity,” and even he most optimistic views have price parity arriving sometime after 2025 the earliest. However, price parity may not be the key to understanding when EV sales will overtake gasoline engine sales. There are quality differences between electric vehicles and ICE vehicles inherent to the type of propulsion. A price parity analysis would be more appropriate for hybrid or plugin hybrid models, since hybrids are generally otherwise indistinguishable from pure ICE vehicles. Tesla started at the high end of the market, where subsidies should have less of an impact, and it is the leader there. And electric cars are poised to take the lead in performance. And consumers have placed approximately 500,000 reservations for Tesla Model 3, and the majority of these purchases will occur after the phase out of the federal electric-vehicle tax credit. This implies that at least some consumers view the Model 3 as currently at price parity. However, some of the reservation holders may be expecting large future improvements in Autopilot and the Supercharger network, which may be part of a bubble mentality towards Telsa. State tax credits are also probably contributing to reservations. An optimistic view should focus on battery technology and price improvements combined with the fact that the Model 3 is currently at “price parity” for a significant number of consumers, and there are many reasons to believe that price and quality improvements in electric vehicle will strongly outpace improvements in ICE vehicles. 

A point in favor of an optimistic view is that possibility for increased range from drafting, or platooning. Teslas get worse range in highway driving because of increasing air resistance at higher speeds. I went through a phase of intensely following professional cycling and learned about the problem of air resistance at higher speeds and the importance of drafting (it’s the reason why in most Tour de France stages the cyclists are packed closely in a flying V formation). Greater air resistance at higher speeds makes the road trip – range problem worse (see my first post, I think the road trip issue is one of the biggests impediment to EV uptake right now). But with Autopilot, Tesla could theoretically have several cars draft off of each other in a line. The Autopilot software could automatically alternate the leading “pace car” and monitor the battery level of each car. The cars would have to be very close to each other to get a good drafting effect, and this would be something of a safety concern, but it should be okay if Autopilot were good enough. Wikipedia calls this “platooning” and according to MythBusters, drafting off of a semi-truck could save between 10% (drafting at 100 ft) and 39% (drafting at 10 ft) on fuel efficiency. These tests could be overstating efficiency gains since they were behind a truck, or understating the gains since the test was at 55 mph, slower than normal highway speeds. The more Teslas on the road (or Tesla semi-trucks, or semi-autonomous trucks that communicate with Teslas?, or just normal cars that communicate with Teslas??), the more Tesla could take advantage of the possibility to increase range. Maybe Apple’s autonomous systems and / or car project is related to this drafting car-to-car communication project – I’ve long thought and still think that Apple should buy Tesla. Car-to-car communication and drafting would have large gains for highway driving electric vehicles by possibility eliminating the need to recharge for trip lengths that fall under the added range. For ICE cars, such car-to-car communication would have smaller marginal gains, since refueling is much less of a hassle, and the price of gas is already low. For ICE trucks this could still be a big deal. 

The Financial Times Alphaville article notes a few other reasons to be bearish about Electric Vehicles, including the increasing costs from subsidies and the loss of income from gasoline taxes. While these points seem true, I think it misses the huge gains from decreased pollution in cities and the smaller gains from quieter urban environments. These externalities from gasoline cars probably justify existing taxes and subsidies.

Michael Pettis Not Very Fox-Like on Alphachat

I really like the Financial Times Alphachat podcast with Cardiff Garcia. The most recent guest is Michael Pettis and he makes a lot of bold and interesting claims, many related to the current account surplus of Germany and China that has been in the news recently. I think Pettis is influenced by Minsky (who I am not familiar with) among others, and I have just started his book The Great Rebalancing. Trade imbalances have been in the news recently, with The Economist article, see reaction from Scott Sumner, and also this older piece from David Andolfatto

I’m sure Pettis knows much more than I do on these topics but I find him far too certain in his analysis. He resembles the hedgehog when recent events like the financial crisis and unorthodox monetary policy demand fox-like thinking. He analyzes almost everything around country level savings rates, and assumes that savings finance investment. This minimizes endogenous money and credit creation – it ignores the actual process by which banks create money by creating loans. Pettis or the hosts could have noted that loanable funds / financial intermediation theory has come under criticism (for over 100 years) and Pettis bases many of his claims on this doctrine. I find the critics of loanable funds theory persuasive, and the Financial Times has pointed out (multiple times) how economists have misunderstood the role of finance in money creation. Even if Pettis does not agree with the critics (and I’m sure he doesn’t), I think he could admit the high degree of uncertainty around the issues of savings, investment and capital flows influence the global economy. Borio and Disyatat have a very helpful paper on the savings glut theory that deals on the distinction between financing and current account flows. When prominent economists can reach opposite conclusions, I think it shows how little we actually know about the topic.

