Falsifiable, Literature, Model, Science

Pump Six and the Perils of Speculative Fiction

I just finished Pump Six, a collection of short stories by Paolo Bacigalupi. A few weeks prior to this, I read Ted Chiang’s short story collection, Stories of Your Life and Others and I couldn’t help but be struck by the contrast between them. Ted Chiang writes stories about different ways the world could work. Paolo Bacigalupi writes stories about different ways the future could happen.

These are two very different sorts of speculation. The first requires extreme attention to detail in order to make the setting plausible, but onceyou clear that bar, you can get away with anything. Ted Chiang is clearly a master at this. I couldn’t find any inconsistencies to pick at in any of his stories.

When you try to predict the future – especially the near future – you don’t need to make up a world out of whole cloth. Here it’s best to start with plausible near future events and let those give your timeline a momentum, carrying you to where you want to go on a chain of reason. No link has to be perfect, but each link has to be plausible. If any of them leave your readers scratching their heads, then you’ve lost them.

Predicting the future is also vulnerable to the future happening. Predictions are rooted in their age and tend to tell us more about the context in which they were made than about the future.

I think Pump Six is a book where we can clearly see and examine both of these problems.

First, let’s talk about chains of events. The stories The Fluted Girl, The Calorie Man, The Tamarisk Hunter, and Yellow Card Man all hinge on events that probably seem plausible to Bacigalupi, but that feel deeply implausible to me.

The Fluted Girl imagines the revival of feudalism in America. Fiefs govern the inland mountains, while there is a democracy (presumably capitalist) on the coasts. This arrangement felt unstable and unrealistic to me.

Feudal societies tend to have much less economic growth than democracies (see part 2 of Scott’s anti-reactionary FAQ). Democracies also aren’t exactly great at staying calm about atrocities right on their doorsteps. These two facts combined make me wonder why the (Coloradan?) feudal society in The Fluted Girl hasn’t been smashed by its economically (and therefore, inevitably militarily) more powerful neighbours.

In The Tamarisk Hunter, the Colorado River is slowly being covered by a giant concrete straw, a project that has been going on for a while and requires massive amounts of resources. The goal is to protect the now diminished Colorado River from evaporation as it winds its way into a deeply drought-stricken California.

In the face of a bad enough drought, every bit counts. But there are much more cost effective ways to get your drinking water. The Colorado river today has an average discharge of 640m3/s. In a bad drought, this would be lower. Let’s say it’s at something like 200m3/s.

You could get that amount of water from building about 100 desalination plants, which would cost something like $100 billion today (using a recently built plant in California as a baseline). Bridges cost something like $3,000 per m2 (using this admittedly flawed report for guidance), so using bridges to estimate the cost, the “straw” would cost about $300 million per kilometer (using the average width of the Colorado river). Given the relative costs of the two options, it is cheaper to replace the whole river (assuming reduced flow from the drought) with desalination plants than it is to build even 330km (<200 miles) of straw.

A realistic response to a decades long California drought would involve paying farmers not to use water, initiating water conservation measures, and building desalination plants. It wouldn’t look like violent conflict over water rights up and down the whole Colorado River.

In The Calorie Man and Yellow Card Man, bioengineered plagues have ravaged the world and oil production has declined to the point where the main source of energy is once again the sun (via agriculture). Even assuming peak oil will happen (more on that in a minute), there will always be nuclear power. Nuclear power plants currently provide for only ten percent of the world’s energy needs, but there’s absolutely no good reason they couldn’t meet basically all of them (especially if combined with solar, hydro, wind, and if necessary, coal).

With improved uranium enrichment techniques and better energy storage technology, it’s plausible that sustainable energy sources could, if necessary, entirely displace oil, even in the transportation industry.

The only way to get from “we’re out of oil” to “I guess it’s back to agriculture as our main source of energy” is if you forget about (or don’t even consider) nuclear power.

This is why I think the stories in Pump Six tell me a lot more about Bacigalupi than about the future. I can tell that he cares deeply about the planet, is skeptical of modern capitalism, and fearful of the damage industrialization, fossil fuels, and global warming may yet bring.

But the story that drove home his message for me wasn’t any of the “ecotastrophes”, where humans are brought to the brink of destruction by our mistreatment of the planet. It was The People of Sand and Slag that made me stop and wonder. It asks us to consider what we’d lose if we poison the planet while adapting to the damage. Is it okay if beaches are left littered with oil and barbed wire if these no longer pose us any threat?

I wish more of the stories had been like that, instead of infected with the myopia that causes environmentalists to forget about the existence of nuclear power (when they aren’t attacking it) and critics of capitalism to assume that corporations will always do the evil thing, with no regard to the economics of the situation.

Disregard for economics and a changing world intersect when Bacigalupi talks about peak oil. Peak oil was in vogue among environmentalists in the 2000s as oil prices rose and rose, but it was never taken seriously by the oil industry. As per Wikipedia, peak oil (as talked about by environmentalists in the ’00s, not as originally formulated) ignored the effects of price on supply and demand, especially in regard to unconventional oil, like the bitumen in the Albertan Oil Sands.

Price is really important when it comes to supply. Allow me to quote from one of my favourite economics stories. It’s about a pair of Texan brothers who (maybe) tried to corner the global market for silver and in the process made silver so unaffordable that Tiffany’s ran an advertisement denouncing them in the third page of the New York Times. The problems the Texans ran into as silver prices rose are relevant here:

But as the high prices persisted, new silver began to come out of the woodwork.

“In the U.S., people rifled their dresser drawers and sofa cushions to find dimes and quarters with silver content and had them melted down,” says Pirrong, from the University of Houston. “Silver is a classic part of a bride’s trousseau in India, and when prices got high, women sold silver out of their trousseaus.”

Unfortunately for the Hunts, all this new supply had a predictable effect. Rather than close out their contracts, short sellers suddenly found it was easier to get their hands on new supplies of silver and deliver.

“The main factor that has caused corners to fail [throughout history] is that the manipulator has underestimated how much will be delivered to him if he succeeds [at] raising the price to artificial levels”

By the same token, many people underestimated the amount of oil that would come out of the woodwork if oil prices remained high – arguably artificially high, no thanks to OPEC – for a prolonged period. As an aside, it’s also likely that we underestimate the amount of unconventional water that could be found if prices ever seriously spiked, another argument against the world in The Tamarisk Hunter.

This isn’t to say that there won’t be a peak in oil production. The very real danger posed by global warming and the fruits of investments in alternative energy when oil prices were high will slowly wean us off of oil. This formulation of peak oil is much different than the other one. A steady decrease in demand for oil  will be hard on oil producing regions, but it won’t come as a sharp shock to the whole world economic order.

I don’t know how much of this could have been known in 2005, especially to anyone deeply embedded in the environmentalist movement. As an exoneration, that’s wonderful. But this is exactly my point from above. You can try and predict the future, but you can only predict from your flawed vantage point. In retrospect, it is often easier to triangulate the vantage point than to see the imagined future as plausible.

Another example: almost all science fiction before the late 00s drastically underestimated the current prevalence in mobile devices. In series that straddle the divide, you often see mobile devices mentioned much more in the latter books, as authors adjust their visions of the future to take into account what they now know in the present.

Writing is hard and the critic will always have an easier time than the author. I don’t mean to be so hard on Bacigalupi, I really did enjoy Pump Six and it’s caused me to do no end of thinking and discussing since I finished reading it. In this regard, it was an immensely successful book.

