Pitching Conservatives on Ditching the Police State

During Mao’s Cultural Revolution of the 1960s, the Red Guards were a radical student-formed paramilitary organization. Inspired by the push to recreate a new communist China and destroy the “Four Olds”, the Red Guards targeted anyone deemed sympathetic to intellectual or bourgeois ideas. Red Guard tactics quickly devolved into property damage, violence, and torture. Thousands were murdered. The Red Guards were both condoned and condemned by the central government, but they were unable to be controlled. The movement in some cases spiraled into a civil insurrection that was eventually defeated by the army in 1968. Lawlessness, destruction of private property, disregard for human life–clearly, conservatives and I agree that the terror campaign of the Red Guards was a moral abomination. But I have a question for conservatives:

Were the Red Guards bad because of their ideology, or because of their violence?

Well, what if we kept the ideology, but got rid of the violence? Apparently, the “Party for Socialism and Liberation” is an American political party that advocates communism. Their positions (perhaps unsurprisingly) are rather odious, but as far as I can tell, they are not actively going around the country murdering small business owners for owning capital. This is clearly morally superior to the Red Guards.

What about the reverse? What if there were paramilitary organizations in the United States operating outside the law, but they happened to not be communists? What would their moral standing be? They wouldn’t target people holding onto traditional values, like the Red Guards did, but that would not make citizens feel better that hundreds of people were being murdered.

American police forces are Red Guards without communism.

Too harsh? Let’s take a look into this phenomenon where we’ll find outrageous situation after situation. In 2011, cops killed a former marine with no criminal record, Jose Guerena, while his wife and children were hiding in a closet from the unidentified intruders. Guerena, while armed, never removed his safety from the gun. Naturally, he was hit over 20 times by police. No evidence was found in his home of any illegal activity. The warrant was served suspecting that Guerena was selling marijuana. At the time medical cannabis was legal in Arizona, but someone selling it without a license? Better have a no-knock raid. Police settled out of court for civil damages, and a County Deputy is quoted saying ” the officers performed that day in accordance with their training and nationally recognized standards”.

In 2013, three off-duty police officers working as security guards in a Frederick County, Maryland movie theater were asked to remove a patron who was attempting to see a second showing of a movie without paying. The customer, Ethan Saylor, was 26 years old and had Down Syndrome. The officers refused the help of his aide, arrested Saylor, and in the process, fractured his larynx, resulting in his death from asphyxiation. A grand jury found no wrongdoing, and I could find no citation indicating the officers had lost their job. The family settled for $1.9 million (the judge did decline to extend qualified immunity for the officers).

In 2015, police destroyed a Colorado man’s home in pursuit of a shoplifter armed with a handgun. When I say destroyed, I mean “… the tactical team bombarded the building with high-caliber rifles, chemical agents, flash-bang grenades, remote-controlled robots, armored vehicles, and breaching rams”. The house was condemned afterwards leaving Leo Lech homeless. The city compensated him with $5000. He took the city to court and an appeals court ruled in 2019 that he was entitled to no compensation, as the police were acting within their police power, not taking items as part of an investigation.

I could go on and on: a homeless person beaten to death by officers, Massachusetts state police using military helicopters to spot single marijuana plants, a retired unarmed Sunday School Teacher shot four times in her car (cop lied on his report, was convicted of manslaughter, served two years), but there was one final story that stayed with me.

In May 2014, one night just past 2 AM, police in full SWAT gear served a no-knock warrant in a small Georgia town. A roommate of an informant they had never used before had apparently bought methamphetamine at the house earlier in the day. Ignoring the minivan parked in front with the car seat in it and the kid sized play pool, police assumed no children were present in the house despite no actual surveillance having been conducted. Their target, it turned out, wasn’t present and when he was apprehended later in the morning, he was not armed. Nonetheless, they easily obtained a no-knock warrant on the flimsiest of information. Police broke into the house and threw a flashbang grenade inside where no less than four children under the age of ten were sleeping. It landed in the playpen of the youngest, a 19 month old baby, and exploded. The police found no contraband or illegal items, but the infant was put into a medically induced coma. A grand jury declined to indict the deputy who obtained that warrant (naturally), and the county paid out a multi-million dollar out of court settlement.

American police have been a threat to freedom for a long time and in many forms. Violent no-knock raids on unsuspecting families, drug enforcement that both fails to stop drug use while also stacking up bodies, executions of unarmed and nonthreatening citizens because they don’t obey police orders, burning infants, the stories sound unbelievable. They are clearly an out of control, unaccountable paramilitary force.

Let’s talk about the recent protests, the catalyst for this post. Left leaning protesters have focused on police abuses’ connection to racism. I stated earlier that American police are Red Guards without the communism. Leftists are extending that argument; instead of Red Guards murdering people for the Cultural Revolution, protesters point out that American police are murdering people due to systemic racism. John Oliver details this argument here. Libertarian critiques of the police state have tended to categorize the ideology of the police in terms of authoritarianism, or opposition to personal liberty; hence the focus on the enforcement of drug laws that infringe on individual autonomy.

Naturally, libertarians emphasize a narrative where police abuses can be counteracted by libertarian ideology while progressives emphasize an alternative narrative that can be solved by social justice. If you are a conservative, you are likely to be suspicious of these critiques as they seem self-serving. Nonetheless, both critiques have a solid basis: Black Americans are killed at a disproportionate rate, and many unnecessary deaths clearly occur when serving drug warrants on citizens who have done nothing violent. Moreover, compared to the Red Guards, American police are not nearly as ideologically cohesive, yet they remain a powerful and unaccountable force as we’ve seen. This creates a dangerous situation where many ideologies and interest groups have an incentive to influence the police for their own ends.

But let’s recall what we stated earlier. Are unaccountable paramilitary groups bad because of their ideology or their violence? I argue, it’s their violence.

And we’ve seen from the previous examples, American police are remarkably violent. But those were just anecdotes. Here is part of a table of countries, and highlighted in blue is the rate of police killings per 10 million people. The United States at 46.6 is surrounded by renowned criminal justice systems like Iraq and the Democratic Republic of the Congo and trails standout nations like Iran.

Wikipedia link

This is bad.

We don’t actually know the amount of people killed by the police in the U.S. because the government doesn’t require police departments to track that data. The data linked in the Wikipedia article is from the Fatal Encounters database, which suggests about 1800 people killed every year by American police. The Washington Post database focuses only on police shootings and indicates around 1000 people are shot to death by the police each year.

To drive the point home, let’s first consider developed countries. The United States is by far the worst developed country in terms of police killings per capita, but the distant second place is Canada (which is so far up on the table, I couldn’t include it on the screengrab). Canadian police kill around 10 people per 10 million population compared to America’s 46.6–nearly a five fold decrease. In fact, American police are significantly outperformed by those of Pakistan, a country where military coups are commonplace, and which only had its first peaceful transition between elected governments in 2013.

The Washington Post database suggests about 25% of those killed by police are Black, which is disproportionately high for their percentage of the U.S. population. But for the sake of argument, let’s pretend that we could wave a magic wand and prevent all Black killings by the American police. Even if we did that, our rate of police killings would remain 350% of the next highest developed country. And to be clear, Canada is the second worst developed country we have data on, most others are much better.

I want to reiterate that last point: American police killings are not 350% of Canada’s total police killings, but 250% higher deaths per capita even if there were no more Black victims of police violence.

In fact, even if this hypothetical scenario of drastically reduced police killings, the rate of American police violence would remain much worse than a country like Egypt’s. In 2018, Human Rights Watch wrote of the Egyptian election:

Since President Abdel Fattah al-Sisi secured a second term in a largely unfree and unfair presidential election in March, his security forces have escalated a campaign of intimidation, violence, and arrests against political opponents, civil society activists, and many others who have simply voiced mild criticism of the government.

This is the country that has a significantly less deadly police force than us. This is disgraceful.

So if you’re a conservative, and you feel left-wing (or libertarian!) activists have a differing ideology than you, you’re probably right. But I think you may have much more in common on this issue that you might initially believe. The question we must agree on isn’t “which ideology should an unaccountable paramilitary force within our borders have?”, the question is “do we want unaccountable paramilitary organizations murdering hundreds of citizens a year?”

If think you might have some common ground with reformists, here are some simple ways to make the police more accountable:

A 2020 Policy Platform Proposal

It’s election season so it’s time to start talking electoral politics again. The Trump administration has been particularly successful in ignoring policy discussions in favor of political point scoring. This isn’t too surprising given Trump’s lack of consistent ideology, apart from perhaps opposition to free trade and immigration. Impeachment has also helped focus attention on Trump’s political situation rather than his policies or lack thereof. Don’t get me wrong, I think there is a strong non-policy case against Trump, and I think in particular Congressman Justin Amash has done an excellent job in articulating why Trump’s behavior is concerning.

However, I think there is also a policy-based critique of Trump. In order to properly make that case and compare Trump’s policies to Biden’s or other candidates, we must establish a foundation declaring which problems are most important, and what policies could be used to implement them. Criteria for these policy ideals include some utilitarian calculus, i.e., how to improve the lives of the most people in the largest way. Thus, the first of these policies is actually a meta-policy, a way to improve congressional power to pass laws and run the state. Changing the way we make policy can affect all of our future policy making.

Countering this interest in utilitarian idealism is a preference for some political feasibility; in other words, while I might prefer to emphasize revolutionary changes that significantly improve the country (changing all of our voting systems to approval voting or quadratic voting or switching taxation to be based on land value), I’ve left them off this list because they are not just unpopular, but in fact virtually never discussed. If you find a particular policy interesting, please follow the links in that section for additional policy discussion and details.

Finally, there is uncertainty here, and I’ll mention other policies that didn’t make this list at the end. Trying to filter major talking points out of a broader range of political ideas is difficult. Policies and political philosophies are interconnected, and where I’m drawing boundaries must be arbitrary. Nonetheless, these ideas should form a good basis for uniformly judging candidate policies.

Congressional Power

Any policy platform has to address the fact that our current system for governance, for crafting and enacting policy, is deeply flawed. We have uncompetitive and broken elections, we have bad ways of choosing candidates, and we have too much power in the executive branch. Executive authority compounds our problems by making each election a stark singular choice between polarized sides instead of a well rounded government built on a legislature with many interests represented. I can’t fix all of these in this policy platform, so improving the balance between the president and Congress seems like a good place to start.

The entire budget for the legislative branch, including congressional staff, offices, and congressional agencies like the GAO and CBO, is about $5 billion. Congress is then responsible for oversight and legislative action for the entire $5 trillion federal government. The CBO has a mere 250 person staff, and it can’t even research and score all Congressional bills. This is absolute insanity.