I also find the inequality point from the first of the two podcasts questionable. Pettis argues that inequality contributes to over-saving, since the wealthy tend to consume less as a percentage of their income. That seems true enough, but assuming labor markets are relatively efficient and high salaries are not based on rents or regulatory arbitrage, higher salaries should go to those who provide goods that are cheaper or more appealing to a large number of consumers. It would be strange if Bill Gates and executives at Microsoft got rich from bringing about better and cheaper computers for more people to enjoy, but the growth in inequality caused overall consumption to decline and sowed the seeds of a financial crisis. I find this dialectic capitalism model not very persuasive.

Tesla Model 3 Quick Thoughts

I have several iPhone pricing follow up posts, but Tesla news is always exciting, so I put down my quick thoughts on the Model 3.

The Model 3 looks nice, and the reviews I’ve seen mention it drives nicely and has a premium feel. I don’t like that you have to look to the middle of the dashboard for the speedometer, but I imagine it’s something you can get used to. It’s a fast car, and definitely a strong BMW 3 series competitor. German car makers have a lot to worry about, with the Model S and X at the high end and now the Model 3 in the compact luxury segment. Ideally for Tesla, the Model 3 would start eating in to the economical segment dominated by the Japanese automakers. In some states with large subsidies and a strong supercharger network, I think the Model 3 comes close to this possibility.

It looks to me like the Model 3 with the upgraded battery pack and glass roof is the most attractive package, and that configuration would definitely eat in to Model S sales. Right now if I had to choose between a Model S 75 and a Model 3 with upgraded battery pack and glass roof, I would choose the Model 3 even if they were priced the same. The Model S 75 has a range around 250 Miles and costs $69,500 (glass roof standard), the Model 3 with upgraded battery pack and glass roof has a range of 310 miles and costs $49,000. We’ll see if there will be upgrades to the Model S – ideally bigger battery packs for better range would help differentiate the Model S. Could the Model S be the first electric car to break 400 mile range- it would be a great way to get Tesla even more publicity. Tesla could increase the price of a 400 mile range car and still find buyers who value the added range. Also, even with a higher up front price, cost of ownership may not increase as much since models with higher range will probably retain value better, as the electric vehicles gradually get larger and more efficient batteries. 

My preference for greater range is molded by wanting to be able to drive from New York to Cleveland with just one stop for recharging. Since the trip is around 460 miles, the Model 3 range would get me close to making the whole trip with just one 30 minute stop at a supercharger station. On Tesla forums, I frequently see questions about range and going for longer drives to visit relatives etc. A lot of people make occasional road trips, and these trips right now are the main thing discouraging electric vehicle purchases. It seems many Tesla buyers have been pushed to the top of the line Model S 100 because of its superior range (around 330 miles). It’s funny since I probably make the Cleveland to New York trip at most 2-3 times per year, yet it weighs heavily on any purchasing decision. As the Tesla forums and conversations with my friends show, I think most normal car buyers put a lot of weight on the road trip range-anxiety issue, and don’t really care about performance. The Model 3, with 0-60 time around 5-6 seconds, is fast enough, already faster than almost all non-performance cars. Model S type performance seems excessive and mostly for the tiny proportion of previously Porsche-buying enthusiast types. There’s nothing wrong with enthusiasts, I just think marginal performance benefits past the Model 3 are attractive to a very small proportion of car buyers. 

I think now that the range of the Model 3 is set, the biggest improvement to enhance the appeal of the Model 3 a larger supercharger network, and secondarily faster supercharger times. Imagine if superchargers could charge the battery half full in 15 minutes, and you never had to worry about finding one. I’m surprised Tesla hasn’t partnered with gas stations or convenience stores yet. Expanding slower charging infrastructure would help alleviate range anxiety for road trips. Imagine if every Shell station had a “semi-charger,” and 25% of them were specially branded “Shell supercharging stations” (say that three times quickly). Looking at the supercharger network map as it is now, in many areas a road trip would not be feasible, and would-be Model 3 buyers would have to go through the unpleasant hassle of renting a gasoline-engine car. 

In a later post, I’d like to get in to detail about Tesla’s valuation. As many have pointed out, the valuation is astronomical considering Tesla has made no money yet. I’m interested in what part of the current valuation is based on Tesla’s lead in autonomous driving, the possible ride-sharing service, or its lead in electric charging infrastructure. Each of these could make a strong stand alone business, like Uber and Lyft, so it’s interesting that Tesla’s aggregation is viewed so favorably by the market.