Epistemic Status: The math is Falsifiable, the rest is a Model.

Model, Politics

The Pitfalls of One-Man Rule

In light of the leaks about Michael Flynn, just about everyone, from America’s allies to its intelligence officers, seems to be reconsidering how much intelligence they share with Donald Trump’s White House. I can’t think of anything more damaging to President Trump’s ability to govern than various domestic and allied agencies (semi-)publicly mulling whether or not to share information with him.

It’s not that I think this will cause irreparable damage to his public image. At this point, you can be swayed by other people’s opinion of Trump or you can’t. Trump’s base doesn’t care what a bunch of intelligence geeks in suits think about him. They just want to see jobs come back.

It’s just that Trump is already beginning to experience one of the most significant failure modes of single-person rule: isolation.

One of the little talked about virtues of democracy is how its decentralizing tendency makes isolation of key decision makes much more difficult. Take Canada as an example. There are 338 Members of Parliament, each based in a different geographic region and expected to regularly travel there and respond to the concerns of the local residents. Each MP also has several aides, responsible for briefing them and keeping them in the loop. Cabinet Ministers have all of this, plus they’ll have one or two MPs acting as their assistants in matters of their portfolio. Prime Minister Justin Trudeau is accountable to his constituents, his cabinet, his MPs, and through them, all of Canada.

It is very difficult to influence Mr. Trudeau’s decision making by influencing the information he receives. Government agencies can attempt it, but Mr. Trudeau is broadly popular, which makes this much more difficult. To hide information from a leader, you need a quorum. While this can be accomplished by a vocal minority, it becomes very difficult to gather even this in the face of enthusiastic majority support.

In addition, the diverse information channels Mr. Trudeau has access to mean that he is very likely to hear about any notable news that leaks out a department, even if his chief of staff or one of his cabinet ministers doesn’t want him to.

This has the effect of making power struggles somewhat transparent. In general, power among the elites is apportioned based on the results of elections and measured in terms of Members of Parliament and political capital (or, more concretely, opinion polling and what this means for re-election chances). All of this information is a matter of public record. Anyone who wants to know what elite faction is currently dominant and how much political capital it has left can find this out with a simple Google search.

On the other end of the spectrum, we have the Vatican. Pope Francis was elected through an opaque process and few Catholics saw his election coming. The power games of the cardinals are hidden from most congregants and their reasons for voting how they do are between them and their god. Francis will reign until he dies or steps down, but the power games of the elites (read: the cardinals) haven’t stopped. Instead of jockeying for position directly, they will jockey by trying to control the flow of information to Francis. If one group of cardinals succeeds (or even partially succeeds), they will wield significant invisible influence.

This same sort of opacity is what makes the “science” of Kremlinology appealing. Without clear information, it takes a hundred subtle hints to figure out who has power (and perhaps even more critically, who is being listened to by those with power).

Right now, it seems like Donald Trump is in a situation that is closer to the Pope than the Prime Minister of Canada.

This normally isn’t the case for presidents. They’re deeply embedded in the fabric of a party and have multiple channels for information – as well as multiple factions they depend on for support. Trump lacks both history and (in his mind, at least) dependency. The route of last resort for information to travel to the president is through donors. Trump has closed off this route by believing he’s entirely self-made.

All of this means that Trump is at serious risk of being controlled by one or two influential advisors. If this happens, there really are limited options for his party to bring him back in line and coordinate on a legislative agenda if the interests of those advisors don’t align with the interests of the Republican party.

This is what should be keeping congressional Republicans up at night. Trump should be staying up at night wondering about what his agencies are refusing to tell him.

Governments have to rely on veritable armies of analysts to keep them swimming in the data they need to act. You want to launch an airstrike on a suspected terrorist? You’re going to need a dozen people to correlate a hundred small tidbits of information to positively identify them with enough time to spare to launch a cruise missile or a drone.

These people tend not to be that loyal to any particular party (at least when it comes to how they do their job). While the heads of departments are often political appointees, their deputies are career men and women who have come up through the civil service. Whatever they lack in loyalty to parties, they make up for in loyalty to the system. This is generally enough to allay any fears about them hiding information or failing to perform their role.

Enter Donald Trump, who seems like he might just try and rip the whole system down around their ears. Do you think they’re going to stand for that? If you can’t believe that they have conviction and a genuine loyalty to the system, at least believe that they have some instinct for self-preservation. Career civil servants rely on the system for a paycheck, after all.

Imagine you’re an intelligence officer, fairly high up. You know how much of a threat Russia is. You’ve been watching them for a decade and you’ve seen how they’re gobbling up territory along their borders, trying to reclaim some slice of their lost empire. You think Trump is going to give some of the intelligence you just collected to Russia, blowing the cover of a source or two. So, you hide it. It’s easy enough to do. All you really have to do is flag it as routine, not pass it up the chain of command. It’s almost the same as phoning it in, really.

Imagine you are Trump. Intelligence is drying up. What do you do? You can go yell at your CIA Department head (who might be loyal to you). He or she can go yell at some subordinates. And they’ll promise to do better. They might, for a week or two, or they might not. Maybe you start getting more intelligence, but it’s all of terrible quality.

What do you do? What can you do?

In the end, Trump is one man. He has maybe a hundred people who are personally loyal to him. If we’re generous, we might call it 150. But I think we have to cap it at Dunbar’s number. He can’t count on an unbroken chain of personal loyalty either, because there is a disconnect between the career civil servants and the political appointees.

Trump and all of his henchmen can rant and rail all they want. But at the end of the day, they can’t compel. They can’t hold guns to the heads of every CIA analyst and demand they tell the administration everything they know. They can’t even fire them all. You can’t solve an intelligence shortage by getting rid of all your intelligence analysts. At a certain point, you just have to give up.

Think I’m exaggerating? Think this couldn’t possibly work on Trump? Read Eichmann in Jerusalem and you’ll learn it worked on the Nazis. Where open resistance failed, obstructionism and carefully cultivated laziness succeeded.

Power is in many ways an illusion and a fragile one at that. Break it and you might not be able to put it back together. If Trump threatens the CIA (or any other agency; you can also image the DoJ taking forever to close an investigation or the EPA having a bunch of trouble finishing an inspection and giving an all clear) and fails to deliver on his threat (likely), then the jig is up. He’s lost all ability to change anyone’s behaviour through threats.

So, this is the problem Trump faces. He has the presidency and he intends to use it to make sweeping changes to America. But without close cooperation with lawmakers, his term is going to look a lot like an attempt at one-man rule. Certainly, this should be frightening for everyone who cares about checks and balances in America.

But it should also frighten Trump’s supporters. One man rule is a terrible system of government. If Trump makes a serious go at it, his cabinet and advisors will be at each other’s throats (when he isn’t around) in next to no time and he’ll face persistent (but impossible to end) resistance from almost every Federal department. I don’t know how exactly Trump plans to make America great again, but I bet he isn’t prepared for large scale passive resistance.

The final remaining question then is: will this resistance show up, or are the early rumours exaggerated. On this point, the world is watching and hoping that the ordinary civil servants of America display the requisite moral courage to passively resist Trump’s most damaging requests.

Epistemic Status: Model

Data Science, Politics

Thoughts (and Data) on Charity & Taxes

The other day, I posed a question to my friends on Facebook:

Do you think countries with higher taxes see more charitable donations or fewer charitable donations? What sort of correlation would you expect between the two (weak positive? weak negative? strong positive? strong negative?).

I just crunched some numbers and I’ll post them later. First I want to give people a chance to guess and test their calibration.