Congress needs to be able to wield its muscle. It should not be relying on executive branch bureaucracies as unbiased experts evaluating their own performance. It should have a better staffed research arm which can oversee all aspects of the massive American bureaucracy. Congressmen also need to have more and more policy-focused positions on their own staff, along with fewer committee assignments. National Affairs has an excellent in depth discussion of the thinking behind this brief overview. Legislators are currently underpaid amateurs who spend half their time outside of Washington focused on other things besides governance. This does not allow for knowledgeable congressional oversight of the federal government.

Cato also has some excellent ideas for strengthening Congress such as having a standing committee to review executive overreaches from statutory law, and forcing votes on major rules as implemented by regulators or bureaucrats. Other ideas include expanding the congressional calendar, making a new Congressional Regulatory Service to oversee the regulations made by independent and executive agencies, and requiring all civil asset forfeitures to be deposited into the Treasury to be spent by Congress, not the executive.

Unfortunately, even despite a recent impeachment trial this is simply not a major political issue in this year’s campaign, and no candidate is running with strengthening Congress as a priority. In fact, there are essentially no meta-policy ideas being floated. Yet ideas are not hard to come by!

Liberalizing Immigration

The U.S. immigration system is terrible (see section 8 here). It is esoteric, slow, and requires a complete overhaul. It should have a focus on a merit-based system rather than nation-of-origin and family ties as it does now. It should be simpler for high-skilled workers to be hired by American companies and it should definitely be easier for young workers, educated at excellent American colleges, to be hired by American companies and remain in the United States where they can pay tax dollars for decades.

Why is this so high up on the list?

This is a matter of national security. China is a growing power, but crucially, it cannot expand its influence or economy through immigration. The Chinese state has largely decided that ethnicity matters, and China is not seeking to create a multicultural amalgamation to improve the world, but rather a nationalist state. The U.S. isn’t restricted in this way; anyone can be an American. Immigrants are also more likely to start businesses and take risks. That means the most creative and ambitious people in the world can come to the United States and contribute to our culture, knowledge, technology, and wealth. Moreover, these remarkable people already want to come here. Increased dynamism and economic growth also makes the rest of our geopolitical challenges easier; it means the national debt is less of a burden, and national defense spending can be higher in absolute terms while costing less of a percentage of GDP.

This is also perhaps the best and simplest way to improve the world quickly. It’s extremely difficult to improve nations with poor institutions, yet people who struggle in developing nations can be immediately more productive if they are transplanted to the U.S. And of course many are quite willing to do so, uprooting their entire lives for a chance at the American Dream. We can pursue limitations on their access to public money, or a simple tax upon immigrating, but nonetheless we should be voting to improve the world in the most altruistic and nationalistic way possible: expanding legal immigration in order to make more Americans!

Federal Incentives to Build More Housing in U.S. Cities

This is a specific policy taken from the Niskanen Center’s Will Wilkinson. Cited on this blog before, he suggests giving federal money to urban areas that add large amounts of new housing stock. Why? Because American cities are absurdly expensive to live in, yet new housing is extremely difficult to develop due to overregulation and zoning laws.

The impact of our poor housing policy is enormous. Economists suggest housing constraints have lowered U.S. GDP by as much as a third over the last 50 years. Think about that. We could be missing a third of GDP because millions of people who wanted to move somewhere for a better job couldn’t find a place to live. It’s clear that the most productive areas in the U.S., especially cities like New York and San Francisco, are prohibitively expensive, keeping out potential new productive workers.

Wilkinson’s suggestion isn’t the only possible policy solution; another is to change zoning to be hyper local, composed by residents of a single street or city block. This would allow experimentation and innovation, instead of immovable local land interests which keep out future non-residents who can’t vote in today’s elections.

While the viable solutions are still up for the debate, the impact is clear: the lack of housing development in U.S. cities due to overregulation may be the single greatest barrier to economic growth, thus earning its inclusion on this short list of policies.

Decriminalization of All Drugs

Ever since Pete Buttigieg announced his support for this policy, I’ve had it circled for inclusion on this list. The War on Drugs has been a colossal failure, has not reduced drug use, and has radically increased prison populations. There have been extraordinary costs to the taxpayer in both civil liberties and assets. Massive application of state force has helped to give a monopoly in funding to the most bloodthirsty and gruesome organized criminal elements in the world, including terrorists. There have even been spillover effects as governments crack down on prescription pain killers, leaving patients in agony.

This policy is wrong morally, practically, and economically. It is not the place of the state to determine what substances informed adults can consume or inject. It is also abundantly clear the state has zero capability to halt the trade or consumption of drugs. Rather, enforcement of drug laws have bolstered a black market where information is asymmetric and scarce, endangering all involved. The only thing the state has succeeded in doing is making organized crime more financially viable. The resulting conflict in Mexico has killed over 150,000 people, making it one of the largest conflicts of the 21st century behind only the Iraq War, Syrian Civil War, and Darfur. It is this monstrous loss of human life as a result of changeable government policy that places this item so high on this list.

And of course it goes without saying that this massive assistance to organized crime is occurring at great financial cost. Estimates for enforcement, prosecution, incarceration, and military interventions are as high as $50 billion a year. State prohibition of private mutually consensual transactions also requires erosion of our rights in ways that frustrate measures of concrete financial cost. The ACLU notes extensive surveillance has been justified under the guise of drug enforcement while increasingly militarized police forces have abused their power to break into homes unannounced or preemptively shoot victims all in the name of stopping transactions among consenting adults. It’s time to end this failed policy.

Catastrophic Risk

It’s clear today that the federal government does not respond well to large disasters. Perhaps too much relies upon the whims of the executive who happens to be in power, but it seems likely that we could institutionalize better responses to catastrophic events. Yes, this includes pandemics, but also major earthquakes, solar flares, artificial intelligence, and even plans for averting nuclear war (for a more detailed analysis, read Toby Ord’s recent book, The Precipice).

This is a highly neglected problem and thus one of the highest impact policies we could undertake (climate change could go here, but it has not been quite as neglected a topic as other risks, so I’ve detailed it later). At the beginning of 2020, I would not have included this in the list of top policies, not because it was low impact, but simply due to the fact that it was not discussed as a major political issue. The failings of the federal government to respond to a deadly virus have pushed catastrophic risks into the mainstream. While the likelihood of any given catastrophe is low, it is the enormous impact of the tail-risk that should concern us; preparing now will mean the difference between devastation and mere hardship.

We should look to create public commissions to investigate our preparedness for various catastrophic events, identify what can be done now for relatively small budgets with larger payoffs when a disaster comes, and then pass legislation that enshrines this knowledge institutionally in ways that do not rely on the whims and competence of whomever happens to be president. It is vital that any commissions include our preparedness for other challenges besides pandemics; preparedness for unexpected events is not selected through democratic pressures, and perhaps this has resulted in our current difficult situation with COVID-19. It would be wise to use this opportunity to prepare not just for the next viral outbreak, but for other unlikely events as well.

Other Topics

There are arguments for inclusion of a lot more policies. I’ll run through several more quickly.

It matters a lot who the president appoints to the Federal Reserve, and that they are extensively qualified and independent. I’ve left it off of this list mostly because we’ve lucked out and it seems Trump’s appointments haven’t been that different from normal. When odd choices were floated, they were largely quashed. Independence is obviously still at risk with the president tweeting criticism of his own appointees, so this issue shouldn’t be overlooked, but given that I treat it like a pass/fail grade, we can reasonably hope this will be a “pass” for all candidates in 2020. I wish I could say that more definitively, but I can’t.

Healthcare is a huge part of the federal budget and has an outsized impact on the economy. We also don’t have great solutions, but this is another issue that could easily have made the list. The most important aspects are stopping reliance on employers providing health insurance (which makes it much harder for workers to take risks and switch jobs), and expanding coverage for the least well-off. How we do that is difficult to answer in such a small space, but I’m wary of radical changes that seek to quickly re-imagine the U.S. healthcare industry from the top down.

Climate change is a potentially expensive disaster waiting to happen. If the past months have taught us anything, waiting for disasters to happen is not the correct strategy. Instituting a small carbon tax seems like a good place to start. It can be refunded to taxpayers equally, or even made to incentivize carbon sequestration programs with refundable tax credits for carbon taken from the atmosphere.

Free trade has had a massive impact on reducing poverty worldwide, while also improving the economies of all countries around the world. There’s also some evidence for reduced chances of wars between important trading partners. Aligning American and Chinese commercial interests through trade will be a vital part of avoiding a war between these world powers. Free trade is also a vital vehicle for continuing the pattern of global poverty reduction seen in the last 30 years.

U.S. interventions in the Middle East have been one of the largest contributors to excess deaths from U.S. policy. Obviously there is high uncertainty over whether many conflicts would have continued even without American intervention, but that seems unlikely in at least a couple large instances (the Iraq War being the biggest one). U.S. support of regimes like Saudi Arabia also seems to show negative payoffs from a humanitarian calculus. It also does not seem that larger 21st century goals like opposing authoritarianism in China and avoiding large scale wars are served through Middle Eastern interventions.

Candidates’ Priorities Matter Too

While this is a nice policy platform, ultimately the goal is to judge candidates by their relationship with these policies.

A major problem for this approach of separating out policies isn’t that most people running for office oppose these positions, but that they might be indifferent or even positive on these high impact policies while still focusing on other completely radical ideas. Elizabeth Warren’s many proposals come to mind here. There are some meritorious critiques in Warren’s proposals; competition is vital to a well functioning market, and some of her ideas could enhance competition. But many are far more radical with, at best, unknown effects on competition and the economy generally. These include the eradication of private equity, the changing of corporate boards, and an unprecedentedly large wealth tax which could significantly curtail investment. If Warren scored highly on the top policies put forward here (she does alright on immigration, housing, and drug policy), how do we balance that with the relatively radical (and I’d argue unhelpful) economic proposals she made the centerpiece of her campaign?

Unfortunately, we have to take those points seriously and note that while I have tried to rank these policies in a somewhat utilitarian, impact- centered way (policies within the Overton Window that help the most people by the greatest amount), radical policies that backfire could have very high impacts that shove aside the ideas proposed here.

And that goes for both parties. If Trump did well on these policies (unlikely, yes), but then also centered his campaign on radical ideas like defaulting on the national debt, shutting off the internet, or throwing away nuclear arms control treaties, then not implementing those policies might become the highest impact.

There is a lot of uncertainty that remains; some of these policies could be higher on the list, and I’ve likely excluded some that are high impact that have not yet occurred to me. Major policies could matter in the future that we just haven’t encountered. And of course these are only policy preferences; as noted in my last post, simple competency is an important factor as well. Despite all of these caveats, this an important step in laying a foundation of policy discussion and analysis against which we can measure candidates. Electoral politics is messy and tribal; discussions confound concise and consistent frameworks, but when they do swerve towards policy, these points should help form the questions that need to be asked.

We’re All State Capacity Libertarians Now

Tyler Cowen kicked off this year with a heavily discussed blog post defining what he calls “State Capacity Libertarianism“.

I didn’t get it.

Cowen described a moderate libertarianism while adding the nebulous concept of “state capacity”, in particular noting that a capable or powerful state was different from a tyrannical state. I read the words, but I assumed he was describing a “left libertarianism“, where a typical position might be to push for the government to collect efficient taxes, and then distribute those to everyone via a basic income, and avoid otherwise messing in the economy. I’m open to engaging with something like that, but I didn’t grok how Cowen’s take was different or what this view offered.

And then there was the worst pandemic we’ve seen in the 21st century. Huge swaths of the American economy have been shut down, and this disease is still likely to kill the equivalent of an entire year’s worth of influenza fatalities in a single month. Pandemics are an obvious market failure scenario, with difficult to control externalities, and thus we have state institutions to deal with this and pick up the slack. They failed.

Policy Failures

I looked up my first tweet about COVID-19 (back then it was just novel coronavirus or nCoV). It was on January 30 in a reply to Robert Wiblin. My first standalone tweet was a few days later:

Now, I wasn’t warning people that this would be bad or telling people to start prepping. But it was certainly something occupying my mind, as I tweeted about it a dozen times in February. I did start mildly stockpiling food and household items starting in mid-February. Ultimately, I consider my actions to be a failure. I didn’t short the market (although I didn’t buy any index funds either), I didn’t warn enough of my close friends and family to start gathering supplies, and I didn’t advocate loudly for changes in government policy. I was partially concerned about looking alarmist, which in retrospect was a really silly thing to be concerned about.

Nonetheless, this blog is a hobby. I have a day job which isn’t concerned about government policy. My question is about the people whose day job is pandemic preparedness. If an anonymous blogger with a few hundred twitter followers can be concerned about a possible pandemic, then where were the people who are paid specifically to deal with this eventuality?

Why didn’t the CDC and FDA start a program to quickly approve new COVID testing in mid-February instead of mid-March? In an exponential growth situation, that is a long time. Why didn’t they start pushing hospitals to create new isolation wards in February? Why didn’t the CDC start putting together lists of events it strongly urged local governments to cancel when the first case appeared at local hospitals? Why did the NBA have to unilaterally suspend its season only after a player was diagnosed in March? Why was there confusion between who was in charge and what steps should be taken?

Beyond lack of distancing and lockdowns, there were several other failures. Firstly, the CDC created their own test which differed from previously used ones in the world. It was shipped to public labs around the country and…didn’t work. The FDA compounded the problem, disallowing any tests except the one the CDC had created. The University of Washington virology lab was able to create their own test and applied to the FDA for approval. The FDA said it couldn’t approve the test until the lab had demonstrated it wouldn’t return false positives for other dangerous coronaviruses. Perhaps this would be a good idea in more normal circumstances, but it was an asinine requirement if you were trying to head off a pandemic where every day saw exponential growth in a virus spreading across the country. The lab started using the test through a research loophole, but the story gets crazier:

Still, Greninger complied. He called the CDC to inquire about getting some genetic material from a sample of SARS. The CDC, Greninger says, politely turned him down: the genetic material of the extremely contagious and deadly SARS virus was highly restricted.

“That’s when I thought, ‘Huh, maybe the FDA and the CDC haven’t talked about this at all,’” […] “I realized, Oh, wow, this is going to take a while, it’s going to take several weeks.”

By this point, there were already over 50 confirmed cases of COVID-19 in the United States and still, nobody but the CDC was permitted to conduct testing.

It was on February 28th, a full 8 days after the UW virology lab had applied for their test to the FDA, that the agency finally allowed other tests besides the CDC’s. Of course, most other labs hadn’t started making a test yet. What a disaster.

What’s interesting to me about the testing fiasco was its independence from President Trump. I think Trump is a bad leader, incompetent, unintelligent, a bad judge of character, and has terrible policies, but ultimately the point of a capable bureaucracy would be to have experts on hand with the proper plans in place to implement them regardless of political leaders. We shouldn’t have to recreate all the accumulated knowledge of the government every time there is a new president. Here, Congress had created agencies to oversee health and pandemic responses, the scenario arose…and then those agencies actively made things worse.

Of course, we can’t allow President Trump off the hook. He decided to dismantle the NSC pandemic response team which Susan Rice had set up, combining it into a single directorate with arms control, weapons of mass destruction terrorism, and global health. The president is entitled to manage the National Security Council bureaucracy as he wants, as it’s part of the EOP. It was claimed that there was too much bloat leftover from Obama’s NSC, and so this should have streamlined communication, which may have sounded reasonable. Nonetheless, whatever changes Trump implemented were a colossal failure. Managing the CDC, FDA, FEMA, and National Institute of Allergy and Infectious Diseases to coordinate in a pandemic is crucial, and it did not occur until weeks after it should have. We should ask if the NSC, which varies from president to president, should even have this coordination power or if Congress should implement a more permanent response hierarchy.

Moreover, the president himself downplayed concerns about the virus the entire month of February, and even promised that 15 cases would soon be down to “close to zero”. This is an idiotic statement but is not unexpected. Trump has long ignored expert advice, lied to the public, and spoken erratically about how his administration will execute at his direction. The difference is that while the first 3 years of his presidency had few crises that weren’t self created, a pandemic actually requires decision making and policy implementation. Trump claimed that he would surround himself with good managers, and yet his White House has been plagued with scandals, including one that resulted in his impeachment. Ultimately, Trump is responsible for the leaders of the agencies that failed so spectacularly to manage this crisis. When asked about the testing fiasco, Trump said “I don’t take responsibility at all”.

Finally, the Obama administration had even created a “pandemic playbook”, a 69 page document you can review at this link which details decision making rubrics, key agencies that need to be consulted, and which questions need to be asked. It’s apparently unclear if senior leaders at government agencies even knew the playbook existed.

“State Capacity”

The typical consequentialist libertarian critique runs something like “the government doing things is bad because the state has poor incentives”. However, with this pandemic, we have instances where the state did too much, like where the FDA got in the way of scaling life-saving testing, but also where the state did too little, like coordinating responses of expert state agencies, or providing guidelines to local governments. A consequentialist libertarian critique would have a hard time dealing with these two failure types, and as somewhat of a consequentialist libertarian, I didn’t understand Cowen’s “state capacity libertarianism” until I was faced with these current government failures.

In a pandemic, every individual and company wants to continue working and consuming as much as possible, but each action also endangers non-parties to the transaction (e.g. other grocery shoppers or attendees at a basketball game). Here was a role for government to play, and yet it completely failed in that role. While a more typical libertarianism might treat the state as a necessary evil from which helpful action is capricious and rare, state capacity libertarianism suggests there is an expected duty for the state to play where there are negative externalities like the spread of a pandemic. The government’s failure here should thus be taken more seriously under state capacity libertarianism.

This also opens up a discussion about “state management” that previously I have often avoided. For example, it’s quite onerous to file taxes in the United States. Some libertarians argue that this is actually a good thing, and in fact making these state interactions difficult helps to convey to citizens how useless the state is, and they will thus be supportive in making the state smaller. Other libertarians might argue that added difficulty in the form of coerced taxation is adding additional rights violations on top of an already immoral action. A state capacity libertarian approach wouldn’t have this confusion; in this view, taxes ought to be collected in the most economically efficient way possible (land value tax or Pigouvian externality taxes would be a good start), and that method should be carried out competently and easily by the state. The IRS would just calculate your Pigouvian externality tax and send you a bill without requiring each citizen to compile their own information, spend hundreds on tax advice and then be punished when the amount is incorrect.

Relatedly, the management ability of electoral candidates sometimes comes up, and I think I may have even argued in the past that if congress or the president are less capable, then they would pass fewer laws. If most laws are net negative, then perhaps this incompetence would be good. However, elected officials do not just pass laws, they impact how government services are carried out. When a high risk, externality-laden scenario arises, like a pandemic, the management aspect of governance rises in importance.

Many presidents, especially Trump, have campaigned on the idea that they are good managers. Until now, I mostly shrugged at these claims, since governing is most often about policy and posturing than actual management. Yet the rare scenarios where management is required have outsized impacts; having fire insurance doesn’t matter until you actually have a fire, but then it matters a great deal. If we are concerned about catastrophic risk, then government management is a vital skill to be evaluated. State capacity libertarianism strongly favors evaluating this skill.

Libertarian Critiques Remain

Finally, we should quickly cover that a strong libertarian critique of government intervention into markets remains. Private property based markets with competition have much better incentives for rapid development and innovation. The FDA’s intervention to thwart testing probably killed thousands, especially compared to an alternate world where additional testing kits were explicitly allowed beginning in mid-February. Libertarians have been criticizing FDA policy on slow drug approval for decades though, and COVID patients will not be the FDA’s only victims this year.

And that’s not the only place where the state is dangerous: the War on Drugs, for example, has resulted in the deaths of hundreds of thousands as prohibition empowers organized crime, makes drugs more dangerous, incarcerates people for non-violent crimes, creates more violent interactions between citizens and police forces, and on and on. U.S. government policy has also resulted in long-lasting deadly wars in the Middle East with little to no concrete benefits for the U.S. globally, and I’m sure your local libertarian could continue on with a very long list.

This pandemic places into stark realization that government both needs to get out of the way in some areas to save lives, and also competently carry out the tasks only it can do. A Cowenian marriage of state capacity and libertarianism is the way forward.

Observations on Impeachment

Impeachment is a highly political process. I want to walk through the impeachment process and trial and try to articulate my own thinking.

The Transcript

Let’s start with the phone call on July 26th. Trump released a transcript of himself asking the head of state of another country to investigate a conspiracy theory that Ukraine has a copy of a server of Hilary Clinton’s emails. The intelligence community believes this is Russian propaganda and National Security Council official Fiona Hill testified to this under oath. Trump also brought up a political rival during an official call with the head of state of another country, and asked President Zelensky to investigate them.

There’s a lot to discuss here. Apart from the intelligence community’s views of the Ukrainian server conspiracy theory, I think it reflects poorly on Trump’s…mental state? priorities? that he is still trying to investigate theories around Hillary Clinton’s emails four years after his own election, which he won! It doesn’t seem like he is prioritizing implementation of American policy, but rather stuck in the irrelevant past.

Also of note, Trump brings up Viktor Shokin, who, as far as I can tell was widely believed to be corrupt, yet Trump seems to think his dismissal was unfair. There’s also the discussion of Hunter Biden, Joe Biden’s son. Hunter seems to be a screw up and a bad human being. Yet, as Senator Romney pointed out, there was no evidence that the Bidens’ actions were criminal although certainly morally questionable. In particular, Joe Biden had a conflict of interest, but I don’t think anyone can argue Viktor Shokin should have remained in office…except Trump.