I was doing research for a future blog post on libertarianism and wanted to check one of the fundamental assumptions that many libertarians make: in the absence of a government, private charity would provide many of the same social services that are currently provided by the government.

I honestly wasn’t sure what I’d find. But I was curious to see what people would suggest. Answer fell into four main camps:

  1. Charitable giving and support for a welfare state might be caused by the same thing, so there will be a weak positive correlation.
  2. Tax incentives for charitable donations shift the utility of donating, such that people in higher tax countries will donate more, as they get more utility per dollar spent (they get the same good feelings from charity, but also receive a bigger rebate come tax time). People who thought up this mechanism predicted a weak positive correlation.
  3. This whole thing will be hopeless confounded by other variables and no conclusion would survive proper controls.
  4. Libertarians are right. Taxes drain money that would go to private charity, so we should see a strong(ish) negative correlation.

I was surprised (but probably shouldn’t have been) to find that these tracked people’s political views. The more libertarian I thought someone was, the more likely they were to believe in a negative correlation. Meanwhile, people who were really into the welfare state tended to assume that charitable donations and taxes would be correlated.

In order to figure out who was right, I grabbed the most recent World Giving Index and correlated it with data about personal income tax levels (and sales tax levels, just to see what happened).

There are a number of flaws with this analysis. I’m not looking for confounding variables. Like at all. When it comes to things as tied to national character as charity and taxes (and how they interact!), this is a serious error in the analysis. I’m also using pretty poor metrics. It would be best to compare something like average tax rate with charitable donation amount per capita. Unfortunately, I couldn’t find any good repositories of this data and didn’t want to spend the hours it would take to build a really solid database of my own.

I decided to restrict my analysis to OECD countries (minus Turkey, which I was missing data on). You’ll have to take my word that I made this decision before I saw any of the data (it turns out that there is essentially no correlation between income tax rate and percent of people who donate to charity when looking at all countries where I have data for both).

Caveats aside, what did I see?

There was a weak correlation (I’m using a simple Pearson correlation, as implemented by Google sheets here, nothing fancy) between the percentage of a population that engaged in charitable giving and the highest income tax bracket in a country. There was a weaker, negative correlation between sales tax and the percent of a population that engaged in charitable giving, but more than 60% of this came from the anchoring effect of the USA, with its relatively high charitable giving and lack of Federal sales tax. The correlation with income tax rates wasn’t similarly vulnerable to removing the United States (in fact, it jumped up by about 12% when they were removed).

Here’s the graphs. I’ve deliberately omitted trend lines because I’m a strong believer in the constellation test.


All the data available is in a publicly viewable Google Sheet.

I don’t think these data give a particularly clear answer about the likelihood of private charity replacing government sponsored welfare programs in a hypothetical libertarian state. But they do suggest to me that the burden of proof should probably rest on libertarians. These results should make you view any claims that charitable giving is held back by the government with skepticism, but it should by no means prevent you from being convinced by good evidence.

I am happy to see that my results largely line up with better academic studies (as reported by the WSJ). It seems that if we look at the past few decades, decreasing the tax rates in the highest income brackets have been associated with decreasing charitable giving, at least in the United States. Whether this represents a correlated increase in selfishness, or fewer individuals donating as the utility of donating decreases is difficult to know.

The WSJ article also mentions that government grants to a charity reduce private donation by about 75% of the grant amount. I don’t know if this represents donations that are lost entirely, or merely substituted for other (presumably needier) charities. If it’s the first, then this would be strong evidence for the libertarian perspective. If it’s the latter, then it means that many people intuitively understand and accept the key effective altruism concept of “room for more funding“, at least as far as the government is concerned.


Finding good answers to the question of whether private charity would replace government welfare turned out to be harder than I thought. The main problem was the quality of data that is easily available. While it was easy to find statistics good enough for a simple, limited analysis, I wasn’t able to find a convenient table with all of the data I needed. This is where actual researchers have a huge advantage over random people on the internet. They have access to cheap labour in the volumes necessary to find and tabulate high quality data.

I’m very glad I posed the question to my friends before figuring out the answer. It never occurred to me to consider the effect of tax incentives on charitable giving. I’m now of the weakly held opinion that the main way taxes affect charitable donations is by offsetting the costs with rebates. I’m also fascinated by the extent to which people’s guesses tracked their political leanings. This shows that (on my Facebook wall, at least) people hold opinions that are motivated by a genuine desire to see the most effective possible government. Differing axioms and exposure to different data lead to differing conceptions of what this would be, but everyone is ultimately on the same team.

I will try and remember this next time I think someone’s preferred government policy is a terrible idea. It’s probably much more productive to try and figure out why they believe their policy objectives will lead to the best outcomes and arguing about that, rather than slipping into clichéd insults.

I was also reminded that it’s fun and rewarding to spend a few hours doing data analysis (especially when you get the same results as studies that get reported on in the WSJ).

Ethics, Politics, Quick Fix

Don’t confuse constitutional rights with social norms

When Ken over at Popehat gets into a full-on rant about people who don’t understand rights, I’m often sympathetic. It was Ken who made me understand that people who don’t understand rights are a threat to everyone. When many people are misinformed about their rights, those rights become easier to take away.

When Scott at Slate Star Codex talks about good social norms, I’m very keen to listen. Scott helped me understand that social groups are worth cultivating and that it’s a good idea to think about how your group norms will change your experience of interacting with people.

So, when Tessa linked me to a Slate Star Codex post where Scott disagreed with Ken, I had some thinking to do.

The Slate Star Codex post is a response to a piece Ken put up after the furor around Justine Sacco’s tweets a few years back. Ken is defending the right of everyone else on Twitter to say whatever they like in response to Justine Sacco’s thoughtless tweets. The particular part Scott highlights is:

The phrase “the spirit of the First Amendment” often signals approaching nonsense. So, regrettably, does the phrase “free speech” when uncoupled from constitutional free speech principles. These terms often smuggle unprincipled and internally inconsistent concepts — like the doctrine of the Preferred First Speaker. The doctrine of the Preferred First Speaker holds that when Person A speaks, listeners B, C, and D should refrain from their full range of constitutionally protected expression to preserve the ability of Person A to speak without fear of non-governmental consequences that Person A doesn’t like. The doctrine of the Preferred First Speaker applies different levels of scrutiny and judgment to the first person who speaks and the second person who reacts to them; it asks “why was it necessary for you to say that” or “what was your motive in saying that” or “did you consider how that would impact someone” to the second person and not the first. It’s ultimately incoherent as a theory of freedom of expression.

Scott disagrees. He argues that there is a spirit of the First Amendment and it’s summed up by Eliezer Yudkowsky with: “Bad argument gets counterargument. Does not get bullet. Never. Never ever never for ever.”

Scott asks to imagine at what point damaging responses become appropriate:

What does “bullet” mean in the quote above? Are other projectiles covered? Arrows? Boulders launched from catapults? What about melee weapons like swords or maces? Where exactly do we draw the line for “inappropriate responses to an argument”?

Scott’s eventual line in the sand is: “Bad argument gets counterargument. Does not get bullet. Does not get doxxing. Does not get harassment. Does not get fired from job. Gets counterargument. Should not be hard.”

I’m sympathetic to what Scott was trying to do here, but ultimately, I’m on the side of Ken.

Scott wants to talk about the spirit of the First Amendment, which is fine. But the spirit he wants to read into it is divorced from the reality of constitutional rights. The First Amendment, like many of the rights in the US Constitution, is a negative right – it prevents the government from acting in a certain way, rather than saying it must provide people with a certain thing. The US Government can’t stop you from saying what you want, but it has no obligation to make you heard. If everyone ignores you, the government will not intervene.

It’s pretty weird to try and read a positive spirit into a negative right. The framers of the Bill of Rights knew when the rights they were setting down were negative rights. They understood the difference between negative and positive rights. To claim that the spirit of a definitely negative right is actually positive feels like an unfair attempt to halo a set of normative ethics (or perhaps aesthetics) with the positive affect that many Americans hold for their constitution.

As far as the government is concerned, as long as people are debating and silencing through legal means, there actually isn’t a distinction between trying to debate and trying to silence. Neither type of speech can be stopped. And I think it’s trivially easy to come up with examples for why neither should be stopped as a matter of routine (if you need inspiration, think of what your worst political enemies call “hate speech” and shudder about it being banned).

Luckily, negative speech and association rights and the government monopoly on force means that it is really hard to credibly threaten people’s freedom of association, so Scott is free to build a subculture that shares his beliefs about normative ethics. A subculture is free to demand positive rights for all members within the context of subculture related discussions and has free association as the perfect tool for enforcing it.

I’m glad that this is what rationalists are trying to do and I like our subculture and all, but we can’t claim that our weird norms are universal positive rights. I know this is a common thing for subcultures to do, but it’s embarrassing.

Politics, Science

Special Topics in Nuclear Weapons: Laser Enrichment

In an effort to make my nuclear weapons post series a one stop resource for anyone interested in getting up to speed on nuclear weapons, I’ve decided to add supplementary materials filling any gaps that are pointed out to me. This supplementary post is on laser enrichment.

Enrichment is one of the more difficult steps in the building of certain nuclear weapons. Currently, enrichment is accomplished through banks of hundreds or thousands of centrifuges, feeding their products forward towards higher and higher enrichment percentages.

Significant centrifuge plants are relatively big (the Natanz plant in Iran covers 100,000m2, for example) and require a large and consistent supply of energy, which often makes it possible spot them in satellite imagery. The centrifuges themselves require a recognizable combination of components, which are carefully monitored. If a nation were to suddenly buy up components implicated in centrifuge design, it would clearly signal its intention to increase its enrichment capacity.

Recently, laser enrichment has emerged as an additional vector for proliferation. Properly called SILEX (separation of isotopes by laser excitation), this new technology has the potential to make enrichment (and therefore proliferation easier). This post discusses how laser enrichment works and puts the threat it represents in context. It’s both a summarization (and simplification) of the recent paper on laser enrichment in published in Science & Global Security by Ryan Snyder and the product of my extensive background reading on nuclear weapons.

How It Works

Like gas centrifugation, laser enrichment requires gaseous uranium hexafluoride (Hex). While the preparation of uranium hexafluoride doesn’t represent a significant technical challenge (compared to all of the rest of the work of building a nuclear weapon), it’s still the sort of work that most reasonable chemists try to avoid. “Requires work with a poisonous, corrosive, radioactive gas” will never be a selling feature of enrichment work.

Laser enrichment also requires a large laser capable of outputting 10.2µm light (which must be converted to 16µm using Raman scattering off of H2 gas), capable of pulsing 30,000 times per second. This appears to be just barely possible with current technology and impossible with off the shelf technology. It’s the sort of thing that would have to be custom assembled.

Also requiring custom assembly is the enrichment cell, which must have a nozzle capable of injecting a supersonic stream of uranium hexafluoride in such a way as to minimize post-injection expansion. The cell also must have an optically transparent window for your laser to shine through and must have several egress lines – peripheral ones for enriched product and a central one for the jet to flow out of.

Finally, if you want to make this maximally efficient, you’re going to need a mirror set up so as to have your laser pass through the gas twice. This corrects for the circular shape of the laser. Without this mirror, you won’t have enough coverage at that edges of the gas and you’re only going to operate at 78.5% of the maximum efficiency.

The whole setup looks like this:

Image Credit: A Proliferation Assessment of Third Generation Laser
Uranium Enrichment Technology

Once you’ve assembled all of this, you’re good to start enriching.

Remember, natural Hex is largely made up of 238UF6 and is only about 0.7% 235UF6. The purpose of enrichment is to increase the percentage of 235UF6 in the gas until it is almost entirely made up of this isotope of uranium.

The process SILEX uses to achieve this is relatively simple. You run the Hex and a carrier gas (the paper says SF6) through this system at supersonic speeds and low temperatures while pulsing the laser so as to hit the jet just as it leaves the nozzle. If you’ve tuned your wavelength as directed, then photons from the laser will kick any 235UF6 molecules they hit into a heightened vibrational state (called the v3 vibrational mode), while doing nothing to the 238UF6 molecules that make up most of the Hex.

235UF6 in the v3 vibrational mode will eventually revert to a lower energy (or “ground”) state, but it is unlikely to spontaneously revert to a ground state during the few milliseconds it takes to traverse the cell. For the purposes of SILEX, 235UF6 in the v3 vibrational mode will remain in that mode unless something acts on it to change it. To improperly anthropomorphize a particle for a second, this is “bad” for the excited 235UF6, because it “wants” to be at a lower ground state.

The excited 235UF6 could get external “help” from a collision with 238UF6 (this collision would allow it to release a photon and revert to its ground state), but this is unlikely if you keep the overall concentration of UF6 in the carrier gas low (the paper recommends 5%). This is in fact exactly what is done, because efficiency is maximized when 238UF6 doesn’t get a chance to collide with 235UF6.

When you put Hex in a carrier gas like SF6, you’re going to see the formation of transitory dimers. These are temporary, weak bonds between one Hex molecule and one SF6 molecule. These bonds are fairly stable, unless the Hex is in the v3 (or similar) vibrational mode. If dimer formation occurs between v3 235UF6 and SF6, the dimer is very short-lived. The excited 235UF6 dumps all of its extra energy into the dimer bond, resulting in a lot of recoil; both the 235UF6 and the SF6 go flying apart in opposite directions. It’s the dimer formation that causes a very different outcome from a simple collision with 238UF6.

This recoil tends to push 235UF6 to the edges of the stream. A skimmer positioned around the outlet collects this enriched product. Note that it won’t be entirely enriched; the outside edges of the jet will have plenty of 238UF6 because the jet is going to be mostly 238UF6 – or at least, it will be when natural or lightly enriched uranium is the input.

If you were doing this on an industrial scale, you’d set a bunch of these cells up in series, with the enriched product of each cell being the feed for the next. In this way, you’d get the same sort of cascade towards higher enrichment as you would with centrifuges.


Laser enrichment might be more space and energy efficient than centrifuge arrays.

I have to say might because there’s some uncertainty here. A few key parameters that determine ease of proliferation using SILEX are missing. This isn’t because of censors removing them for security reasons. It’s because this technology is so new that there are serious question marks hanging over it. SILEX has shown promise in lab scale experiments, but there doesn’t yet exist any proof that SILEX will be superior to centrifuge enrichment when it comes to enriching uranium on an industrial scale. Given that the pilot project has been stalled since GE pulled out, it may be quite a while before we know if SELIX will fulfill its promise or not.

It looks like a SILEX would allow a country with technology on the level of Iran to enrich the same amount of uranium with only 59% of the floor area. This would make enrichment a bit easier to hide, but would do nothing to stop leaks. It was human intelligence, not satellite photos that allowed the west to discover the work at Natanz.