Moreover, it doesn’t seem to be a matter of national security or indeed foreign policy that Joe Biden be investigated by the Ukrainian government even if he were directly implicated in a crime. Instead, I would expect the FBI would just investigate him for wrongdoing. That seems much more effective if you wanted to get to the bottom of it. Moreover, looking at the testimony presented by Trump’s defense, it doesn’t seem like a crime was committed.

Yes, when making a public announcement about Ukrainian prosecutors, the Vice President should acknowledge all conflicts of interest. He didn’t, and that seems bad. While it isn’t a violation of a particular law (the respective law would have to be pretty specific), it reflects pretty poorly. In fact, there are many parallels to Trump’s actions. I’d go so far as to say that it’s inconsistent for one to believe that Joe Biden’s actions are wrong while saying Trump’s are not. They are almost identical.

Both are accused of undertaking policy decisions that they claim are in the national interest while appearing to have personal conflicts. The differences are that (1) Trump is currently in office, while Biden is not, and (2) the House of Representatives seems like the correct place to investigate whether Trump had a conflict of interest, while the Ukrainian government absolutely should not be in charge of investigating Biden. And it definitely shouldn’t be incentivized with U.S. taxpayer funded military aid; it should be done by a law enforcement agency. Finally, I would be remiss not to mention that Trump has an even closer parallel with the Hunter Biden situation: his own children, in particular his son-in-law who was given a prominent position in the west wing. If Hunter Biden’s actions in Ukraine are worth using the office of presidency in such a way, what are we to make of Trump’s own nepotism?

Returning to the call, the Trump defense team pointed out that no quid pro quo was mentioned in this phone call. This is true, and also seems irrelevant. The use of the office of the president to encourage foreign governments to investigate political rivals and conspiracy theories is most certainly an abuse of office, regardless of whether a quid pro quo occurred. Whether this is grounds for removal from office we can address later.

Witnesses and Evidence

More circumstantial evidence for this being a political abuse of office include testimony from several witnesses. Most interesting is from Gordon Sondland who stated that a quid pro quo did exist offering the President of Ukraine a White House visit in exchange for an announcement of investigations into the Bidens. He also stated that John Bolton, Mike Pompeo, and Mike Pence were all aware.

Although not a witness, Mick Mulvaney gave a press conference confirming that military aid authorized by Congress was held up in order to get Ukraine to investigate the DNC server.

Also of note is that the White House released the aid to Ukraine on September 11, only days after Congress announces an inquiry into Rudy Giuliani’s involvement in Ukraine and possible interference with US policy. No one has made any attempt to explain what particular evidence of the Ukrainian government fighting corruption came to light on that day which made the White House approve the aid.

Finally, the House asked (not subpoenaed) John Bolton to testify which could have turned the circumstantial evidence into specific testimony against Trump, at least according to the testimony of Gordon Sondland. The Trump administration blocked those witnesses from testifying, citing executive privilege. The Constitution gives the House the sole power of impeachment, not the executive the power to overrule their investigation. The House could definitely have taken this to the courts with an official subpoena, but they decided not to, I suspect because they felt that impeachment was somewhat politically toxic given the President’s support in the Senate. That’s a political calculation. I think the political case against Trump would have been stronger if both the House and the courts sided against the President. They decided not to the and I think the case against Trump for obstruction was weaker because of this procedural choice. Nonetheless, I can’t see any legal argument that would side with the President; if so, the House’s impeachment power is useless. Therefore, the obstruction of Congress charge certainly seems appropriate.

However, I need to take the Democrats down a peg; the stated reason for not taking the subpoena to court was that Trump presented an immediate threat to our democracy because of his election interference attempts. Democratic impeachment managers argued Russia had interfered in the 2016 election and Trump had encouraged this. I find this completely unconvincing. Russia favored a Trump victory over Hilary Clinton, and I’m sure they spent a bunch of money trying to achieve this, but I have never bought into the narrative that Russia can control the outcome of U.S. elections through Facebook ads. It’s ludicrous. Democracy is powerful because it utilizes disparate information from voters; if you think voters have to be shielded from information, even misinformation, then you don’t believe democracy is a force for good in the world. You instead prefer some sort of government where gatekeepers determine what information voters receive and then voters are allowed to vote with that limited information.

Trump’s attempts to use the Ukrainian government to help him win reelection seem to be an abuse of office. But I don’t think he is a threat to a fair election. The biggest threat to a fair election is our entire electoral system.

Senate Trial

Trump maintained that the impeachment inquiry was a “witch hunt” and a “hoax”. Under this view, it’s not surprising that he opposed John Bolton testifying at his Senate trial. Of course, the President’s claims do not address the significant (although circumstantial) evidence gathered against him. Moreover, Bolton supposedly implicates Trump in his new book. The only real explanations for the President’s behavior is either that there was a vast conspiracy, including many witnesses, his own handpicked ambassador to the EU and massive donor Gordon Sondlond, his own handpicked National Security Adviser John Bolton, his own handpicked Chief of Staff Mick Mulvaney, and presumably his own Secretary of State Mike Pompeo (although he wasn’t asked to testify). Or, Trump is guilty of doing some pretty shady stuff that he doesn’t want people to testify to. The President maintained that it is the first scenario. This seems to be quite unlikely given the circumstantial evidence.

But there’s an easy way to check, just have the Senate ask Bolton to testify. Senate Republicans declined to do this. I have a hard time explaining this in a good faith way. Some maintained that the evidence presented by the prosecution was inadequate. Unless Gordon Sondlond and Mick Mulvaney just spontaneously made up the same story despite being integral parts of the Trump’s administration, this seems hard to believe. John Bolton is a well-respected lifelong Republican, serving in the Bush and Trump White Houses. He is a strong opponent of the Obama backed nuclear deal with Iran, a defining Republican foreign policy position in the last election. Any attempt to say Bolton is a left-wing sympathizer is bizarre, and yet virtually no Republicans voted to hear his testimony.

I believe what is actually happening is that there is significant political pressure from Republican voters to end the impeachment process. Even if a Republican Senator believed Trump to be guilty, to stay in office, they must survive a primary challenge from a pro-Trump challenger, which would surely win in a Republican primary against someone who voted against Trump in impeachment. One could easily argue that the Senators are doing the democratic thing by following their voters’ interests, but it would not mean that Trump was innocent of these charges. It would simply shift blame from the Senators to our current democratic system, which is apparently unable to deal with a president who abuses the powers of his office.

Other senators, like Lamar Alexander of Tennessee stated that the prosecution made a compelling case of misbehavior, but that it doesn’t rise to the level of impeachment. We will get to this defense in a minute.

Non Defense Arguments

It’s worth taking a minute to discuss some points I have heard often, but are not actual defenses of President Trump. The first is the critique of hypocrisy, which is pretty common in these partisan times. Hypocrisy is an excellent way to impugn the motives of your political opponents, but it doesn’t address the object-level arguments. The way I’ve heard this phrased is that presidents have been expanding executive authority for decades, and now Democrats are only calling out Trump’s abuses of power because they don’t like his tribal affiliation, which is more than the usual Right-Left divide. In other words, Trump is a jerk to his political enemies and that’s the reason the House impeached him. Reason Editor Nick Gillespie has espoused something like this view, and tied it in with a libertarian point which is that Democrats don’t actually care about executive authority, they just don’t like Trump. This is a double standard other presidents have not been held to.

Closely tied with this critique is that House Democrats have wanted to impeach Trump from the beginning and were just waiting for a chance they could exploit. Note, neither of these related points address whether Trump actually abused his office. I think it’s absolutely true that Democrats have wanted to impeach Trump, but that doesn’t have any bearing on whether or not he did bad enough acts that he should be removed from office. This argument would be equivalent to Al Copone’s lawyers arguing that the prosecution had been wanting to catch him for a long time and so it’s irrelevant that he broke the law. This is not an argument. The only thing that matters is what Trump did, and the evidence isn’t good.

Returning to the abuse of executive power, I fail to see how continued abuse and concentration of executive authority over time means we should continue allowing abuse of authority. This view basically says that if one government official gets away with abuse, then we have no standing to ever challenge their successors abuses again. This makes no sense to me.

I’ve written pretty extensively about the problem with increased executive authority that’s accrued to the president. A thousand wrongs don’t make the next wrong a right. Instead, trying to restore some of the rule of law one piece at a time sounds like a good idea. And we should be using this opportunity to recruit more who didn’t use to care about executive authority into the tent.

Allowing unchecked executive power to own the libs is a dumb strategy.

Finally, there’s a lot of talk about the whistleblower who wrote a letter that helped to start this investigation. According to Republicans, this whistleblower was a partisan who didn’t like Trump. I have to admit, I have no idea how this argument is supposed to work or how it could possibly be relevant, but it seems so common, I have to include it. However the House finds out about presidential abuse, they have the sole authority of impeachment, meaning they can call witnesses and investigate wrongdoing. They’ve done so, and the witnesses have implicated Trump in abusing his office. The whistleblower’s testimony isn’t necessary. Why Rand Paul keeps talking about the whistleblower seems to just be a distraction from the evidence of Trumps wrongdoing.

Defenses

To summarize, apparently Trump sent his personal lawyer to Ukraine with intention to dig up dirt on his political opponent and even met with Ukrainian officials to achieve this end. Trump then used his office and capacity as president to directly bring up his political goals with the Ukrainian president on an official call. Witnesses and Mick Mulvaney say that military aid and a White House visit were conditioned on investigations into a discredited conspiracy theory about the DNC servers (which are not a matter of national security) and also an investigation of his political rival’s son from several years ago (also doesn’t seem to be a matter of massive national security importance).

This seems pretty bad.

I have a low tolerance of abuse of power. I think Obama was horrendous in his abuse of the office of the presidency including when the IRS targeted conservative groups, and when the administration targeted journalists with the Espionage Act. I think these could pretty easily be classified as impeachable. In that light, I don’t see how the evidence against Trump is much different.

We’ve already covered how the disinterest in the Senate on hearing from John Bolton is pretty suspect. But let’s talk about the actual defenses given by the president’s legal team. There are quite a few.

Some seem pretty specific and weak; Ukrainian President Zelensky said he was not pressured to investigate the Bidens or Crowdstrike. Of course, if your entire presidency is based on opposing Russia and you need U.S. help to maintain that stance, and the Senate is entirely controlled by Republicans, why would you risk antagonizing Trump who viciously attacks his own officials if they ever cross him (see Jeff Sessions, Gordon Sondlond was fired). If Trump was removed, it’s certain that Zelensky could quickly make friends with any future Democratic president regardless of his current positions, and so it makes sense that he would offer to support Trump publicly. This public position seems to count for little compared to actual testimony of American witnesses under oath.

More interesting I think was the argument that there could be no impeachment without a statutory crime. This could be promising, but as Ilya Somin at The Volokh Conspiracy points out, holding up Congressionally approved funds is a violation of federal law. Moreover, if you had to pass a law specifying every possible way that the president should be constrained then we would have no limits on the presidency.

Returning now to Senator Lamar Alexander, who argued against subpoenaing John Bolton by saying the President acted inappropriately but his actions didn’t rise to the level of impeachment. Supposing what the President did was not impeachable, why would this preclude getting more information from a witness? Senator Alexander didn’t know the extent of Trump’s actions prior to the House investigation and the testimony of witnesses. He says Trump’s actions were inappropriate, then it would seem judicious to get additional information to make sure no further wrongdoing had occurred. The Senator’s position is completely incoherent. Moreover, his beliefs about the world don’t explain what we see happening: Gordon Sondland was fired from his position as ambassador to the EU. If Alexander is correct and Trump did some things wrong but nothing impeachable, why was Sondland fired?

Trump maintains everything was a vast conspiracy. I’ve noted before, this is bizarre and would mean that everyone Trump happens to hand-select for prominent positions in his administration turned on him with the exact same beliefs about how his administration operates, supported by tons of circumstantial evidence, including Trump’s own phone calls. However, if you wanted to maintain such a narrative, you’d fire everyone who was in on this conspiracy, including Gordon Sondland, who testified to the existence of a quid pro. Of course, you might also act this way if you were actually guilty of abusing your office. Perhaps only John Bolton could have told us the difference. One thing that does not explain this evidence is Senator Alexander’s position: “Trump did things but they weren’t that bad”. If so, why purge the administration? If the actions revealed by Sondland weren’t a big deal (and honestly I kind of thought Sondland thought this) then what did he do wrong from Trump’s perspective? I don’t think Senator Alexander can explain this, and thus I think his position makes little internal sense.

Moreover, I think he’s wrong on the object level as well. The actions undertaken by Trump are serious. He held up Congressionally authorized aid for personal political reasons. Separation of powers is a vital part of our limited government. If the executive can simply kidnap funds authorized by Congress, then we have no limits on government power. We are no longer a limited Republic where the rights of individuals and the minority are defended against an overzealous majority. Instead we are simply electing a despot every four years who will terrorize his political enemies. If what Trump did was an acceptable use of the office of the president, can the president just deny funding for things until people do his political bidding? Could Obama have denied highway funding to red states until they agreed to drop lawsuits against Obamacare? Could he have held up funds until Congress authorized his strikes in Libya? If Trump could do the same for Ukrainian aid, I’m not sure what the difference is, or how any of this could be called limited government or separation of powers.

Finally, there is the argument that this matter should be left to the voters. It is, by definition, a very democratic argument. Clearly, of course, such an argument could not always make intuitive sense; if a president decided that he had the power to cancel elections and declare himself permanent dictator, then the voters can no longer give any input. This is the tact the Democratic impeachment managers took. I’ve already stated my skepticism about it.

However, we do not have to go there; the Constitution doesn’t indicate that impeachment can only be used if the president endangers elections. Instead it states that he can be impeached for treason and other high crimes and misdemeanors. It is clear that Congress can remove a president who misbehaves, not just about election endangerment. And this makes sense; the separation of powers in our Constitution means that both the president and Congress can claim separate democratic mandates, even conflicting ones. The president isn’t elected dictator for four years, but given limited powers and told to work in conjunction with Congress to exercise authority. If the president misuses that power, it’s clear Congress has the authority to remove him, even in an election year. To believe otherwise is to believe that there are no limits on presidential authority.

Conclusion

It seems clear to me that congressional Republicans are backing Trump due to political expediency rather than the facts as they appear. Yes, it’s true, we don’t have much in the way of direct testimony of what Trump knew and when. But we have ample circumstantial evidence to warrant taking a closer look. Republicans failed to do so, declining to hear from literal Republican heroes like John Bolton and Mick Mulvaney. I understand the reality of their politics but I don’t understand why Republican voters aren’t concerned about the massive power abuses going on in the White House and what that could mean when the other party gets into power. I’ve searched hard for another explanation for the current state of affairs, even looking at Trump’s own defense team, and I found them entirely unconvincing.

Grading 2019 Predictions

I make predictions every year to put empirical tests on my model of the world. I tend to do a lot of predictions, in order to get a larger dataset, and at the end of the year, I grade them. These were made last year in March.  I’ve placed levels of confidence for each prediction with the odds I would bet on those outcomes in the vein of Bryan Caplan. I’ve created a chart at the end to show my calibration versus perfect calibration.

  1. Trump Approval Rating end of year <50% (Gallup): 95% ✔️
  2. Trump Approval Rating end of year <45% (Gallup): 90% (was 45% exactly)
  3. Trump Approval Rating end of year < 40% (Gallup): 70%
  4. US will not get involved in any new major war with death toll of > 100 US soldiers: 60% ✔️
  5. No single terrorist attack in the USA will kill > 100 people: 95% ✔️
  6. The UK will not leave the EU this year: 80% ✔️
  7. North Korea will still be controlled by the Kim dynasty: 95% ✔️
  8. North Korea will not conduct a nuclear test this year: 60% ✔️
  9. North Korea will not conduct a missile test this year: 60% They conducted 10, with several launching many missiles
  10. North Korea will not agree to give up nuclear weapons entirely, contingent on US troops staying in the Korean peninsula: 99% ✔️
  11. North Korea will not agree to give up nuclear weapons as a result of any negotiations: 90% ✔️
  12. Yemeni civil war will still be happening: 70% ✔️
  13. S&P 500 2019 >10% growth (from 2506 on Jan 1): 60% ✔️
  14. S&P 500 will be between 2400 and 3100: 80% (80% confidence interval) was 3231
  15. Unemployment rate December 2019 < 6%: 80% ✔️
  16. Unemployment rate December 2019 < 5%: 70% ✔️
  17. WTI Crude Oil price up by 10% (from $45.41): 70% ✔️
  18. Price of Bitcoin in dollars up over the year (Coinbase – 3823 Jan 1): 70% ✔️ was $7163
  19. Price of Bitcoin < $8,000 (does not double): 60% ✔️
  20. Price of Bitcoin > $1900 (does not lose half value): 70% ✔️
  21. Price of Bitcoin < $12,000 (does not triple): 70% ✔️
  22. Drivechain opcodes not soft-forked into Bitcoin: 80% ✔️
  23. No drivechains soft-forked into existence: 99% ✔️
  24. US government does not make Bitcoin ownership or exchange illegal: 95% ✔️
  25. Self-driving cars will not be available this year for general purchase: 95% ✔️
  26. Self-driving cars will not be available this year to purchase / legally operate for < $100k: 99% ✔️
  27. I will not be able to buy trips on self-driving cars from Uber/Lyft/Waymo in a location I am living: 95% ✔️
  28. I will not be able to order groceries on self-driving cars in a location I am living: 90% ✔️
  29. I will not be able to buy a trip on a self-driving car from Uber/Lyft/Waymo without a backup employee in the car anywhere in the US: 80% ✔️ This is tough. You can get self driving cars in Phoenix, but only if you’re part of the Waymo beta and so far they are free, so no “buying”.
  30. The artificial general intelligence alignment problem will not be seen as the most important problem facing humanity: 99% ✔️
  31. Humans will not be in lunar orbit in 2019: 99% ✔️
  32. SpaceX Falcon Heavy rocket will launch again this year: 90% ✔️
  33. SpaceX will bring humans to low earth orbit: 60%
  34. SpaceX will test the “Starship” mock up this year: 70% ✔️ (pretty sure I just meant this giant water tower thing, not a real launch)
  35. Mexican government does not pay for wall: 99% ✔️ (lol)
  36. Border wall construction not complete by end of 2019: 99% ✔️ (some construction occurred, mostly replacing existing wall)
  37. National Debt increases by >$1 trillion (from
    $21,943,897,000,000): 90% ✔️ (was $23.201 trillion on Jan 1 2020)
  38. There will not be a significant decrease in trade barriers between US and China from pre-2017 tariff levels: 90% ✔️
  39. Democratic RCP front runner will not be Bernie Sanders: 80% ✔️ (front runner on Jan 1 was Biden)
  40. Democratic RCP front runner will not be Kamala Harris: 80% ✔️
  41. Democratic RCP front runner will not be Beto O’Rourke: 80% ✔️
  42. Trump not removed from office or resign: 95% ✔️
  43. Trump not impeached: 70% I was not expecting this
  44. No CRISPR edited babies will be born: 80% (it turns out the researcher responsible for the two 2018 CRISPR edited babies had already treated a third unborn child in 2018 when the story broke. Apparently the third baby was born in 2019 if you carefully read Xinhua, so technically this prediction is wrong, although I meant no other researcher would do anything. Remember to properly word your predictions!)
  45. No full year US government budget will be passed (only several months spending): 90% ✔️ (they basically only do continuing resolutions now)
  46. Some tariffs raised: 90% ✔️ (like a bunch)
  47. Trump administration does not file a lawsuit against any news organization for defamation: 90% ✔️
  • I got 4 of 6 predictions correct at 60% confidence
  • I got 7 of 9 predictions correct at 70% confidence
  • I got 7 of 9 predictions correct at 80% confidence
  • I got 8 of 9 predictions correct at 90% confidence
  • I got 7 of 7 predictions correct at 95% confidence
  • I got 6 of 6 predictions correct at 99% confidence

Overall, not bad at all, and we should note that from last year’s grading, my 60% confidence predictions have tended to be overconfident. I only had 6 of those predictions this year, so actually 66% is the closest I could have been to perfect calibration. 70% also ended up being a bit overconfident, but a single additional missed prediction here would have dropped me down to 66% as well. Had I moved one of my correct 70% predictions to 80%, I would have been perfectly calibrated.

Combining this data and data from last year gives:

  • I got 11 of 15 predictions correct (73%) at 60% confidence
  • I got 14 of 20 predictions correct (70%) at 70% confidence
  • I got 14 of 17 predictions correct (82%) at 80% confidence
  • I got 14 of 15 predictions correct (93%) at 90% confidence
  • I got 16 of 17 predictions correct (94%) at 95% confidence
  • I got 10 of 10 predictions correct (100%) at 99% confidence

In 2018, as I noted in the post last year, I should have made some of my 60% predictions at a higher confidence, but other than that, these predictions are remarkably well calibrated if I do say so myself.

I hope to post my 2020 predictions soon.

How can we use our resources to help others the most?

This is the fundamental question of the Effective Altruism movement, and it should be the fundamental question of all charitable giving. I think the first fundamental insight of effective altruism (which really took it from Peter Singer) is that your donation can change someone’s life, and the wrong donation can accomplish nothing. People do not imagine charity in terms of “investments” and “payoffs”, yet GiveWell estimates that you can save a human life for somewhere in the magnitude of $2500.