The error bound on SILEX energy consumption is large enough that it’s unclear if there would be a power consumption benefit or cost for rogue states switching to SILEX from indigenous centrifuge technology. State of the art American centrifuges still beat SILEX on floor space and they may beat it in energy use.

Estimates for SILEX efficiency span an order of magnitude and in the upper two-thirds of that range it seems to be a clear winner (in terms of amount of energy required per percent enrichment). I couldn’t see any consensus on the relatively likelihood of high vs. low actual efficiency, but I would personally bet that a lot of the probability distribution exists near the bottom of the allowed efficiencies. I haven’t worked in nuclear science, but I have done chemistry, and my experience is that few processes perform as well on an industrial scale as you might expect from efficiency calculations done at laboratory scale.

Enrichment with SILEX is quite possibly easier than enrichment with centrifuges. That is to say, even if SILEX doesn’t allow rogue nations to enrich more efficiently, it might allow them to enrich at all. SILEX requires some advanced optics knowledge and the lasers needed aren’t exactly available off the shelf, but they are easier to design and build than specialized enrichment centrifuges.

Before centrifugation became the preferred method of isotope separation for nuclear weapons (and nuclear energy), gaseous diffusion was used. Gaseous diffusion plants use truly prodigious amounts of space and energy. There is absolutely no way that these things can be hidden or disguised as something else.

With the advent of centrifuges, proliferation became significantly easier. Countries used to be faced with no good path to a functioning bomb. Plutonium is relatively easy to acquire and separate, but it is very difficult to build a successful implosion weapon (and impossible to do so without alerting anyone with test detonations). Uranium was relatively difficult to enrich, which closed off the option of a simpler gun assembly weapon (it is impossible to build a gun assembly weapon using plutonium).

If you want a nuclear arsenal and don’t care that gun assembly weapons are wasteful and less useful for staging, then the advent of uranium enrichment via centrifugation was a boon to you. Gun assembly weapons don’t even necessarily require test detonations, which allows for the (slim) possibility of entirely clandestine nuclear arsenals – assuming enough uranium can be secretly enriched.

SILEX may eventually exacerbate this problem, to the point where any country with access to uranium could conceivably build a relatively low yield bomb (say a dozen or two kilotons).

At present, the technology is too new for this to be true. SILEX almost certainly has a few kinks left to be worked out. Trying to work them out at the same time as your country builds a new nuclear program isn’t ideal. Best to wait for India or Pakistan to figure them out and then leak them to you in exchange for favours or missile technology (this has been North Korea’s approach to nuclear weapons and it has worked quite well).

In a decade, SILEX may make proliferation even easier. I don’t think it will make it easy to the point where Al Qaeda or Daesh can attempt to build nuclear weapons (can you imagine Daesh setting up a high-energy laser laboratory in Raqqa?). But I do worry that countries like Saudi Arabia or the Philippines might see the calculus around proliferation change enough to justify their building of a small arsenal of uranium weapons.

That would be a disaster for world peace and stability.

Governments are already reacting to threat posed by SILEX by adding necessary components to export ban and international watch lists. If any nation buys up a bunch of laser components over a short time without a good explanation, the international community will now suspect enrichment. I’m sure there are many men and women in the basement of the Pentagon and CIA headquarters now watching all laser equipment sales for more subtle signals of gradual stockpiling. Don’t think for a second that SILEX somehow represents a cheat code for proliferation. It’s still untested and unproven and governments and international organizations are already taking steps to reduce the proliferation risk.

Most nuclear technology is dual use. Uranium enrichment by centrifugation has made proliferation easier. It also increased the energy return on investment from burning uranium in power plants from ~40x to over 1500x (see here if you want to double check my calculations). Because of centrifugation, nuclear power plants could permanently end our dependence on oil if coupled with new battery technologies (and upfront capital and political will to build them).

SILEX could further increase the energy return on investment, making nuclear power plants even more economical. But SILEX also has the potential to make proliferation easier. It’s still a new, experimental technology and it might not even pan out. Until we know for sure, it is certainly best for the world to proceed with caution.

main topic index
Model, Politics

Checks and balances can’t last if they make governing impossible

There is an interesting post by Professor Bryan Caplan spinning limited government as an insurance policy against wild swings in political climate. You should go read the whole thing, but I’ll summarize for the lazy.

Professor Caplan makes his case using a thought experiment with an angel. This angel talks to you during Obama’s inauguration and offers you a bargain. The terms are simple If you accept, neither Obama nor Trump will be able to get much done. You trade away Obamacare and in exchange you don’t get Trump’s immigration policies. Professor Caplan frames this as a form of political insurance, a guarantee of mediocracy instead of potentially wild swings.

Professor Caplan points out that this insurance (which might be sounding pretty tempting to you right about now) is actually similar to the concept of limited government, something we already know how to achieve. From his post:

If you want the insurance of limited government, there are well-tested mechanisms to deliver it.  You all know them.  Supermajority rules require more than a majority to act.  Division of powers makes it hard for government bodies to accomplish anything on their own.  Judicial review allows judges to invalidate acts of government.  Federalism greatly reduces the cost of “voting with your feet.”  If you think these institutions aren’t working, the obvious solution is to strengthen them.  Impose more supermajority requirements.  Divide more powers.  Overturn legislation that fails to get support from six, seven, eight, or all nine Supreme Court Justices.  Make states pay for their own spending with their own taxes, not federal grants.

Prof. Caplan then briefly remarks on the known difficulties of attaining any of this. I’m actually not going to comment on that part of the post, because that isn’t what interested me.

Instead, I want to talk about limited government.

The thing I find most difficult when discussing limited government is that everyone wants a “limited” government to do different things. Many libertarians take it on faith that a limited government would naturally contain a police force to protect private property and contractual rights within the territory of the state and a military to protect that territory from outside threats. On the other hand, if you ask a centre-left policy wonk (hi!) about limited government, they’ll tell you that a limited government shouldn’t have much in the way of an army but it should be willing to run insurance programs in response to market failures. Radical leftists might want some redistribution without police or moral laws, while Christian fundamentalists might want strict morality laws but limited taxes.

(I want to pause here and point out that “limited government” as a concept must be backed up by a specific set of mechanisms by which it is limited. Throughout this essay I will treat “limited government” as the philosophy that a government shouldn’t be able to do whatever it wants and “checks and balances” or “separation of powers” as some of the specific mechanisms that are used to achieve that philosophical aim.)

With so many different (and sometimes mutually exclusive) ideas as to what a limited government actually means, you can’t just say “we have limited government” and expect voters to leave it at that. No matter how many checks and balances your government must contend with, voters are going to want to see it govern and they’re going to want to hold it accountable.

Voters in general do a good job of holding government accountable to their wants. Because of this, politicians tend to do what the voters want. For all of their reputation otherwise, politicians are actually quite good at keeping their promises. This seems to hold true irrespective of checks and balances. If you click through to the link, you’ll see that politicians follow through on their campaign promises roughly as often in the US (with a lot of checks and balances) as they do in Canada (which gives politicians a “get out of the constitution free” card in the form of the notwithstanding clause) or the Netherlands (which has the character more of a unicameral state than a true bicameral state like the US).

Faced with voters who have made demands and a need to achieve most of those demands if they wish to keep their jobs, politicians need to get things done. Checks and balances don’t change this simple fact. But checks and balances seem to have a lot of influence on how things get done.