Many American households donate that much to charity every year, and simply put, if the charities we donate to don’t try to maximize their impact, our donations may not help many people, when they could be saving a life.

This post is a short reminder that we have researched empirical evidence that you can make a difference in the world! The EA movement has already done very impressive work on how we might evaluate charitable giving, why the long term future matters, and what the most important and tractable issues might be.

Apart from the baseline incredible giving opportunities in global poverty (see GiveWell’s top charities), the long term future is an important and underfocused area of research. If humanity lives for a long time, then the vast majority of conscious humans who will exist will exist in the far future. Taking steps to ensure their existence could have massive payoffs, and concrete research in this area to avoid things like existential risk seems very important and underfunded.

I write this blog post not to shame people into donating their entire incomes (see Slate Star Codex on avoiding being eaten by consequentialist charitable impacts), but rather to ask donors to evaluate where you are sending your money within your budget and to see if perhaps the risk of paying such a high opportunity cost is worth it. Alma maters and church groups are the most common form of charity Americans give to, but the impacts from these areas seem much lower than donating to global poverty programs or the long term future.

Finally, part of this blog post is simply to publicly discuss what I donate to and to encourage others to create a charitable budget and allocate it to address problems that are large in the number of people they impact, highly neglected, and highly solvable. I thus donate about a third of my budget to GiveWell as a baseline based on evidence backed research to save lives today. I then donate another third of my budget to long term causes where I think the impact is the highest, but the tractability is perhaps the lowest. Top charities I’ve donated to here include the Machine Intelligence Research Institute for AI alignment research, as well as the Long Term Future Fund from EA Funds.

The last third of my budget is reserved to focusing on policy, which is where I believe the EA movement is currently weakest. I donate money to the Institute for Justice, as they work on fairly neglected problems in a tractable way, winning court cases to improve civil liberties for U.S. citizens. I also like the Center for Election Science as they work to improve the democratic processes in the US. It would be great to be able to move good policies to polities with bad institutions (i.e. many developing nations), but that problem seems highly intractable. It may be that the best we can do is create good institutions here and hope they are copied. I’m open to different ideas, but I am a relatively small donor and so I believe that taking risks with a portion of my donations in ways that differ from the main EA thrust is warranted. This is by far my most uncertain category, and thus usually I will not entirely fulfill my budget for policy charities. I plan on giving anything remaining to GiveWell.

There are many resources from the Effective Altruism community, and I’ll include several links of similar recommendations from around the EA community. If you haven’t heard of EA charities, consider giving some of your charity budget to GiveWell, or other EA organization you find convincing. If you don’t have a charity budget, consider making one for next year. Even small amounts a year can potentially save dozens of cumulative lives!

Podcast Recommendations October 2019

Last year I wrote up a post discussing my recommended podcasts, and I figured it was about time to update my list. Podcasts have grown significantly in the last 10 years to the point where I honestly haven’t listened to terrestrial radio stations for several years. Podcast distribution is decentralized, and the barrier to entry is low. We live in a world where if you have a niche interest, there’s going to be a podcast and several YouTube channels covering it.

But since podcast discussion is decentralized, my most common method of hearing about podcasts is through other people. In that light, I have created this list of recommendations. It is loosely grouped with podcasts I have listened to longer and/or enjoy more at the top, with more recent podcast discoveries or podcasts whose episodes I have found hit or miss towards the bottom.

I’d also like to take a second to recommend a method of podcast listening: have a low barrier to skipping an episode of a podcast that you otherwise enjoy. This was actually a recommendation by 80,0000 Hours podcast host Rob Wiblin. He encourages his listeners to skip podcast episodes if they find it uninteresting because he’d rather they continue to enjoy the pieces of content from the podcast that they do like, rather than feel like they have to slog through parts they don’t. Moreover, there is just so much good content out there, you should never waste your time with something you don’t find interesting. And now the (slightly sorted!) list:

Reason Podcast

First up, the Reason Podcast includes several different types of excellent content. My favorite is the Monday Editor’s Roundtable which usually includes Katherine Mangu-Ward, Matt Welch, Nick Gillespie, and Peter Suderman. It’s well-edited, sharp, witty, and always tackles the latest news of the week from a libertarian perspective. In the last few years I often find myself wondering if the political world has lost its mind, and on Mondays I’m able to get the message that yes, everyone has gone crazy, but you’re still not alone, there are these four libertarian weirdos who are right there with you. Moreover, Nick and Matt’s obscure 70s and 80s pop cultural references and cynicism play well off of Katherine and Peter’s more techno-libertarian science fiction vibe.

However, that’s not the only content here! There are many interviews from presidential candidates to authors and professors. Audio from the monthly SoHo Forum debates are also posted, and I always listen to at least the opening statements (audience Q&As are less interesting to me). Overall, I almost never skip an episode of the podcast and they produce a ton of great content!

80,000 Hours

80,0000 Hours is an effective altruist organization researching how people can do the most good with their careers. The effective altruist movement does great work, and I think anyone seriously interested in making a difference in the world should be aware of it and the approach with which effective altruists analyze the world. But more than that, this podcast is just more awesome than other interview shows. Rob Wiblin, the host, is excellent at interviewing. He presses the guests on issues but is also willing to accepting strange concepts about the world and follow them to their interesting conclusions.

The interviews are also long, sometimes resulting in 3 hour episodes. This is on purpose, as they can cover in depth why people have the beliefs they do, and what specialized knowledge they have accumulated working in niche roles. Sample episodes include Vitalik Buterin (founder of Ethereum) on ways to revamp public goods, blockchains, and effective giving, Paul Christiano (AI alignment researcher at OpenAI) on messaging the future, increasing compute power and how CO2 interacts with the brain, and Philip Tetlock (author/inventor of Superforecasting) on why forecasting matters for everything.

This one is perhaps a bit more intense than some of the more chill “people hanging out” podcasts, but I listen to every episode.

EconTalk

EconTalk is centrally an economics podcast hosted by Russ Roberts. It’s funded by the Library of Economics and Liberty and Roberts leans libertarian, but he is a courteous and thoughtful interviewer. He knows his biases and acknowledges them during discussions. The podcast strays into many related fields, not just economics; Russ is interested in personal philosophy and introspection as well.

As of late, Russ has particular concerns about the economics field and how free market policies fall short of what we might hope for. In particular, he has discussed themes of societal disillusionment and isolation that simple “material” concerns that dominate economic metrics cannot capture. I wouldn’t say I always agree with Russ and certainly not with all of his guests, but I can say I listen to almost every episode because there are so many good insights discussed.

The Fifth Column

I recently heard the term “Dive Podcast”. This is an excellent description of The Fifth Column, a talk show hosted by Kmele Foster, Matt Welch, Michael Moynihan, and Anthony Fisher. All lean various degrees and shades of libertarian, and discuss the news and/or critique the ever continuous stream of takes in print media, television, online, Twitter, etc while in various states of inebriation. This is much less of a cerebral lecture and more of a “rhetorical assault” as Kmele calls it.

I find the show incredibly entertaining, often informative, and very funny. I listen to all episodes as soon as they are posted.

Hello Internet

Hello Internet is another talk show, hosted by YouTubers CGP Grey and Brady Haran. It isn’t really related to any topics we cover here on the blog, but it is nonetheless entertaining and charming. Unlike The Fifth Column, there is no alcohol involved in the making of this podcast, but it does have an amusing self-grown culture and language.

For example, there is an official flag of the podcast after a referendum of users was held, but one of the losing flags is occasionally taken up by rebellious listeners. There are also unofficial official birds of Hello Internet (the Reunion Swamphen with limited edition t-shirts). Topics covered include YouTube, technology, but also the various interests of Brady and Grey, such as mountain climbing or Apple products. There’s no simple way to convey this podcast, but I do recommend it, and I do listen to every episode.

Rationally Speaking

Rationally Speaking is an interview show hosted by Julia Galef, founder of the Center for Applied Rationality and who I’ve heard described as one of the major pillars of the rationality community. Like Russ Roberts of EconTalk, Galef is an excellent, fair, and thoughtful interviewer. However, the subjects of these interviews are much broader than EconTalk’s admittedly broad discussion of economics. There is a general focus on the philosophy of why we believe what we believe. I do tend to skip more episodes of Rationally Speaking than I do of previously mentioned interview podcasts, but I estimate I still listen to 90% of all episodes, and I would absolutely recommend this very accessible podcast to everyone.

The Economist Editor’s Picks

This one is pretty straightforward. In a world where we tend to get news continuously from the internet or our smartphones, this podcast is a short, ~20 minute weekly selection of important topics from a global perspective that you might not know much about, and that may have gotten swept away in the torrent of your daily information deluge. The Economist is certainly opinionated, but I think does a good job of promoting moderate, liberal ideas that would improve the world. This podcast is an excellent way to expose yourself to some of those simple important concepts in a global context.

Anatomy of Next

From Founder’s Fund, this is a bit of an outlier podcast on here. It’s much more of a series of scripted journalistic pieces or lectures rather than recorded unscripted discussions between people. However, it is quite ambitious in its ideas. The latest season, entitled “New World” which finished up in early 2019, is about how to build a human civilization on Mars. Anatomy of Next explores everything, most of which does not exist yet, but perhaps could. There is terraforming, genetic engineering, sci-fi launch concepts, etc.

I wouldn’t say this podcast is for everyone, but if you feel like you are missing out on human optimism, where people talk about settling Mars with technology that doesn’t exist and yet remain incredibly compelling, this is a podcast you should definitely check out. Also, thanks to Nick Gillespie and Reason for interviewing Mike Solana and letting me know about this podcast in the first place!

Building Tomorrow

Building Tomorrow is a podcast about technology and innovation, and how that is leading to and interacting with individual liberty. It’s hosted at Libertarianism.org which is a project of the Cato Institute. I only recently discovered this podcast and thus it is lower down on my list only because I haven’t had a chance to listen to as many episodes as I would like. Nonetheless, every episode I have listened to is really great! Of course, this program is the perfect niche for me to enjoy, but I would definitely recommend it to anyone who enjoys this blog.

Conversations with Tyler

Tyler Cowen co-hosts one of the most popular econ blogs in the world, Marginal Revolution, and, of course, he is quite an accomplished economist and author. I have recently discovered his podcast, and it’s pretty wonderful. I admit, I don’t listen to every episode, as it turns out Cowen’s and my interests diverge somewhat, which is quite alright. On the episodes that I do find interesting, Cowen is an excellent, although unorthodox interviewer. I rarely go into an episode knowing much about the interviewee or even thinking that I’d really enjoy the topic, but I am always impressed.