During good years, the governments of Canada and America function relatively similarly. Legislation begins in the House, passes to the Senate, and is eventually signed by the Head of State. But during bad years the two countries look nothing alike. When America is wracked by partisanship or crisis, things still get done. But they get done by methods (read: kludges) that erode the very checks and balances that made them so difficult to do in the first place. Take this example (from The Economist): “The job of White House counsel was created to provide the president with sound legal advice; it has ballooned into a battery of lawyers—almost 50 under Barack Obama—whose task is to find legal cover for whatever the president wants to do”. America is a common law jurisdiction. Every justification President Obama made stick has become a permanent weakness in the nation’s checks and balances.

My observation is that both presently and historically, any attempt at checks and balances that makes governing too odious is eventually circumvented, ultimately leading to a government that is much less constrained than if it had instead been kept to less strict limits from the start. As specific examples, I provide Weimar Germany, the Roman Republic, and as we’ve discussed, the United States of America.

If you want durable limited government, then you have to be prepared to define limited government broadly and make your checks and balances effective only against extreme changes. Canada does a good job of it. Our system makes change possible but radical change incredibly difficult. The notwithstanding clause does allow the government to suspend certain parts of the constitution, but it has a deliberate sunset clause – any legislation passed with it lasts only five years, which is conveniently the maximum amount of time between elections. Actually modifying the constitution is prohibitively difficult, requiring the consent of the federal government and at least seven provincial legislatures (compromising at least half of the population).

A lack of serious resistance to reasonable policy proposals means that the Canadian Constitution and its checks and balances is at no serious risk of erosion. Meanwhile, the debate in America isn’t about whether the constitution is eroded, just about whether it is eroded to the point of ineffectiveness.

This is the choice you face with checks and balances. You can make them powerful and try to contend with the inevitable erosion they’ll face as politicians try and get things done in the face of them. Or you can make it easy for politicians to implement enough of their agenda to keep their voters happy and watch as they give up on any part of it that isn’t simple.

Viewed through this lens, Prof. Caplan’s suggestions are supremely harmful. They’re the sorts of things that will incentivize politicians to go hunting for any constitutional workaround they can find. I can’t think of anything that will hasten the demise of the American constitution more quickly than measures that make obstructionism easier.

All this is to say I think the rant I had when I first saw this article was justified:

The problem with checks and balances (as I keep articulating) is that they are fine in good times, but in bad times people need to be able to change stuff and you won’t have a supermajority to agree on what things to change. So you get people going from Point A (where we currently are) to Point B (where they want to be) in a way that is supremely damaging to the norms that we need for stable growth in the remaining 95% of cases.

If 51% of the population want something, they will probably eventually get something in that general area. We can control, however, where in that area they land. Do we want to restrain them temporarily, then have them get it in the most extreme form? Or do we want them to work within the system, be able to easily get it in a watered down form, and then dissolve for want of a remaining common enemy?

My overriding desire is for checks and balances that are reasonably effective over the lifetime of a nation. I support limited checks and balances because I truly believe that in the long run, this leads to governments being most constrained in their actions.

Epistemic Status: Model

Model, Politics

On Political Norms and Scandals

Prime Minister Justin Trudeau rang in 2017 with an ethics scandal. Electoral reform and President Trump might have pushed it out of the news, but it still bears talking about.

Maybe it’s just that my memory is fuzzy before 2004, but I feel like there was a point in Canadian politics when scandals weren’t a run of the mill occurrence. It seems like we’ve been treated to a non-stop parade of them since the sponsorship scandal. There was the In-and-Out scandal, that time Maxime Bernier left classified documents with his Hell’s Angels girlfriend, that horrible mess with Afghan detainees, the Robocalls (and associated criminal charges!), the F-35s, the senate, and now the Aga Khan.

There’s also been a host of minor scandals that didn’t even make it into this list, like the $50 million of G8 money spent to make Tony Clement’s riding prettier and that time the PM was rebuked by the International Commission of Jurists. To be fair, the governing Liberals have already had a few minor scandals of their own, like the cash for access dinner that maybe had something to do with a foreign bank getting approval to expand into Canada and Elbowgate (although the real scandal there may have been that we had to hear about it for days on end).

In the face of all of this, I have a confession to make. I haven’t cared about a political scandal in Canada since 2012. Trudeau could give $50 million to an ad agency while blowing up two gas fired power plants and performing lewd acts on a pig and I probably wouldn’t even bat an eye. And honestly, aside from a few partisans in the comments sections of online news, I bet you most Canadians feel the same way.

this is about what it would take for Canadians to be surprised
Not pictured: The pig

Why? Why the apathy in the face of serious and unethical behaviour?

I’d bet on a decade of scandals.

People aren’t outraged anymore because outrage requires norms to be violated. Unless Mr. Trudeau was doing cocaine off the seats of the Aga Khan’s helicopter, this latest scandal is just more of the same. If I lived in Toronto, even coke would be more of the same.

Rona Ambrose pulled off a passable imitation of outrage in question period (despite spending her winter vacation with a billionaire as well) and Mulcair is as angry as he’s always been, but for all their earnestness, they don’t realize that the Canadians who elected Trudeau aren’t going to turn on him without a truly novel scandal. Sure, he promised to fix Ottawa. But if we get one or two less scandals every four years, that will feel like a fix. For his voters to feel betrayed, Trudeau is going to have to do something really bad. And we just aren’t there yet. I don’t even think electoral reform got us there.

Critics called Mr. Harper “The Teflon Man” after no scandal would stick to him. They assumed that this was because of some aspect of Mr. Harper’s personality. They forgot to examine if the problem was larger than that. What if no scandal stuck to Mr. Harper not because of some strange and sinister aspect of his personality, but because of some failure of ours.

What if the lack of any truly damaging scandal during the Harper years wasn’t because of a deal he struck with Mephistopheles? What if it was because we no longer had norms left around that kind of scandal? What if the sponsorship scandal made petty corruption and favouritism boring and banal? What if we’ve lost the ability to expect better, so accept what we get as all we deserve? The answer to all these hypotheticals looks a lot like what we’re seeing right now.

Certainly getting people to even frown about Trudeau’s scandals is an uphill fight for the opposition. The people who voted for Trudeau like him –his approval rating is relatively high, at 48%– in spite of cash for access, his moving expenses, and the ridiculous hullabaloo about elbows. Trudeau’s supporters don’t seem to want to think bad things about him. Worst of all (for the opposition), a lot of Canadians think Trudeau is on their team.

Some of the time, politics is about teams. Our last election was fought on values, so it is perhaps more than likely that this is one of those times. When politics is about teams, you forget about the bad things your own team does, because you really, really don’t want to give the other team any ammunition.

When Harper was in power, the Conservatives didn’t want to hear anything about his scandals. They had a bunch of convincing reasons why they really weren’t that bad. I think they bought their own reasons too. The Conservatives were embattled, locked in what was to them a fight for the nation’s soul. Stephen Harper’s overriding goal was to change the relationship between Canadians and their government. In service of this goal, the Conservatives couldn’t afford to falter. They couldn’t afford to spend any time off of their message of economic stability. They had no time for contrition, not when there were elections to win.

But no time for contrition meant no time for reflection. And as is its eventual wont when out of power, the Liberal party engaged in a lot of self-reflection. They made their whole campaign about ending the nastiness coming from Ottawa. They built up a coalition of people who had been sidelined by Mr. Harper. And on the back of this team, they made their way back to power.