There are some additional podcasts I listen to sporadically, but either don’t fit the context of this blog, or I haven’t listened to enough episodes to recommend them here. Nonetheless, it’s worth mentioning that I have listened to a handful of episodes from the Neoliberal Podcast, and I hypothesize that if I wrote this list again in 3 months, it would likely be here.

If you have any podcast recommendations, please tweet at me or leave a comment! I’m always interested in more podcasts.

A Twitter Digression on Trade and China

This week I saw an absolutely horrendous take on Twitter by Chris Arnade and felt compelled to discuss it. This is partially because the positions are incorrect, but also because his discussion itself was in bad faith and actively worsens our dialogue. Are there bad takes on Twitter every minute? Yes, but hey, I saw this one.

Here is the thread:

First, let’s start with the object-level fact that hundreds of millions of people were lifted out of poverty in China. The intentions of the advocates of free trade isn’t relevant to this fact, what’s relevant is whether a particular policy improved the well being and life expectancy of millions of people. This is equivalent to claiming Jonas Salk was only in it for the name recognition, and therefore, we shouldn’t have used the polio vaccine.

Next, Arnade is simply wrong about the intentions of his political opponents, claiming they only support free trade because of greed. He then obfuscates the target of his accusations by using the ever popular term “elites”. Free traders have been talking about the moral benefits of trade forever. Friedman wrote Capitalism and Freedom in 1962, 10 years before Nixon’s visit to China, and The Economist was founded to repeal the Corn Laws in 1843. Saying Milton Friedman was defending free trade just to make money off of China’s market liberalizations decades later is just lazy argumentation.

So Arnade is wrong about their intentions, but are these even the “elites”? As far as I can tell Chris doesn’t have a good definition of elites. I noted in the past:

He claimed on his interview on EconTalk, that while elites are abandoning faith, it remains an important aspect of life to more everyday people. This divide is not borne out by the data. Income doesn’t predict church attendance, and according to Gallup, the difference in church attendance between college educated and non-college educated is within the margin of error. If you want things that predict church membership, you should use age (young people are less religious) or political ideology (those identifying as “conservative” are far more likely to go to church than those calling themselves “liberal”).

If Arnade doesn’t have a good definition of elites, then it seems pretty duplicitous to then claim “elites” have any particular position since we can’t identify who he is talking about. Even if such a group existed, surely there would be many different positions and ideas within this group. Not for Chris though, everyone in this group is one in the same. And it seems especially malicious to then claim that not only does this group with no definition exist, but they have specific stated incorrect claims! In fact, Arnade has identified these claims from an imaginary group as fake and then reveals the “true” beliefs which are, of course, simply greed. This is not just a strawman, ladies and gentleman, but indeed strawwomen and strawchildren too.

It’s plausible that someone could have disagreements with free traders, but just ignoring their arguments and claiming they’re only after money is a terrible way to learn and improve our understandings of the world. We should be engaging with each other’s ideas sincerely, not attributing hidden values to people we disagree with. I guess I find this especially upsetting because EconTalk, Russ Roberts’ podcast, does such a good job of emphasizing those values of charity and understanding, and Arnade was recently a guest on that podcast. To see someone who was treated very charitably turn around and be so underhanded on Twitter is quite disappointing.

Let’s return to China. Arnade discusses a “deal” where the U.S. allowed human rights abuses hoping that democracy would follow. I know of no one who has ever said that. In fact, the opposite occurred: the Cultural Revolution killed 500,000 people, ending in 1976 and following the Great Leap Forward which killed some 18 million people in the lowest estimates. Since trade liberalizations began, nothing on that scale has occurred again.

The interest in China from Nixon was as a tool in the Cold War against the Soviet Union. Democracy was really not the goal. Moreover, the U.S. didn’t dictate to China to liberalize its economy, the market liberalizations were largely from within Deng Xiaoping’s government. In fact, I’m not really sure the U.S. would even be able to extract human rights improvements through protectionist policies. What is the model we would base it on? Cuba? Iran? North Korea? Venezuela? All have become wonderful bastions of human rights following American sanctions.

Also, I should bring up the simple libertarian critique that even if you have the perfect policy, the perfect knowledge of exactly how a foreign government will react to sanctions and trade agreement details, government is not an impartial executor of policy. Democratic forces and interest groups will always mutate policy as it passes through government, and it will not be implemented in the idealized fashion you might like. For example, what does the phrase “U.S. foreign policy human rights record” conjure up?

It’s true that China has not become a democracy, while many foreign policy types certainly believed it could happen, particularly in the 90s. Chris seems to think he predicted this outcome (not cited). Suppose you knew this, would you change policy? In a choice between a poorer China without democracy and a richer China without democracy, it seems we should choose the richer one because, you know, we want good things for all humans, not just people who live in the same country as us. 

Finally, let’s get to the economics here. There is no “us” who “exported our factories”. Individual firms make decisions in a complex economy. And those decisions have been that as a percentage of the total, manufacturing jobs peaked in the 40s prior to the establishment of the People’s Republic of China:

Arnade seems to understand that as he cites the recent scandal in the NBA where the Houston Rockets GM tweeted in support of the Hong Kong protesters, and quickly deleted it under a firestorm. Yet, Arnade switches between U.S. trade policy, governed by Congress (allegedly), and private firms making profit maximizing decisions without bothering to differentiate them.

It’s unclear exactly how Arnade wanted U.S. policy to intervene to stop private firms from making their own decisions. He cites an upcoming book, but provides no details. He also cites anti-labor policy, and I have heard similar discussions from economist Noah Smith, and many others associated with the new neoliberal movement. My problem is that given all the weird deceitfulness and strawmanning, I have no reason to trust Chris when we finally get to the policy discussion. I agree with him that the NBA or Blizzard caving to the Chinese government is a bad thing, but saying more robust industrial policy would have changed that is a non-sequitur.

Moreover, there are still tons of benefits from trade with China. The smartphone revolution changed the way we interacted with the world, and mostly in good ways I believe. This happened in part because of Chinese manufacturing allowing anyone to buy a highly complex piece of technology for cheap. On the other hand, technologies we have a dimmer view on today, like social media, are entirely U.S. grown.

There are problems in the world today, but we need to improve our level of dialogue if we want to solve them. Refusing to engage with well known arguments that critique your position and instead going on uncharitable Twitter threads is something we should avoid.

Artificial Intelligence and Existential Risk

The purpose of this post is to discuss existential risk, and why artificial intelligence is a relatively important aspect of existential risk to consider. There are other essays about the dangers of artificial intelligence that I will link to throughout and at the end of this post. This essay is a different approach that perhaps will appeal to someone who has not seriously considered artificial general intelligence as an issue requiring civilization’s attention.

This is a second edition of this post, updated in 2019, streamlining my essay from last year.

Impact as a Function of Resources

This blog is often concerned with political discussions, and political fights are divisive. They also tend to optimize for controversial topics and to overshadow more impactful policy debates. For example, abortion debates are pretty common, highly discussed political issues, but there have been almost no major policy changes since the Supreme Court’s decision 40 years ago.  The number of abortions in the US has declined since the 1980s, but it seems uncorrelated with any political movements or electoral victories. If there aren’t meaningful differences from different political outcomes, and if political effort, labor, and capital is limited, these debates seem to distract from other areas that could impact more people.

One way we might tackle the most impactful decisions we face is number of lives at stake. This immediately points to issues like cancer, suicide, and car accidents, with trade offs like average age (cancers tend to be later than car accidents) and tractability (cancer breakthroughs are rare and expensive, autonomous cars may be closer and relatively underinvested in).

There is also a question of timescale. Over short time periods, the concern about catastrophic events isn’t very high, but over time, the chances may arise to a level we should worry about, especially given the very high costs. For example, nuclear war has thankfully been rare, but even a single event could be quite deadly compared to most modern political concerns. Research into how and why nuclear wars have been rare and how to keep them that way might be an excellent use of resources to avoid a catastrophic event.

Existential Risk

So what about the extreme long term? What about not just catastrophic events, but existential risk, i.e. the death of all people on Earth? This blog’s philosophy takes consequentialism as a founding principle, and if you’re interested in the preceding questions of what policies are the most helpful, and where we should focus our efforts, you’ve already accepted that we should be concerned about the effects of our actions. The worst possible event, from a utilitarian perspective, would be the extinction of the human race, as it would not just kill all the people alive today (making it worse than a catastrophe that only kills half of all people), but also ends the potential descendants of all of humanity, possibly trillions of beings. If we have any concern for the the outcomes of our civilization, we must investigate sources of existential risk.

Restating to make things more intuitive: assume it’s the year 2300, and humans no longer exist in the universe. What is the most likely cause of our destruction, and then how likely is that cause? I’m selecting the year 2300 because it seems highly likely that humanity’s capabilities will be radically different at that point, but it is also far enough into the future that current political discussions or planning will likely not be thinking about many of these threats.

Wikipedia actually has a very good article on Global Catastrophic Risk, which is a broad category encompassing things that could seriously harm humanity on a global scale. Existential risks are a strict subset of those events, which could end humanity’s existence permanently. Wikipedia splits them up into natural and anthropogenic. First, let’s review the non-anthropogenic risks (natural climate change, megatsunamis, asteroid impacts, cosmic events, volcanism, extraterrestrial invasion, global pandemic) and see whether they qualify as existential.

Natural climate change and megatsunamis do not appear to be existential in nature. A megatsunami would be terrible for everyone living around the affected ocean, but humans on the other side of the earth would appear to be fine. Humans can also live in a variety of climates, so natural climate change would likely be slow enough for some humans to adapt, even if such an event causes increased geopolitical tensions.

Previous asteroid impacts have had very devastating impacts on Earth, notably the Cretaceous-Paleocene extinction event some 66 million years ago. This is a clear existential risk, but you need a pretty large asteroid to hit Earth, which is unusual. Larger asteroids can also be more easily identified from further away, giving humanity more time to do something (push it off path, blow it up, etc). Extinction events every few dozen million years are unlikely to occur in the next 300 years, and when adding the ability of humans to alter some asteroid trajectories, it seems highly unlikely that an asteroid impact will destroy humanity in the next 300 years.

Other cosmic events are also low probability. Gamma-ray bursts are pretty devastating, but they’d have to be close-by (with a few hundred light-years at least) as well as aimed directly at Earth. Neither of these is likely within the next million years.

Volcanism is also something that has the potential to be pretty bad, perhaps existential level (see Toba Catastrophe Theory), but it is also pretty rare.

An alien invasion could easily destroy all of humanity. Any species with the capability to travel across interstellar space with military ambitions would mean they are extremely technologically superior. However, we don’t see any evidence of a galactic alien civilization (see Fermi Paradox 1 & 2 and The Great Filter). Additionally, solving this problem seems somewhat intractable; on a cosmic timescale, an alien civilization that arose before our own would likely have preceded us by millennia, meaning the technology gap between us and them would be hopelessly and permanently large. We’ll leave the probability on this one with less certainty, but it’s unclear what we can do today to mitigate this risk.