By casting things in terms of a fight for the soul of the country, the Conservatives gave the Liberals their best defense against scandals. They now have the boogeyman of Mr. Harper and his nastiness and cuts to brandish at any member of their coalition who thinks about leaving.

Currently this team is ascendant. So when the Conservatives crow about scandals, they find themselves offside on public opinion. Only the Conservative base wants to entertain the notion of scandals from Mr. Trudeau’s government, because serious scandals would mean the return of the Conservatives. For all who dread that eventuality, scandals must be ignored.

It gets even worse for the Conservatives though. Their earlier willful blindness means they’re just waking up to the fact that Canada is plagued by scandals and it hurts. I don’t know how they do it, but they have an amazing ability to fail to see any of the excesses of the previous government. Watch Rona Ambrose criticize Trudeau. She doesn’t make a single attempt to defend or acknowledge the previous government’s record. She just ignores it, as if it didn’t happen or isn’t worth mentioning.

Maybe this is a rhetorical trick that politicians can do and I’m naively falling for it. Maybe Rona Ambrose knows full well that Stephen Harper was no better than Justin Trudeau, but can’t acknowledge it if she wants her criticism to have teeth. If so, props to her as an actor. When I watch Rona Ambrose, I see a woman who believes that only she can see clearly. As far as I can tell, from her vantage point, Mr. Trudeau is violating a number of norms that the Conservatives resolutely defended for a decade.

But what Rona Ambrose sees as clarity, the rest of Canada sees as myopia. We know that the Conservatives were plagued by scandals for a decade and that the present state of affairs is certainly no worse than the previous one. So the Conservatives come across as sanctimonious. They’re behind the curve. And insouciance will always be almost synonymous with power (and therefore, in popular culture, with cool). The Conservatives care and they care visibly and impotently and a lot of the country can’t help but see this as weakness.

Even worse than being a loser is being a hypocrite and the Tories look like hypocrites as well. We all remember when the Tories were the ones explaining away scandals, not decrying them. Many people are angrier with the perceived hypocrisy than they are about the actual scandals.

Whether the ultimate reason is hollowed-out norms, team-based politics, or rage at Conservative hypocrisy, we’re in a shitty status quo. I didn’t want to find myself completely numb to Canada’s Prime Minister facing an ethics investigation.

Luckily, my partner Tessa still cares about scandals in Canada. And since I spend a lot of time talking with her, she prodded me about my breezy attitude towards scandals. She didn’t quite prod me into caring, but I at least I got to a state of meta-caring. I now care about my lack of care.

We got here because of norms. Political norms are a fragile thing. In Canada, they’re still damaged. We need a decade with as few scandals as possible to give them some time to recover.

I don’t think we’re going to get that. Not like this.

If we can’t count on politicians to police themselves, we have to find ways to make them accountable. We can’t let scandal norms become entrenched. I’ll let David Schraub explain:

There is an extraordinarily narrow range of levers through which one can be compelled to act in Washington: impeachments, being voted out of office, mandatory court orders … it’s not all that large, and it doesn’t cover all that much. Much of what we take for granted our government will do is not legally compelled, but is based on politicians following established patterns of political culture. Among those patterns is that a major scandal will lead to an investigation and some measure of accountability. But nobody forces Congress to launch an investigation, and nobody forces administration officials to resign or even acknowledge scandals reported in the media.

David is writing about America where the situation has become particularly dire. But Canada is also seeing norms deteriorate. That’s why we have to fight for them. Even in mild-mannered Canada, accountability can’t exist without a public outcry.

We have to call or write to politicians and tell them when we’re displeased. We have to press them to acknowledge their mistakes and promise not to repeat them. And we have to be prepared to vote against politicians who won’t desist, even if we like them, even if it means our side losing sometimes.

I’m not at the point of committing to vote against Mr. Trudeau. I think it’s still amateur hour with the Liberals and that they deserve some time to get over their learning curve. I think a lot of Canadians are in the same boat. I bet that Trudeau’s approval rating stabilizes or bounces back over the next few months instead of continuing to fall [1].

So the Liberals have some time, but they don’t have all the way to 2019. You get your first election on promises. The second has to be backed up by results.

Even though I think Trudeau and his Liberals deserve some time to sort themselves out, I don’t think they deserve my complacency. I’m going to post about every scandal, whether it’s on social media or on my blog. I’m going to talk with my friends about the importance of political norms and my displeasure with scandals. When I feel important norms are being violated, I’m going to write to my MP. I’m now committed to actively standing up for our norms, no matter who is in power.

Will you join me?

Epistemic Status: Model

[1] I’m expressing this guess as three ranges I think the approval rating will fall into on the first survey conducted after April 1st, given that the approval rating is currently 48% ^. The ranges and my associated confidence in them are:

  • 38-60% (90% confidence)
  • 42%-57% (70% confidence)
  • 46%-54% (50% confidence)
Advice, Software

A Dialogue on Lateral Thinking

Cast: The Hare (interviewer #1), The Coyote (interviewer #2), and The Tortoise (interviewee).

Hare: Okay, that wraps up the technical portion of the interview. Now we want to ask you some lateral thinking questions.
Tortoise: Lateral thinking questions?
Coyote: You know, questions that challenge your ability to come up with non-obvious solutions? Or when you find a solution by throwing out all your assumptions? Here at Acme Corp., we pride ourselves in coming up creative solutions to problems.
Tortoise: Okay…
Hare: We’ll start off easy [1]. Acting on an anonymous phone call, the police raid a house to arrest a suspected murderer. They don’t know what he looks like but they know his name is John and that he is inside the house. The police bust in on a carpenter, a lorry driver, a mechanic and a fireman all playing poker. Without hesitation or communication of any kind, they immediately arrest the fireman. How do they know they’ve got their man?
Tortoise: Hmmm… Well, if they can immediately tell the profession of everyone in the room without saying anything, presumably everyone is dressed in clothes emblematic of their profession. I think the jumpsuits firefighters…

The Hare snorts in amusement.

Image Credit: MaxPixel

Tortoise: As I was saying, the jumpsuits that firefighters wear often have nametags on them. So they arrest him because he’s wearing a jumpsuit that says “John”?
Coyote: Not quite right, I’m afraid. You see, all the other poker players were women.
Tortoise: I don’t really see how… You aren’t next going to ask me the one about the surgeon operating on her son are you? This isn’t secretly the HR section of the interview?
Hare: No, you’re already through that.
Tortoise: Because I assure you, I do think women can do anything men can.
Coyote: We weren’t questioning that at all.
Tortoise: It just seems weird to have a problem that hinges on interpreting fireman in an archaic way. I’m given to understand that it’s essentially gender neutral now, although I thought firefighter was preferred… Sorry, this is a tangent. You said you had more questions.
Hare: How about you ask the next one Coyote.
Coyote: A man lives in the penthouse of an apartment building. Every morning he takes the elevator down to the lobby and leaves the building. Upon his return, however, he can only travel halfway up in the lift and has to walk the rest of the way – unless it’s raining. What is the explanation for this?
Tortoise: That’s a really odd way for an elevator to work. What time period is this?
Coyote: That isn’t important to the question.
Tortoise: Okay, I know this is a digression, but how can more information not be important to the question? Like this is basic scientific method. You formulate a hypothesis that is capable of being falsified by the world. Then you see if the world falsifies it. And you keep doing it until you have a hypothesis that doesn’t get proven false. It’s basically the same as test-driven development, come to think of it. Didn’t you just ask me if I did TDD?
Hare: We do TDD. We also write code on computers, not whiteboards. The interview doesn’t map quite perfectly to the job, but we have found how candidates perform on these questions often correlates well with their later performance.
Tortoise: Okay, if you say so. Um… maybe the elevator uses water as a counterweight, but the counterweight leaks a lot and only really works properly when it rains? I really don’t have a better answer.
Coyote: Nope. The man is a little person. He can only reach the button for his penthouse when he has an umbrella with him.
Tortoise: Couldn’t he just carry an extendable pointer with him and use that?
Hare: If it’s any consolation, I got this one wrong too. Now I like to visualize the guy as the kind of person who is so lazy he won’t do anything he doesn’t absolutely have to, even if doing it would save him time.
Coyote: Hare keeps insisting that getting this one wrong is a badge of pride, because it proves that you have a huge inferential gap to that kind of laziness.
Tortoise: That’s the most reasonable thing I’ve heard in the last five minutes.

Coyote snickers nervously.

Image Credit: Jitze Couperus

Coyote: Um… Well, one last question: Assume there are approximately 7,000,000,000 (7 billion) people on Earth. What would you estimate to be the result, if you multiply together the number of fingers on every person’s left-hands?
Tortoise: Oh wow, that’s… oh, I see! Zero. There’s plenty of people with no fingers, so the result has to be zero.
Coyote: I’m glad you got that one. When people try and write out a solution the whole thing gets out of hand.

Tortoise groans.

Image Credit: Eric Kilby

Hare: Actually, that brings up an interesting point. Are puns a form of lateral thinking?
Coyote: I don’t think so. I think puns are more of a brute force thing. You know roughly what you want to say and what you’re punning off of, so you can do a brute force search in your head for likely combinations.
Tortoise: Puns feel closer than those questions. When I think of lateral thinking, I think of a much less constrained solution space. I felt like those problems just tested my ability to be clever in the exact right way on command.
Coyote: What does having one answer have to do with it?
Tortoise: Well look at your first question. A priori, I don’t think any reasonable person would consider my hypothesis less likely than the “correct” answer. If I was in the room with the police, then obviously my answer would reflect reality less closely than the answer that John is the only man in the room. But without that information, it really is just luck if you stumble onto the one right answer. Anchoring effects mean it’s hard to generate multiple plausible answers. And because you’re anchored onto the “right” answer, you view my answer as obviously incorrect.
Hare: You honestly did fine on this section. You really don’t have to argue about your score.
Tortoise: It’s not about score, it’s about the principle of the thing! Even if lateral thinking is a useful quality in a software developer–
Hare: What makes you think it isn’t?
Tortoise: Lateral thinking, by definition, involves unexpected solutions. Answer me honestly: do you like reading code where someone did the unexpected thing? I’d rather an ugly but conventional solution to an unexpected one that’s “elegant”. I mean, there’s a reason no one uses Perl any more, right? “More than one way to do it” is a horrible philosophy for code that has to be read by many people!
Coyote: Tortoise has a point there, Hare. I remember when Perl programmers used to brag to me that they could write a web server in one line of code. It made me think of the scientists in Jurassic Park. Just because you can do something doesn’t mean you immediately have to go out and do it.
Hare: I don’t think programmers need to use lateral thinking all the time. But aren’t there often places where an unconventional solution saves a lot of time?
Tortoise: Name one.
Hare: What?
Tortoise: Name one unconventional solution you’ve implemented that saves time. Bonus points if you don’t need five lines of comments to keep everyone else on the team from being confused by it.
Coyote: I actually have one! I had to round to the nearest half to enable functionality that the designer wanted. I considered this horrible concept with recursive if-statements. But then I found out there was a much better way to do it. You just multiply your number by two, round, then divide by two. It’s one line and perfectly clear if you think about it for a second.
Tortoise: So why didn’t you ask me about that?
Hare: We have an interview script. We can’t just change whenever one of us comes up with a clever solution to a problem.
Tortoise: Then pick a few examples in your code where you actually had to think laterally and give them out as worksheets. At least that has a chance of accurately measuring our lateral thinking. And if we get it right, you have a convenient baseline in Hare here.
Coyote: Actually, I didn’t come up with the solution.
Hare and Tortoise: Stack Overflow?
Coyote: Stack Overflow [2].
All pause to consider this.
Coyote: Huh. So if I understand you correctly Tortoise, your point is twofold. One, lateral thinking isn’t a very important quality in a programmer. Doing things conventionally is actually more useful, because as Guido van Rossum pointed out when he designed Python, code is read more than it’s written.
Tortoise: That’s a really good way of phrasing it. How would you phrase my second point?
Hare: Actually, can I try this one?
Tortoise: Certainly!
Hare: The way we’re testing candidates on lateral thinking is pretty much invalid because there’s only one right answer and it’s the one we’ve already decided on. A better test of lateral thinking would be to give candidates some of the problems where lateral thinking was actually useful to us. And if we do this, we should be open to solutions that are different (or even better) than ours. Like your jumpsuit solution of the John problem.
Tortoise blushes
Tortoise: I think that sums it up well. And thanks.
Coyote: One thing I still feel weird about is treating lateral thinking as not that important.
Tortoise: I may have overstated my opposition a bit. I don’t think it’s unimportant. But I think there are many other skills that are more important and easier to test. Technical writing is a good example. It’s actually pretty easy to test someone’s technical writing and documentation abilities in an interview and that’s a skill a good programmer will use every day.
Hare: Or time estimation! I bet we could incorporate that really easily into the blackboard task.
Tortoise: Oh, that’s a really good one. I’m definitely stealing that for interviews I conduct in the future.
Coyote: I’ve got it!
The Tortoise and the Hare stare at him blankly.
Coyote: I’ve realized why I felt so bad about us not being enthusiastic about lateral thinking. I read a lot of science fiction and a lot of my heroes (like Ender, Breq, and Miles Vorkosigan) use lateral thinking to win. And you know how it is… when you read books like that, you want to imagine you’re like the hero. Admitting that lateral thinking isn’t that important to programmers feels like admitting I’m not like them.
Tortoise: Actually, I think Ender is the best example of convention being useful. A lot of what Ender did with Dragon army didn’t involve lateral thinking so much as it involved taking what Ender already knew worked from all of his experiments with his launch group practice sessions and putting it into systems where others could easily use it. This is the sort of thing you see really great programmers do. They make everyone else’s job easier by designing really good systems that transfer well. It was systems that let Ender win all of his battles but the last
Coyote: That’s a different way of looking at it.
Hare: It seems like in a lot of fields, lateral thinking is what makes people great, instead of really good. I’m think scientists and pro athletes, for example. And maybe that’s true in ours as well. But it also might be that what makes a really great programmer is their ability to build up systems that other members of their team can use.
Tortoise: And lateral thinking does help! Whoever it was on StackOverflow that came up with that answer, or whoever taught it to them, that person had to do some lateral thinking. And they’ve ended up saving a lot of programmers a lot of time with it.
Hare: But if we’re counting up total time saved, I guess Stack Overflow takes the cake, doesn’t it?
Tortoise: Yeah, Joel Spolsky and Jeff Atwood definitely showed how systems are incredibly useful to programmers.
Hare: Well now that that’s all cleared up, we’re actually out of questions. So do you have any questions for us?
Tortoise: So about your benefits package…


[1] All puzzles in this dialogue taken from http://www.folj.com/lateral/ ^

[2] Specifically this question: http://stackoverflow.com/questions/6137986/javascript-roundoff-number-to-nearest-0-5 ^