A global pandemic seems pretty bad, certainly much more likely than anything else we’ve covered in the short term. This is also exacerbated by human actions creating a more interconnected globe. However, it is counterbalanced by the fact that no previous pandemic has ever been 100% lethal, and that modern medicine is much better than it was during the Black Plague. This is a big risk, but it may not be existential. Definitely on our shortlist of things-to-worry-about though.

Let’s talk about anthropogenic risks next: nuclear war, conventional war, anthropogenic climate change, agricultural crises, mineral exhaustion, artificial intelligence, nanotechnology, biotechnology.

A common worry is nuclear war. A massive nuclear exchange seems somewhat unlikely today, even if a regional disagreement in the Korean peninsula goes poorly in the worst possible way. It’s not common knowledge, but the “nuclear winter” scenario is still somewhat controversial, and I remain unconvinced that it poses a serious existential threat, although clearly a nuclear exchange would kill millions. Conventional war is also out as it seems strictly less dangerous than a nuclear war.

For similar reasons to nuclear winter, I’m not quite worried about global warming on purely existential terms. Global warming may be very expensive, it may cause widespread weather, climate, and ecological problems, but I don’t believe humanity will be entirely wiped out. I am open to corrections on this.

Agricultural crises and mineral exhaustion seem pretty catastrophic-but-not-existential as well. These would result in economic crises, but by definition, economic crises need humans to exist; if there are fewer humans, it seems that an agricultural crisis would no longer be an issue.

The remaining issues are largely technological in nature: artificial intelligence, biotechnology, nanotechnology, or technical experiments going wrong (like if the first nuclear test set the atmosphere on fire). These all seem fairly concerning.

Technological Existential Risk

Concern arises because technological progress means the likelihood that we will have these technologies grows over time, and, once they exist, we would expect their cost to decrease. Additionally, unlike other topics listed here, these could wipe out humanity permanently. For example, a bioengineered virus could be far more deadly than what would naturally occur, possibly resulting in a zero survival rate. The cost of DNA technology has steadily dropped, and so over time, we might expect the number of organizations or people who have the knowledge and funding to engineer deadly pathogens to increase. The more people who have this ability, the more likely that someone makes a mistake and releases a deadly virus that kills everyone. An additional issue is that it is quite likely that military research teams are right now researching bioweapons like an engineered pathogen. Incentives leading to the research of dangerous weapons like this are unlikely to change, even as DNA engineering improves, meaning the risk of this threat should grow over time.

Nanotechnology also has the potential to end all life on the planet, especially under a so-called “grey goo” scenario, where nanobots transform all the matter on Earth. This has a lot of similarities to a engineered pathogen, except the odds of any human developing some immunity no longer matter, and additionally all non-human life, indeed, all matter on Earth is also forfeit, not just the humans. Like biotechnology threats, we don’t have this technology yet, but it is an active area of research. We would also expect this risk to grow over time.

Artificial General Intelligence

Finally, artificial general intelligence contains some similar issues to the others: as technology advances, we have a higher chance of creating it; the more people who can create it, the more dangerous it is; once it is created, it could be deadly.

This post isn’t a thesis on why AI is or isn’t going to kill all humans. We made an assumption that we were looking exclusively at existential risk in the near future of humanity. Given that assumption, our question is why will AI be more likely to end humanity than anything else? Nonetheless, there are lingering questions as to whether AI is an actual “real” threat to humanity, or just an unrealistic sci-fi trope. I will outline three basic objections to AI being dangerous with three basic counterarguments.

The first objection is that AI itself will not be dangerous because it will be too stupid. Related points are that AI is too hard to create, or we can just unplug it if it has differing values from us. Counterarguments are that experts disagree on exactly when we can create human-level AI, but most agree that it’s plausible in the next hundred or couple hundred years (AI Timelines). It’s also true that we’ve seen improvements in AI ability to solve more general and more complex problems over time; AlphaZero learned how to play both Go and Chess better than any human without changes in its base code, YouTube uses algorithms to determine what content to recommend and what content to remove ads from, scanning through thousands of hours of video content every minute, GPT-2 was powerful enough to start to translate languages as a side effect of machine learning applied to word prediction. We should expect this trend to continue, just like with other technologies.

However, the difference between other technological global risks and AI is that the machine learning optimization algorithms could eventually be applied to machine learning itself. This is the concept of an “intelligence explosion”, where an AI uses its intelligence to design and create successively better versions of itself. Thus, it’s not just that an organization might make a dangerous technological breakthrough, like an engineered virus, but that once the breakthrough occurs, the AI would rapidly become uncontrollable and vastly more intelligent than us. The intelligence analogy being that a mouse isn’t just less smart than a human, it literally doesn’t comprehend that its environment can be so manipulated by humans that entire species depend on the actions of humans (i.e. conservation, rules about overhunting) for their own survival.

Another objection is that if an AI is actually as intelligent as we fear it could be, it wouldn’t make “stupid” mistakes like destroying all of humanity or consuming the planet’s resources, because that wouldn’t count as “intelligent”. The counterpoint is the Orthogonality Thesis. This simply states that an AI can have any goal. Intelligence and goals are orthogonal and independent. Moreover, an AI’s goal does not have to explicitly target humans as bad (e.g. “kill all the humans”) to cause us harm. For example, a goal to calculate all the digits of pi or solve the Riemann Hypothesis might require as much computing power as possible. As part of achieving this goal, a superintelligence would determine that it must manufacture computing equipment and maximize energy to its computation equipment. Humans use energy and are made of matter, so as a way to achieve its goal, it would likely exterminate humanity, and convert all matter it could into computation equipment. Due to its superintelligence, it would accomplish this.

A final objection is that despite experts believing human level AI will happen in the next 100 years, if not sooner, there is nothing to be done about it today or that it is a waste of time to work on this problem now. This is also known as the “worrying about overpopulation on Mars” objection, comparing the worry about AI to something that is several scientific advancements away.  Scott Alexander has an entire blog post on this subject, which I recommend checking out. The basic summary is that AI advancement and AI alignment research are somewhat independent. And we really need to learn how to properly align AI values before we get human level AI.

We have a lot of theoretical philosophy that we need to figure out how to impart to a computer. Things like how humans actually make decisions, or how to value different moral tradeoffs. This could be extraordinarily complicated, as an extremely smart optimization algorithm could misinterpret almost everything we say if it did not already share our values for human life, health, and general brain state. Computer scientists set out to teach computers how to understand natural human language some 60 years ago, and we still haven’t quite nailed it. If imparting philosophical truths is similarly difficult, there is plenty of work to be done today.

Artificial intelligence could advance rapidly from human level to greater than human very quickly; the best human Go player lost to an AI (AlphaGo) in 2016, and a year later, AlphaGo lost to a new version, AlphaGo Zero, 100 games to none. It would thus not be surprising if a general intelligence achieved superhuman status a year after achieving human-comparable status, or sooner. There’s no fire alarm for artificial general intelligence. We need to be working on these problems as soon as possible.

I’d argue then, that of all scenarios listed here, a misaligned AI is the most likely to actually destroy all of humanity as a result of the Orthogonality Thesis. I also think that unlike many of the other scenarios listed here, human level AI will exist sometime soon, compared to the timescale of asteroids and vulcanism (see AI Timelines, estimates are highly variable, anywhere from 10 to 200 years). Compared to other technological sources of risk, AI is unique because of the introduction of another agent with differing goals. There is also a wealth of work to be done surrounding AI value alignment. Correctly aligning future AI with goals compatible with human values is thus one of the most important challenges facing our civilization within the next hundred years or so, and probably the most important existential threat we face.

The good news is that there are some places doing this work, notably the Machine Intelligence Research Institute, OpenAI, and the Future of Humanity Institute. The bad news is that despite the importance of this issue, there is very little in the way of conversations, money, or advocacy. AI Safety research is hard to calculate in total, as some research is likely done by private software companies, but is optimistically on the order of tens of millions of dollars a year. By comparison, the U.S. Transportation Security Administration, which failed to find 95% of test weapons in a recent audit, costs $7.5 billion a year.

Further Reading

I have focused this essay on trying to convey the mindset of thinking about existential risk generally and why AI is specifically worrying in this context. I’ve also tried to keep it short. The following are further resources on the specifics of why Artificial General Intelligence is worth worrying about in a broader context, arranged by length. If you felt my piece did not go in depth enough on whether AI itself is worth being concerned about, I would urge you to read one of the more in depth essays here which focus on that question directly.

The Immigration Tariff in 500 Words

Immigration liberalization is one of the policies this blog has described as highest impact. It could have massive benefits to both immigrants and native born citizens in the United States and other developed countries. Immigration bypasses the need to solve the extremely difficult problem of “building good institutions” which is a mercurial and sparsely solved goal in development. By moving people directly to societies where good institutions already exist, we don’t have to make them. OpenBorders.info also suggests free movement of people could double world GDP, with smaller migration seeing proportionally smaller but still substantial growth.

The United States is uniquely positioned to absorb immigration. It is the largest developed country by both population and GDP by significant margins (developed country referring to either OECD member or country with HDI > 0.8). By nominal GDP the US economy remains the largest in the world, and by PPP it is second only to China. Unlike China, the US is the only large country with a large foreign born population, and indeed the US has the largest foreign born population in the world at over 46 million. The US also has a long history of immigration contributing to its excellent position as an immigration destination.

Given this blogs inclination towards the benefits of markets, self determination, and individual rights, our default position should be in support of more liberalized immigration. Current immigration policy is geared towards family connections despite much of the potential benefits of immigration stemming from economics. The U.S. also takes in less immigrants as a percentage of its population than other developed nations, despite the previously mentioned advantages the U.S. has in absorbing immigration.

Originating from economist Gary Becker, an immigration tariff would allow prospective immigrants to pay a tax or fee to enter the country and work. We have a somewhat similar although highly limited current system with H-1B visas which are sponsored by companies for employees. Expanding this and accounting for age and level of education, Congress could create a tariff schedule for various immigrants based on potential costs and tax revenue from these immigrants. They could also simply sell off additional green cards after the current legal green card approaches were filled in the current year. The Cato paper linked goes into more detail.

The benefits of any such system would be to guarantee that immigrants with the skills and ability to work productively in the United States would be able to do so, with additional monetary compensation provided up front to the U.S. to avoid any potential risk of those immigrants becoming a net cost on society. This would see benefits in terms of additional labor, entrepreneurship, and economic growth.

The issue with this approach is that immigration is a highly divisive political issue. Republicans would be unlikely to embrace this proposal due to their base’s opposition to apparently all immigration. Democrats may be more interested, but may balk at the notion of people “buying” their way to the front of the line.

Further Reading

For more on why immigration is generally a positive policy: