Technology

Should we let ourselves see the future?

This is a cross-post from my new substack, Calibrations. Cryptography professor and solid Twitter follow Matthew Green asks why we should really care what prediction markets say:

It would seem the perfect first post for a blog titled Calibrations. So are prediction markets any good at predicting things, and should we care what they say?

Prediction markets allow participants to buy shares of an event occurring or not, analogous to a bet or wager. For example, the website Polymarket has political event markets. A “share” pays out if the election occurs as you predict and pays nothing if you predict incorrectly. A market is created by people buying and selling shares on the outcomes of the events, and the price of the share represents the market’s current probability estimate for the outcome. For example, you can buy a share of Kamala Harris winning the presidential election which will pay out if she wins, but will be worth nothing if she loses. A price of 60 cents would indicate a 60% chance of her winning.

Data on Accuracy

So are prediction markets any good? Well, we can actually look at the history of prediction markets and see how often a market that predicts any given outcome actually ends up resolving in that way. Maxim Lott at electionbettingodds.com has a track record page for the prediction markets he tracks.

This is the calibration of the markets. If something that is predicted to happen 60% of the time actually happens that often, we say it’s well calibrated (and thus the name of this blog). We can also see the calibration for some other prediction sites here from calibration.city which has some excellent visualizations:

I suspect some of the systematic bias you see in the chart above and below the 50% mark is from the fact that many of these markets default to 50% and require betters to provide liquidity to move the market away from the start. To make an enticing market to bet in, the market makers have to incorrectly price a market so that bettors have an incentive to wager and earn their winnings, and we see this in the data as a the market makers systematically “betting” the wrong way to start the market. In fact, if you go to the calibration.city page and weight the y-axis for resolutions by market volume, this bias is reduced (since larger volume markets would have less weight on the initial price) although it does not disappear.

Alright, prediction markets are well-calibrated, but does that mean they are accurate? Are they actually predicting anything? Not necessarily!

Suppose we know that Republicans win the presidency about half the time and Democrats win the other half. We check the prediction markets and they always say democrats have a 50% chance to win. This is a well calibrated prediction, but it doesn’t actually tell us much information at any given point in time. We want to know about this particular upcoming election. What we can do then is use a Brier score to mathematically measure accuracy, and not just calibration. The Brier score is the means squared error for a set of predictions and their outcomes, and thus we can use it to measure the accuracy of probabilistic predictions. Here are the Brier scores from calibration.city:

I arranged by market volume and did a time weighted average, but you can also try it yourself. Overall it seems market volume didn’t have much effect, perhaps because the existing Brier scores are already pretty low around 0.15, and also perhaps because higher volume markets could reflect exciting changes in the underlying event leading to both more uncertainty and more market trading.

We can actually decompose the Brier score into calibration, resolution, and uncertainty components, although exactly why this works is beyond my current statistical understanding. This stack exchange post was helpful, but unfortunately I don’t have the time to put together a full script to grab the data from Manifold and run the decomposition.

There’s also real limitations to comparing Brier scores. You’re really only supposed to compare them across the same events; if the underlying event is more uncertain, you would expect higher (worse) Brier scores. But Kalshi, Manifold, and Polymarket are all predicting slightly different things. The legal difficulties prediction market platforms face exacerbates the accuracy measurement problem; PredictIt has to limit the number and size of bets, Polymarket and Betfair are banned in the United States, and Manifold only uses play money. Still, you’d expect much of this to contribute to worse Brier scores. The fact that these are pretty solid is encouraging. If anyone wants to run an actual Murphy decomposition on prediction market data on a narrow topic, I would be interested to see the results.

I think we can still draw some conclusions though:

Prediction markets are well calibrated: if they say something will happen with a given percentage chance, that’s actually a good estimate of how likely it will happen.
Prediction markets have pretty good (low) Brier scores, and that puts an lower limit on how practically bad their resolution decomposition component could be even if they were perfectly calibrated.
Prediction markets do all this despite major regulatory barriers.

The Bigger Picture

Taking a step back though, I think Matthew Green’s critique gets to something deeper: sure prediction markets are well calibrated in the face of uncertainty, but they don’t have any magical insight besides pushing the best guesses to the top. In other words, they’re not actually reducing the fundamental uncertainty in the world. Why is everyone so excited about this?

I think the answer is price discovery.

First let’s take the demand side. Perhaps you’ve looked up a prediction market price for an event and found the price to be…about what you expected.

The Fed has been talking about rate cuts this year, they indicated in their last meeting that it’s likely coming and the market took a sharp tumble last week before recovering. It seems pretty obvious to anyone paying attention that the likelihood of a Fed rate cut by the end of the year is high. But not everyone was paying attention. A prediction market’s existence creates a publicly known “price” of the best estimate of whether an event will occur according to the market participants. Being able to know a well calibrated estimate of an uncertain event is a huge positive externality that prediction markets provide the whole world.

This is a big deal! If you don’t have a price to check, you’re left doing all the research yourself and you only have so much time. Worse, you might have to rely on very non-quantified pundit opinions with no track record. A price cuts through everything to give you a single best guess.

Next let’s take the supply side. Think about the amount of time, effort, resources, money, labor, algorithm design, etcetera that trading firms undertake to get an edge in financial trading. There are potentially billions of dollars on the line if you can correctly predict whether an equity or future will rise or fall in price. Prediction markets can harness these incentives to uncover truth not just about company financial data, but about anything we are curious about. They tie reality and truth about the world to financial incentives. They push us to discover information about what is happening and what will happen.

Unfortunately, they are also legally limited. Event based betting is for all practical purposes banned in the United States even though political decision making is very high stakes. Our lives are worse because of this! We should want our political decision-making processes scrutinized and better understood; prediction markets provide financial incentives to achieve those ends. Prediction markets could be providing us with real time information on whether Congress will pass legislation or how an administration will respond to crises.

Of course, the actual application of prediction markets could be much broader than politics. Nominal GDP futures markets could do a better job informing the Fed of forward looking expectations than stock markets or lagging economic data. Or imagine if we had an ongoing prediction market about possible pandemics on the horizon. That would have been very handy to refer to in January or February 2020 and may have alerted us sooner than simply relying on broad media reporting, which didn’t pick up the potential danger of the pandemic until early March.

Perhaps today prediction markets don’t tell us much besides aggregating some polling data, but if we legalized them, maybe they could tell us a lot more!

Uncharted Territory

The finance industry may be driven purely by profit, but the development of new financial instruments has actually brought all sorts of benefits, some of them very widely distributed. Index funds and ETFs allow regular people with savings access to market returns without needing to expose themselves to specific stock risk and at a fraction of the cost of mutual funds. Foreign exchange and commodity derivatives can help companies reduce their risk by locking in prices now; airlines are trying to run an airline business, not predict oil prices in six months.

Prediction markets could do even more if we let them.

A conditional prediction market is a single trading market with two events allowing investors to trade on the relationship between the two outcomes. For example, you could make a prediction market for a political parties’ election prospects conditional on whether they replace their candidate or not. Democrats took months to come around to the idea that their current presidential candidate might do poorly. If they had been able to compare conditional bets on who would win, they might have been able to select a different candidate much earlier!

Conditional markets could have all sorts of interesting applications. We could answer questions like “what will different policies’ impact be on unemployment next year?” or “which scientific research approach is most likely to improve 5 year survival rates of a major cancer?”. But I think the most important point to underline is that we don’t know because the markets aren’t allowed. It took a while to develop ETFs that regular investors could use for their retirement accounts. If we allowed people to experiment and try things, we could probably create some pretty cool things.

Even today, with all the legal challenges prediction markets face, they offer great value! People on Twitter will continue to claim that prediction markets are systemically biased or simply reflecting quirky beliefs of people overrepresented in the betting pool. I think this is too quick of a dismissal, but to some extent, they’re right! And I think it’s because you have to jump through a bunch of hoops to correct the markets and collect your winnings; as long as the CFTC wages a crusade against PMs, it will be challenging to make much money on them and thus they may often have incorrect prices. But we don’t have to live this way! We could look into the future if we allowed ourselves to.

Observations on Parenting

I’ve recently added a new member to my family and become a first time parent! Here are some scattered thoughts and observations I’ve had about the experience. Beware all are basically projections of my existing biases as seen through a new experience.

Policy does not seem to take into account externalities of children. This is really the umbrella observation guiding many of these points. In a market economy, you pay for someone else’s time via wages. If there’s a new industry that forms from the advance of technology (say software engineer or YouTube creator), you can convince people to provide that labor in the new industry by paying them, and other industries will need to adjust to deal with more demand on the labor pool. They must either reduce output, offer better wages, or advance the productivity of their remaining workers. In this way, the labor pool for a given industry can expand or shrink based on the market forces of that industry, with some lag for education requirements.

However, you can’t change the whole pot of labor with market forces except in specific circumstances. Immigration is one of those, but we’ll set that aside for now. Let’s focus on population growth through new babies. Of course, adding a worker to the economy doesn’t just add to the supply side; new workers also buy things! You may think these cancel out, but I’m doubtful. As Adam Smith observed, the division of labor increases productivity and the more people you have, the larger economy you have and the larger the degree of division of labor and extent of the market, leading to super-linear productivity gains.

But just because some jobs’ wages go up doesn’t mean there are more babies being born; there’s no market feedback mechanism for this! You pay the worker once they are old enough to work; you don’t pay their parents to raise them! This results in a missing positive externality. The invisible hand is willing to pay for more workers to work in America’s massively productive economic engine (which we can see through huge immigration pressures). But there’s no market mechanism to pay parents to provide more workers! Thus the expensive costs are born by the family raising kids and the benefits accrue to the economy broadly, a classic positive externality problem.

Matthew Yglesias adds a related argument in One Billion Americans that is specific to the U.S.; there are positive spillover effects from having a large market in the U.S. which is fueled by an integrated largely free market encompassing hundreds of millions of productive workers. The U.S. economy, because of its size, dominates global commerce as companies seek to sell in the American market, which in turn exports U.S. cultural products and values of liberalism.

However, despite these economic and nationalistic benefits, failing fertility rates and deadlocks on immigration mean U.S. policy hasn’t really prioritized population growth. Assuming they should, it’s interesting to see from a parental perspective how U.S. policy impacts the decision to have children. Largely, it seems they work in the wrong direction.

The Baumol Effect is real. The Baumol Effect or Baumol’s Cost Disease is the rise of wages in jobs that have not seen increased labor productivity simply because of the rising salaries in other jobs. For most of the economy, jobs continually become massively more productive through the huge engine of a free market economy investing in new technology and capital. You expect to see rising wages for jobs in that industry since the workers are now more productive even as the overall cost of goods produced can continue to fall. For example, agriculture used to take up the vast majority of labor in the economy but is now a very small part, yet food is more plentiful than ever because mechanization and logistics is so advanced.

But some jobs have seen little labor productivity improvement, notably childcare; we don’t have robots or software that can take care of babies. So that labor needs to be done by parents or hired out via nannies or daycare. But today nannies, daycare workers, and especially parents have many more productive things to do in the economy because of the advanced capabilities and technology of the market. So the opportunity cost of taking care of children is massive. Being a new parent, you feel this viscerally. One week you are at a high paying intense job automating workloads and trying to streamline business processes that handle billions of dollars a year, and the next week you can’t get any sleep because you have to change diapers.

To be clear, I’m not complaining! Parental leave is actually a nice break despite the lack of sleep. But the cost to e.g. my employer and the economy broadly is almost hilarious. And yet, clearly there’s an incentive mismatch since collectively, the economy benefits if there are more children born in the long-run due to children’s eventual contributions to the economy and to innovation.

One could imagine a policy to try and encourage people to have children at a younger age when the opportunity cost to the economy is lower (since less experienced workers are probably less productive). But of course, the most likely to take advantage of this policy are probably people whose opportunity costs are lower anyway (the least productive workers), which means the benefits are skewed towards the members of the economy least needing this. Moreover, targeting any government policy directly at high earners is going to be an impossible political ask anyway.

Build more homes. One of the first things we realized is that we needed more space. However, our current job situation means we aren’t likely to stay in our current city for very long and so we had opted against buying a house when we had the option a few times given closing costs. There’s also some benefit I find from renting from a corporate run apartment complex with an always on-call maintenance staff. I’m sure at some point I’ll be more interested in DIY type repairs but not right now.

However, there’s not many options for people with children or especially multiple children who would like to live in an apartment. And my city is more affordable than most. This is of course just a small reflection of the overall housing overregulation we are imposing on the entire English-speaking world. The U.S. is actually more affordable than Canada, the U.K., or New Zealand in many cases, yet the U.S. could be so much better. The most frustrating part of this is that it would cost local governments literally nothing to allow more housing to be built. In fact, governments would likely see revenues rise as land value would be improved. Private actors are completely willing to provide desperately needed supply all up and down the market, but local governments have been coopted by literally landed interest groups to keep a lid on housing supply to monopolize already built housing.

It’s evil, it’s expensive, it saps growth, extracts rents for people who are not contributing to the economy, it violates property rights, and also as a smaller side-effect, it makes it harder to have a family! These are awful policy impacts and we should choose to do better.

Preeclampsia is pretty common. Preeclampsia is a condition that can occur during pregnancy characterized by high blood pressure and signs of damage to other organ systems, most often the liver and kidneys. It usually begins after 20 weeks of gestation. The exact cause is not known, but it is thought to be related to problems with the placenta, and it occurs between 5-8% of pregnancies.

I had the misfortune of having to see this firsthand and it was quite scary. Ultimately it led significant symptoms requiring an earlier than expected induced delivery and my child having to be in the NICU for over a week. Everyone is doing well now, but this seems like a major policy problem if we take as given that there are massive spillover effects to having more children. No one would get on a plane if there was a 5% chance of a serious complication requiring experts to be consulted just to land the plane. I’m not an expert on NIH funding categories, but it doesn’t appear that preeclampsia is a major priority. If long term economic growth matters and thus spillover effects from population growth matter, should this cause area be higher?

Artificial wombs would be amazing. The follow up to the last point: there are promising technologies that would solve the problem of preeclampsia and many others. Pregnancy sucks. It’s physically taxing, it can make women feel miserable for months, and there are serious risks of complications. If we can replace natural gestation with artificial wombs, we can likely reduce risks to both baby and mother and just make having children less stressful. It also means there’s less drag on the economy because women don’t have to trade off being in physical discomfort at work for several months or taking that time off.

A year ago ago when a tweet went viral talking about artificial wombs, there was a strange pushback I saw from people claiming it wasn’t “natural”. I investigated some of these claims the best I could on Twitter, but the argument really made about as much sense as you’d expect. “Natural” to me means dying from smallpox and dysentery. There’s nothing good or beautiful about dying in the state of nature. Preventing horrific and common complications for women giving birth in the 21st century is a no-brainer.

Emily Oster is really helpful. Emily Oster is an economist and author who publishes on parenting. Her books are written for non-academics and summarize the current findings for childcare and pregnancy. Her work makes it really clear what the data indicates on what’s important or not. She’s controversial because her interpretation of the data on drinking during pregnancy is that a drink or two a week during pregnancy does not have much negative effect. This is against the recommendation of most doctors in the U.S.

I haven’t actually heard any doctors cite specific studies to refute Oster’s position, but honestly, I’m of the position that alcohol is probably just bad for you regardless even if in moderate amounts the negative effects are small. I often go weeks without drinking anyway, so cutting down some more isn’t particularly difficult. So feel free to be more cautious than Oster on this point, but I don’t think it negates the overall benefits of her work.

Each chapter contains a nice summary of the key points, so if a couple weeks later you can’t remember the takeaways, its easy to look them up quickly and trigger the rest of your memory of the chapters. The only real concern is that her books are somewhat short and so I’ve found it important to have other child-rearing and pregnancy books available for more specific questions. However, Oster also has a LLM trained on her books and newsletter where you can ask specific parenting questions and it will find the citations related to your question and summarize an answer.

Book Review: The Precipice

I have titled my annual blog post summarizing where I donate my charitable budget as “How can we use our resources to help others the most?” This is the fundamental question of the Effective Altruism movement which The Precipice‘s author, Toby Ord, helped found. For a while, Toby Ord focused on figuring out how to fight global poverty, doing the most good for the worst off people in the world. Now, he is focusing on the long term future and existential risk.

The Precipice is fantastic. It’s incredibly well written, engaging, and approachable. It covers a lot of ground from why we should care about the world, what risks humanity faces in the future, how we might think about tackling those risks, and what the future might look like if we succeed.

The Precipice eloquently interweaves fairly philosophical arguments with more empirical analysis about the sources of existential risk and tries to statistically bound them. The book discusses a pretty concerning topic of the potential end of humanity, but it does so with an eminently reasonable approach. The complexities of philosophy, science, probability, epidemiology, and more all are brought into the narrative, but made easily digestible for any reader. I honestly wish Toby Ord could teach me about everything, his writing was so clear and engaging.

The main discussion is never overwhelming with technical details, but if you ever find a point interesting, even the footnotes are amazing. At one point I came up with a counterpoint to Ord’s position, wrote that down in my notes, only to find that the next several paragraphs addressed it in its entirety, and there was actually a full appendix going into more detail. Honestly, this will be less of a book review and more of a summary with a couple final thoughts, because I think this book is not only excellent, but its content is perhaps the most important thing you can read right now. You are welcome to read the rest of this blog post, but if you have found this compelling so far, feel free to stop reading and order Toby Ord’s book posthaste.

Existential Risk

The consequences of 90% of humans on Earth dying would be pretty terrible, and given our relatively poor response to recent events, perhaps we should better explore other potential catastrophes and how we can avoid them. But The Precipice goes further. Instead of 90% of humans dying, what happens if 100% of us die out? Certainly that’s strictly worse with 100>90, but in fact these outcomes are far apart in magnitude: if all humans die out today, then all future humans never get to exist.

There’s no reason we know of that would stop our descendants from continuing to live for billions of years, eventually colonizing the stars, and allowing for the existence of trillions of beings. Whatever it is that you enjoy about humanity, whether that’s art, engineering, or the search for truth, that can’t continue if there aren’t any humans. Full stop. As far as we know, we’re the only intelligence in the universe. If we screw up and end humanity before we get off this planet, then we don’t just end it for ourselves but perhaps we end all intelligent life for the remaining trillions of years of the universe.

Even though I was aware of the broad thesis of the book, I was continually impressed with just how many different angles Ord explores. He early on notes that while we might normally think of a catastrophic extinction event, like an asteroid impact, as the thing we are keen on avoiding, in fact there are several scenarios that would be similarly devastating. For example, if humanity were to suffer some calamity that did not kill everyone but left civilization stuck at pre-industrial technology, that would also preclude humanity from living for trillions of years and colonizing the stars. A 1984 style global totalitarian state would also halt humanity’s progress, perhaps permanently.

Ord also discusses the fundamental moral philosophy implications of his thesis. The natural pitch relies on utilitarian arguments as stated above; if humanity fails to reach its potential, this not only harms any humans currently alive but all future generations. Other arguments against extinction include a duty to our past and what we owe to our ancestors, the rights of those future generations who don’t get to decide for themselves, and the simple fact that we would lose everything we currently value.

The book categorizes three types of risk: natural, anthropogenic, and future risks. Natural includes asteroids, supervolcanoes, and stellar explosions. These are pretty diverse topics, and Ord is quite informative. The story about asteroid risk was particularly fascinating to me. In the 90s, the relatively new discovery of the dinosaurs’ demise led Congress to task NASA with identifying all the largest near-Earth asteroids to see if they pose a threat to Earth. They allocated some money, and NASA tracked every near-Earth asteroid over 10 km in length, and determined that none pose a threat in the next century. They then moved on to 1 km asteroids and have now mapped the vast majority of those as well. The total cost of the program was also quite small for the information provided — only $70 million.

This is one of the rare successes in existential risk so far. Unfortunately, as Ord points out several times in the book, current foundational existential risk research at present is no more than $50 million a year. Given the stakes, this is deeply troubling. As context, Ord points out that the global ice cream market is about $60 billion, some 1000x larger.

I’ll skip the other natural risks here, but the book bounds natural risk quite skillfully; humans have been around for about 200,000 years, so it seems natural risk can’t be much higher than 0.05% per century. Even then, we’d expect our technologically advanced civilization to be more robust to these risks than we have been in the past. Many species survived even the largest mass extinctions, and none of them had integrated circuits, written language, or the scientific method.

That doesn’t mean that all risk has declined over time. On the contrary, according to Ord, the vast majority of existential risk is anthropogenic in origin. Nuclear weapons and climate change dominate this next section. It’s remarkable just how callous early tests of nuclear weapons really were. Ord recounts how there were two major calculations undertaken by a committee of Berkeley physicists before the Manhattan project got underway in earnest. One was whether the temperature of a sustained nuclear reaction would ignite the entire atmosphere in a conflagration (the committee believed it would not). The other was whether Lithium-7 would contribute to a thermonuclear explosion (it was believed it would not). It turns out that Lithium-7 can contribute to a thermonuclear explosion as was found out when the Castle Bravo test was about three times larger than expected, irradiating some 15 nearby islands.

It turned out the other calculation was correct, and the first nuclear explosion in 1945 did not ignite the atmosphere. But clearly, given the failure of the other calculation, the level of confidence here was not high enough to warrant the risk of ending all life on Earth.

Luckily, current risk from nuclear weapons and climate change that would wipe out humanity seems quite low (although not zero). Even a nuclear winter scenario or high sea level rise would not make the entire Earth uninhabitable, and it is likely humans could adapt, although the loss of life would still be quite catastrophic.

Instead, the bulk of the risk identified by Toby Ord is in future technologies which grow more capable every year. These include engineered pandemics from our increasingly powerful and cheap control over DNA synthesis, as well as artificial intelligence from our increasingly powerful and integrated computer systems.

The threat of engineered pandemics is particularly prescient as I write this in August 2020 where SARS-CoV-2 is still sweeping the world. Ord notes that even given quite positive assumptions about whether anyone would want to destroy the world with a virus, if the cost is cheap enough, it only takes one crazy death cult to pull the trigger. Even an accidental creation of a superweapon is a serious risk, as production is cheap and there are many examples of accidental leakages of bioweapons from government laboratories in the past. Unfortunately, we are also woefully unprepared on this front. The Biological Weapons Convention had a budget of $1.4 million in 2019, which Ord notes is less than most McDonald’s franchises.

Risks from unaligned artificial intelligence are similarly related to technical advancements. Ord notes that artificial intelligence has had some impressive achievements recently from photo and face identification to translation and language processing to games like Go and Starcraft. As computer hardware gets better and more specific, and as we discover more efficient algorithmic applications of artificial intelligence, we should expect this trend to continue. It therefore seems plausible that sometime in the future, perhaps this century, we will see artificial intelligence exceed human ability in a wide variety of tasks and ability. The Precipice notes that, were this to happen with some sort of general intelligence, humanity would no longer be the most intelligent species on the planet. Unless we have some foresight and strategies in place, having a superior intelligence with it own goals could be considerably dangerous.

Unfortunately, we are already quite poor at getting complex algorithms to achieve complicated goals without causing harm (just take a look at the controversy around social media and misinformation, or social media and copyright algorithms). The use of deep learning neural networks in more high stakes environments means we could be facing opaque algorithmic outcomes from artificial intelligence that we don’t know if we’ve correctly programmed to achieve the goals we actually want. Throw in the fact that human civilizational goals are multifaceted and highly debated, and there is a great deal of mistakes that could occur between what humans “want” and what a superior intelligence attempts to accomplish. While Toby Ord doesn’t think we should shut down AI research, he does suggest we take this source of risk more seriously by devoting resources to addressing it and working on the problem.

So What Do We Do?

I’ve spent a lot of time on enumerating risks because I think they are a concrete way to get someone who is unfamiliar with existential risk to think about these ideas. But Ord isn’t writing a book of alarmism just to freak out his audience. Instead, starting with the high levels of risk and adding the extremely negative consequences, Ord details how we might begin to tackle these problems. Unprecedented risks come with modeling challenges: if an existential risk cannot by definition, have ever occurred, how can we know how likely it is? We have to acknowledge this limitation, use what incomplete knowledge we can have access to (number of near misses is a good start), and start building institutions to focus on solving these hard problems.

International coordination is a major factor here. Many of these problems are collective action problems. Humanity has found ways around collective action issues with international institutions before (nuclear arms treaties), and so we need to replicate those successes. Of course, we can’t establish new or better institutions unless we get broad agreement that these issues are major problems that need to be solved. Obviously, that’s why Ord wrote this book, but it’s also why I feel compelled to blog about it as well. More on that momentarily.

In this section of the book, The Precipice outlines preliminary directions we can work towards to improve our chances of avoiding existential catastrophes. These include obvious things like increasing the funding for the Biological Weapons Convention, but also discussions on how to think about technological progress, since much of our future existential risk rises as technology improves. We also obviously need more research on existential risk generally.

Finally, I want to wrap up discussing Appendix F, which is all of Ord’s general policy recommendations put into one place. As policy prioritization has long been an interest of mine, I found Toby Ord’s answer to be quite fascinating. I wrote a post a few months back discussing the highest impact policies actually being discussed in American politics in this election cycle. Comparing it to Toby Ord’s recommendations, the overlap is essentially nonexistent except for some points on climate change, which most democrats support such as the U.S. rejoining the Paris Climate Agreement. There’s also a point about leveraging the WHO to better respond to pandemics, and given Trump has essentially done the exact opposite by removing U.S. funding for the WHO, I suppose I should at least include that as relevant policy debate.

I want to emphasize that Ord has 9 pages of policy ideas, and many of them are likely uncontroversial (improve our understanding of long period comets, have the Biological Weapons Convention have a real budget), but our political system is failing to even address these challenges, and I think it’s important to highlight that.

There is room for optimism; human knowledge is improved by discussion and research, and that includes reading and blogging. If you find these ideas interesting, or even more broadly, if you think there are valuable things in the world, one of the most effective activities you could do this year might be to just read The Precipice. Even without the weight of humanity, the concepts, problem solving, and prose are worth the read all by themselves. This is definitely by favorite book I’ve read this year, and I’ve skipped over summarizing whole sections in the interests of time. Ord even has a whole uplifting chapter about humanity’s future potential, and is overall quite positive. Please attribute any gloominess on this topic to me and not the book.

And if you do read this book, it also just makes for intriguing conversation. I couldn’t help but tell people about some of the ideas here (“are supervolcanoes a national security threat?” ), and the approach is just wonderfully different, novel, and cross-disciplinary.

For more on this, but slightly short of reading the whole book, I also recommend Toby Ord’s excellent interview on the 80000 Hours Podcast. On that page you can also find a host of awesome links to related research and ideas about existential risk. I’ll also link Slate Star Codex’s longer review of The Precipice, and places to buy it.

Martian Constitution Brainstorming

This post started out as an attempt to draft an actual constitution for a Martian colony. It quickly became apparent that such a project was pretty difficult and relied on too many variables. Consequently, it’s apparent that before writing a constitution, we have to discuss all the decisions and challenges surrounding what sort of constitution should exist in the first place. Elon Musk talked a while back about possible governing mechanisms and it seemed to me that he had evidently not thought about it very hard.

Any discussion of Martian law isn’t complete without starting with the similar governing international treaties we have right now. This Cato Institute piece by Edward Hudgins from 20 years ago is actually a good place to look. I’ll be summarizing his discussion here.

The Antarctic Treaty was signed by most large countries in the world and leaves Antarctica for scientific research and ecological preservation. Economic activity is essentially banned. Hudgins states: “This clearly is not the model for Mars.” Next is the UN Convention on Law of the Sea, which the US has never signed, although it largely agrees with the treaty as codification of maritime law. Hudgins opposes it citing provisions that could be used to “tax” economic activity in the open ocean and redistributed to inland countries. Practically speaking, I couldn’t find any evidence of such a thing happening, but I’m not familiar with fishing in international waters, and I’m sure someone else who knew more could properly complain about it, as one can about all human governance systems. I would state that this treaty doesn’t seem like a good model for Martian colonies either.

Hudgins next discusses the INTELSAT agreement, but that was privatized into the Intelsat corporation in 2001. Apparently it was an attempt to have an intergovernmental organization that provided communications between countries. I don’t have enough information to see whether it was better or worse one way or the other, but I find it telling that it was privatized and has remained so.

I should note that I’m not the only one that has been thinking about governance on Mars. There was a short-lived (as far as I can tell) conference called the International Extraterrestrial Liberty Conference, mentioned in this BBC article. Scientists and philosophers brainstormed about potential constitutions on Mars as well as legal and economic problems we might see. The astrobiologist mentioned in the article, Charles Cockell, actually has two books published, alluded to in the 2014 BBC article: Dissent, Revolution and Liberty Beyond Earth, and its follow up Human Governance Beyond Earth. Both are over $100 and I run this blog as a fun hobby, so I will not be buying them. Maybe Dr. Cockell has figured out all the problems with human space exploration himself, but I doubt it.

Next, take a look at James P. Howard’s presentation on the topic. Howard has an interesting background of computational statistics and public policy, but I don’t believe Martian law is his main subject matter (he also publicly admits to supporting the Ex-Im Bank??). Nonetheless, his was one of the easier discussions of Martian law I could find. He discusses the history of law in areas of colonization and establishing of states. We can expect that Mars, like past colonization in the New World, will have some sort of government pretty quickly, and it will have some democratic components. Howard also points out that we are unlikely to see a “global” Martian government, especially right away. Colonies will be small, sent by nation-states or companies, and they will be separate. Local self-government will be the most important form, as governance from Earth seems too remote.

A major takeaway from looking into this was that there is little research available in this area, yet plans are being made to go to Mars in a very short period of time. There should be work done on the relevant questions prior to going, but many remain unanswered. Meanwhile Elon Musk is tweeting that his Martian colony will be run by direct democracy, but what about its relationship with SpaceX, the U.S. company? Will it be subsidized? Will it receive free shipments of goods? If it doesn’t, is anyone able to be sued? What about its relationship to the United States? Will there be courts? Will there be a police force? How many people will be transported there before some of these governmental structures are in place? Some institutional structures should be predetermined, but which ones? What will be done on the fly by colonists who are also trying not die from all the lack of air, water, food, and radiation shielding on Mars?

There needs to be a comprehensive research project at SpaceX looking into what sort of legal system will exist on the Martian surface, and what can be planned out. All of these questions are also intertwined. If the colony is reliant on Earth corporations for shipments of goods, then the company has de facto veto power on any legal changes the colonists propose. Protections of colonist rights ought to be written down prior to any life or death crisis arising. Colonists will be presumably given all sorts of training in terms of physics, chemistry, rocketry, healthcare, etc. In a group any larger than a dozen or so, there will soon be the need for law, and I don’t think current efforts have begun to tackle it.

I wanted to go into a deep discussion of the minutiae of different governing and voting systems in order to tinker and figure out which system is the “best”. It’s quite possible that eventually the Martian surface will house many different colonies that can experiment in governance, but right now the concern isn’t that a Martian local government won’t have planned for an efficient voting system, but rather that there won’t be any good planning for governance at all.

Bitcoin and Energy: Everything is Actually OK

I found many arguments that Bitcoin wastes energy to be lacking, so I decided to write up this post. However, it’s gotten pretty technical, so be warned.

TL;DR:

Bitcoin is an economic activity like any other, and thus it has associated input costs that are paid for by its users. Its costs just happen to be very clear and singular (electricity)
I go over several technical ways we could change the Bitcoin protocol to achieve Pareto improvements and why I don’t think they would work, or work only marginally.
I discuss how Bitcoin counterintuitively may help renewable energy rather than just rack up carbon emissions.

Economically Efficient

I’m defining “wasteful” as economically inefficient.

First, why I think referring to Bitcoin’s energy consumption in terms of economic efficiency makes sense.

Bitcoin mining is economic activity. It provides the service of securing and running the Bitcoin network. Like all economic services, there are costs associated with providing it. Reddit is a website that provides the service of an online forum and discussion, and it has associated costs. Some people don’t use Reddit or find that Reddit is a time sink, they might say that Reddit is “wasteful”, after all, they don’t use it, and they could think of better things to do with the resources. In another example, some people might be uninterested in baseball and thus believe that the New York Yankees are a waste of resources. They use space for their stadium, training facilities, and offices, they use energy to run those facilities, and they use advertising space to promote their organization where more useful things could be advertised, like charity. I think this usage of “wasteful” is fine for both Bitcoin and the New York Yankees, but this seems to be more an argument about preferences or about Bitcoin being a bad thing that exists in the world regardless of its resource use. There’s nothing really economically inefficient with Reddit’s existence or the existence of the Yankees; people want the services those organizations provide, they provide them, and they fund it through ads on a website or ticket sales to baseball games.

Bitcoin mining provides something that literally didn’t exist before 2009; the ability to send digital value across the internet with no third parties, or at least no specific third parties. This is technically impressive and apparently highly valuable. To do this, the Bitcoin network had to solve all the problems normally solved by the banking and payment system, including how to prove authentication when sending money, how to check you actually have the money you are sending, how to avoid double spending, how to achieve consensus on the current state of the network and transactions, and how to survive malicious attacks on the network, all without state support, or in fact any third party of any kind. This is remarkable, and to provide this difficult service securely, the network relies on Proof-of-work by miners. Just like the New York Yankees, Bitcoin’s services aren’t used by everyone, yet they still have costs associated with with providing their services, used by many around the world. This work done by miners isn’t wasted any more than the Yankees’ investment in staff, facilities, or their brand.

So how do Bitcoin “consumers” pay for their service? Each transaction has a fee associated with it, which is given as a reward to miners for including the transaction into the next “block” or batch of transactions. Additionally, the protocol slowly adds new Bitcoin into the system by including a block reward for the miner who finds that block. Thus each block has a reward for the miner, which is what Bitcoin users are “paying” to miners to keep the network safe.

In fact, this is where articles that discuss how much energy Bitcoin is using come from: they are taking the current market value of the block reward (12.5 Bitcoin @ $5000/Bitcoin today) and multiplying that out for a year. One block every 10 minutes and 525,960 minutes in a year means 52,596 blocks worth a total block reward of $3.3 billion a year and probably more with transaction fees. Bitcoin mining is competitive; you only get the block reward if you solve the hashing problem first. Consequently, margins are tight. That means there are big incentives to only use the most efficient hardware (efficient in terms of hashes/kilowatt hour) and the cheapest electricity. Depending on your estimate of what miners are paying for electricity, you can divide 8 cents a kilowatt hour or whatever into the ~$3.5 billion to get a these massive energy estimates. Of course, we should note we don’t really know what percent of earnings goes into R&D for ASIC design and manufacturing costs, rent, etc.

But in my view, it doesn’t really matter; all of these costs are paid for by Bitcoin users in order to use the network. The energy input into the Bitcoin network is determined by the block reward and price of Bitcoin; if Bitcoin and Bitcoin transactions are in demand, the block reward is higher, and miners spend more resources on energy and chip manufacturing. If the demand is lower, they spend less. It’s analogous to people who pay to go to Yankees games; if the demand is higher and fans are willing to pay more for tickets, the Yankees can spend that on advertising, improving the stadium, or getting better players, etc.

Technical Improvement Proposals

Next, technical counterarguments would demonstrate that this ability to send value across the internet with no third party can be done for cheaper than is currently available with Bitcoin. I will now go over the best arguments I know of.

No Block Reward

Currently, miners complete work through hashing to solve a difficult problem. The problem can only be completed with brute force, meaning you just need to run as many hashes as possible to solve the problem. The way this is done today is with specialized hardware and electricity as noted above. However, it is a protocol design that Bitcoins are added to the network with each block that is mined (and this is a pretty useful idea for bootstrapping a digital currency when none existed). We can quickly imagine a cryptocurrency that is identical to Bitcoin except no new coins are added, the ~17 million that currently exist are the grand total. Each block only gets transaction fees rewarded to the miners.

It seems in this case that we have recreated all the value of the Bitcoin network for users, but have reduced the input of electricity on the cost side. Ignoring the usefulness of the block reward for bootstrapping Bitcoin, isn’t this an efficiency gain?

It depends on how liquid and efficient the Bitcoin/dollar exchange market is. If Bitcoin was unexpectedly changed so that no more coins were created, the market cap of Bitcoin (total units of account * exchange rate per unit) would soar, by the net present value of all future coins that are no longer going to be produced. It’s hard to say exactly what that is, but it’s about 3.6 million coins, with the majority produced in the next 10 years. The value would be at least a few billion dollars, maybe as high as $10 billion depending on the discount rate. The current market cap is about $87 billion so that’s not a huge percent increase in price, but it’d be notable. That means if transaction fees in Bitcoin remained constant, in dollars they’d increase by a similar amount forever, unlike the fixed block reward which decreases over time.

In an efficient market, we’d expect this increase in transaction costs due to the higher market price of Bitcoin to exactly offset the reduction in block reward. In other words, current Bitcoin holders are “paying” for the block reward though a reduction in net present value of Bitcoins they hold today. Simply removing the block reward doesn’t change that, it just moves value around.

However, markets may not be that efficient. Dollar/Bitcoin exchanges seem relatively liquid, but they are certainly less liquid than traditional stocks. Additionally, I’m not sure you can borrow money from an exchange to invest in Bitcoin, which is also a sign of an underdeveloped market. So we can imagine a hypothetical world where the Bitcoin protocol originally had much smaller block rewards or had reduced its block reward more quickly. In this world, it’s possible we could achieve a net decrease in total energy expended even with a likely higher price. However, I’m not sure we could know which direction the market inefficiencies would go; perhaps the price of Bitcoin would “stick” higher, meaning higher priced transactions outweigh the reduced block reward.

A final point to be made for this counterargument; transaction costs don’t have to remain constant in Bitcoin terms in this hypothetical. If Bitcoin prices go up, transaction costs might remain the same in constant dollars since users will probably continue to demand transaction space on the blockchain at the same level as before. We do however, have empirical evidence that transaction fees in dollars have correlated with the dollar/Bitcoin exchange rate, and even perhaps in pure Bitcoin terms.

If this argument is true (and so far I don’t believe it is), it should also be noted that it would imply Bitcoin will get more efficient over time as it moves towards smaller block rewards.

Inefficiencies Due to Fixed Bitcoin Protocol Constants (Block Size)

In a theoretical free and efficient market, consumers demand goods with downsloping demand curves, producers supply the good with upward sloping supply curves. Where the curves meet, there is a market clearing price and quantity. In Bitcoin, consumers demand transactions (or transaction space) on the blockchain, but producers don’t produce blockchain space; they produce hashrate, or perhaps “security”. The transaction space is fixed by protocol.

This means that total transaction fees could theoretically be lower if the fixed transaction space (i.e. block size) was changed. This would be determined by the slope of the demand curve for Bitcoin transactions. A shallower slope would mean shifting the supply curve to the right will increase the value of total transaction fees, even if each individual fee drops. A steeper one would mean total transaction fees could drop. It’s been pretty common for Bitcoin blocks to be less than the maximum for the past year, so I’m skeptical that a larger block size would lead to a drop in total transaction fees, and I suspect even if it there was a drop, I don’t think it would be massive in magnitude. Nonetheless, I think this is an argument that Bitcoin could achieve the same ends and be slightly less wasteful, but it needs more empirical evidence.

Proof-of-Stake

This section is a reiteration of Paul Sztorc’s “Nothing is Cheaper Than Proof of Work”.

Suppose instead of hashing and Proof of Work, new transaction fees went to current holders of Bitcoin. This “Proof-of-Stake” would demonstrate commitment to the blockchain’s success not through investment of mining hardware but rather direct demonstration of stake in the blockchain through ownership of the currency.

Firstly, I still have some doubts that this allocation of new currency can actually be done without any actual work. Casper (Ethereum’s proposed Proof of Stake system) requires some calculation as to which random number to pick, which determines which staked coins get the transaction fees. It would be highly valuable to affect that calculation, and it seems optimistic to suggest there will be no way to influence it. But Ethereum isn’t my specialty, so I’ll concede it’s actually possible despite it not existing today.

In Bitcoin, miners spend the equivalent of the block reward and transaction fees every 10 minutes in order to compete and be in the best position to obtain that reward. In Proof-of-Stake, Validators still have to deposit coins to be staked. They risk these coins, because if they misbehave and disagree with other Validators, they can lose them. The amount deposited determines likelihood of receiving the transaction fees, and so these are just a form of “bonds”. Returns to bonds are analogous to returns to mining resources. Value locked up in bonds could have been used in other more productive parts of the economy. The opportunity cost isn’t as externally obvious as the electricity used in Bitcoin mining, but it is nonetheless there and it is identical. Locking up value in validation bonds isn’t a permanent thing, whereas investing time, money, ASICs, R&D, and electricity in Bitcoin mining cements that value into silicon and heat which can only be used for one thing. Thus the returns to mining are going to be higher on a per percentage basis to account for the increased risk.

We already pointed out that the cost of Bitcoin mining is a consequence of the block reward. The block reward if Bitcoin switched to a Proof-of-Stake system would still be the same. But because buying validation bonds isn’t as risky as tying up resources permanently into silicon and electricity, there will be significantly more resources tied up in Proof-of-Stake for any given level of block reward/transaction fees (because the market will keep putting more until the rate of return reaches the market rate for the given risk level). There is thus no free lunch with Proof-of-Stake; users of Bitcoin are auctioning off a block reward/block transaction fees worth of value every 10 minutes, and so a competitive market will form to always provide that value at that opportunity cost, whether that cost is through validation bonds or mining.

My view here is agnostic on whether PoS is a “better” system than PoW, just that PoS doesn’t eliminate the mining cost from the system.

“Useful Work”

What if you used Bitcoin mining to do “useful work”? One counterpoint is that mining is already useful work, since Bitcoin users are paying billions of dollars for it a year. Another is that using Bitcoin for some useful work wouldn’t change anything if miners can capture the benefit of the useful work. For example, using mining rigs for heating in the winter allows you to profit more. But this is equivalent to an increase in mining ASIC efficiency which happens all the time. The network uses this extra efficiency to increase the hashrate, the difficulty level adjusts (the network aims to always have a new block average every 10 minutes) and we are back to where we were, same energy used, but now with a higher hashrate.

However, what about useful work that was a positive externality? For example, finding prime numbers? Assuming away all the difficulties with this specific example, like how hashes are much easier to check than prime numbers, if the work resulted in a true positive externality public good, like information becoming public, then that has to be an efficiency gain.

It should be noted that the work can’t be too useful because if it’s profitable enough where any single individual could benefit given the cost to mine, then lots of people would start mining for the benefit of the work itself. In which case, this would be treated again like an increase in efficiency with the difficulty level increasing significantly until the marginal cost of mining again equaled the total marginal revenue of both block reward/transaction costs and the public good. But assuming it’s not usually profitable, the benefits could be so spread out across society that there is no way for an individual to benefit, yet there benefit at the societal level. I just don’t see “finding prime numbers” as fulfilling that value, but I’m open to other suggestions. Given the current value of mining is over $3.5 billion a year, I think the useful work would have to have a value that’s a significant fraction of that to matter in terms of efficiency gain.

Carbon Emissions, Regulatory Arbitrage, and Renewal Energy

Bitcoin mining is location independent. That means it will only be undertaken in locations where the input values are cheapest in the world. We don’t actually know where Bitcoin mining is done, but we have some guesses based on information in blocks mined by companies and where the coins are deposited (see this article). The majority is certainly in China due to proximity of the world’s computer manufacturing base there. Miners in other countries would have to wait for mining material to be shipped to them, which could be out of date by the time it gets there. Eventually, we would expect diminishing returns to slow the rate of improvements in ASICs, which would allow non-Chinese miners to utilize mining equipment before it becomes antiquated. That means they could use their locally low price of electricity to their advantage.

That also means that the Bitcoin network could be optimizing for polluting energy, like fossil fuels that are incorrectly priced (i.e. lack of carbon tax). A country that creates a carbon tax would make fossil fuel energy more expensive, and Bitcoin miners there unprofitable, so they might switch to a country without a carbon tax, thus polluting more. This is a regulatory arbitrage and is an efficiency loss.

However, there are caveats to this argument. One is that many countries, including the ones with the most Bitcoin miners, China and the U.S., never had carbon taxes. Bitcoin blew up there because of their technical advancement and network effects of their tech economies (hardware and software respectively). If they were to implement carbon taxes, and miners then left, that would be an inefficiency brought about by Bitcoin.

Another caveat is that Bitcoin is highly efficient in finding the cheapest energy sources. Many renewable sources of energy are very cheap on a per kilowatt hour basis, and so Bitcoin has actually acted as an incentive for expanding renewable energy (see Morocco).

Bitcoin’s monetary existence, unstable though it is, provides a floor underneath which states can no longer mismanage their currency, or else those states risk their population turning to Bitcoin instead. Similarly, Bitcoin mining’s existence means that there is a floor under which local energy prices won’t be able to drop. This is good, as locally cheap (not globally cheap!) energy means that demand is lower relative to supply in a given area, but it’s too expensive to build transmission lines to other areas where energy is more in need. Compared to a world without Bitcoin mining, mining creates value from cheap local energy which can then be transported digitally. The beneficiaries are the users of Bitcoin who get a payment network that literally didn’t exist before. It is paid for with locally cheap energy around the world that had excess supply. There are also secondary effects as users and miners are better off and the wealth effect on their behavior will be to increase some spending, some of which should enrich people who live in already energy expensive areas. This means some people in expensive energy areas will see a cheaper relative cost of energy.

Final Notes

A couple other arguments that I hear a lot but I don’t consider to be challenges to this view.

Bitcoin mining leads to centralization. This is true empirically, but not an argument that it’s wasteful, just that it’s bad for Bitcoin.
Bitcoin uses a lot of energy. This is basically the argument I’m opposing and it is very common. I’m not saying Bitcoin doesn’t use a lot of energy, I’m saying it provides a service and has associated costs and expenses.
Bitcoin has no use cases. The empirical evidence seems to contradict this, as billions of dollars of Bitcoin transactions happen every day. If you need some more discussion on what Bitcoin is used for, check out my previous post on the subject, or check out this useful page from the EFF on how payment service providers can be used to censor free speech.

Overpopulation on Mars is a Problem Today

This blog is somewhat concerned about machine superintelligence and promotes the idea that we should be researching how to fix the AI goal alignment problem now.

A few years ago, Brian Ng compared concerns about machine superintelligence to worrying about overpopulation on Mars:

If we colonize Mars, there could be too many people there, which would be a serious pressing issue. But there’s no point working on it right now, and that’s why I can’t productively work on not turning AI evil.

I don’t know how seriously anyone took this quote, as it seems like most people would quickly point out that the conditional statement is already almost true; there is no “if” in colonizing Mars, there are people trying to figure out how to get there right now. But the generic version I’ve heard is that worrying about AI is like worrying about overpopulation on Mars, as if overpopulation on Mars would look like overpopulation on Earth. But I don’t see any reason for this to be the case.

Oh, look, it only took 5 months to go from landing 1 person on Mars to Mars being overpopulated.

— Eliezer Yudkowsky ⏹️ (@ESYudkowsky) March 9, 2016

Mars isn’t overpopulated at this moment, but we have the technology to send a person to Mars right now. The Falcon Heavy rocket launched a car into near-Martian orbit in February. NASA has already sent a car-sized payload and landed it on Mars in 2012. So why haven’t we sent a human to Mars? Because if there was a human on Mars, Mars would immediately become overpopulated. A person wouldn’t have food or shelter or water if they just showed up on Mars today. A lot of the work being done to set up any kind of human mission to Mars is to solve the problems you think of when you think of overpopulation, like food and shelter.

Of course, that’s simply a critique of a poor analogy, not a response to underlying point. However, the response is just as compelling a point: just because there is a belief that machine superintelligence is far off, it does not follow that there is nothing to be done now. On the contrary, solving the AI Alignment problem must occur prior to the development of Artificial General Intelligence, or there could be very dangerous consequences. Researching how we might be able to convey to an intelligent agent the complex set of values that cover human beliefs is a daunting task. Working on it now seems like the least that we can do if there is even a small chance of existential risk.

Things Mars Can Export

Imagine an economy on Mars or the Moon. Some people move there, excited about the chance to live on the new frontier of human development. They set up shop and start providing services and goods to each other. They’ll need jobs like metallurgist, construction worker, physicist, farmer, material scientist, physician, etc. The technician with the 3D printer will trade with the computer repairwoman who will trade with the doctor who will trade with the solar panel producer. Certainly very early colonies won’t do much trading and work more as a cohesive unit, with specific team members and job duties, running as a ship’s crew or military base more than a small town. But eventually once the outpost becomes big enough, trade will occur.

What will they use for money? Well, it’s not the point of this post, but I’d like to take a moment for speculation anyway. It’s possible that the centralized group that creates the space colony will see this problem ahead of time and establish their own currency. But it wouldn’t take much for someone to introduce a simple Bitcoin fork to create their own blockchain. Such a currency would only require an application running on a computer, something the colony would likely have lots of. I’ve written before that Bitcoin isn’t that great of a currency at present, but it does have benefits if your local currency is already pretty bad, see Venezuela. On Earth it functions as a floor beneath which local currencies can no longer do worse than. However, on Mars, the local currency would have a lot of uncertainty surrounding it. Maybe a bank has been set up, but why would people trust the bank? It doesn’t have a long history of trustworthy monetary policy like some Earth central banks do.

Suppose SpaceX sets up a Martian colony and creates a bank to hold everyone’s dollars. It’s too long to wait to ping back to Earth (average message/reply round trip is 6-40 minutes) to transfer Earth currency, so you’re left with trusting SpaceX’s bank that they won’t mismanage the currency, even though SpaceX doesn’t have any experience managing currency. Well, what if some enterprising people come over with a “MarsCoin” clone of Bitcoin, running a blockchain on Mars? You can send the code to Earth for audit, then run the code yourself on a local network of servers. If the block time is lower than it takes for a return ping to Earth (almost always true even if you’re using the 10 minute blocks of Bitcoin, which could be shortened), then you have a decentralized, trusted currency that’s much simpler than relying on distant Earth institutions or unproven Martian ones. Of course, then you need a special interplanetary blockchain to move currency between Earth and Mars with a very slow block speed, but we’re getting off on a tangent.

We have established that there will be a way to trade on Mars and between Mars and Earth. Mars will obviously have things it needs to buy from Earth, like difficult to manufacture parts, engineering and consulting services, and likely entertainment. But what will they be able to export to Earth? A country that can only import goods probably can’t sustain itself, unless there was positive immigration. We can thus treat reasons to immigrate to Mars as part of the answer to what Mars can export. I have compiled a list of possible ways the Martian economy can maintain a stable exchange rate with Earth.

Science

Science is the clear primary export of Mars today. As Robert Zubrin states in this excellent video, we believe Mars had liquid water on its surface for a long time, perhaps a billion years. On Earth, life showed up much sooner than a billion years after liquid water appeared. If we go to Mars and find evidence of life, that would help prove that the development of life is pretty common in the universe. On the other hand, if we don’t find much evidence of life, that could help point out that life isn’t very common on in the universe. Evidence of current or past life, could also help us determine whether all life is similar to that of life on Earth, using DNA to create amino acids and so on. These questions are “…real science. This is fundamental questions that thinking men and women have wondered about for thousands of years.”

From an economic standpoint, it’s clear that scientists on Earth are willing to pay billions of dollars for this data. They might send their own scientists to Mars, which would count as positive immigration, or they might hire Martians to conduct the science for them, but it’s clear that scientific interest in Mars is worth billions. It’s also worth mentioning that other celestial bodies would also have this benefit, the Moon likely less so than Mars, and Jupiter’s moons perhaps similarly valuable.

Tourism

Space tourism also seems like an important industry for a space colony, although perhaps not Mars. Andy Weir’s Artemis takes place on the Moon and space tourism is an important export industry. Tourists visit the Apollo 11 landing site and jump around at 1/6 Earth gravity. It seems likely that similar tourism might function on Mars, although there are serious limits. For example, the trip to Mars takes months, would occur in essentially zero-g, and usually is only done one way every two years. Large rooms with windows would seem both vital for any tourism industry and highly impractical without radiation shielding. Vacationing for a week on Mars wouldn’t be practical, unlike taking a week or two week vacation to the Moon. Perhaps a two year long sabbatical would appeal to some people, but probably not most. A more efficient transportation system, such as an ion drive ship between the planets might reduce the travel time or allow for interplanetary travel even when the planets are not close in their orbits. But until then, tourism is likely to be very limited.

Martian Souvenirs

If tourism is limited, perhaps actual Martian rocks will be more highly valued. Those might be much simpler to ship to a mass audience on Earth. Given the monopoly Martians would have on this industry, they could theoretically charge a hefty markup. This is, of course, in addition to the scientific relevance the rocks would have. On the less positive side, there may be regulatory issues for bringing alien rocks to Earth. Fears of alien bacteria that could kill everyone on Earth might cause prohibitions. Such a threat seems unlikely, but import bans are often irrational.

Reduced Launch Costs

Of all the rocky places in the solar system to launch a rocket, the surface of Venus with its Earth-like gravity, sulfuric acid rain, and insane heat, is probably the worst one. But Earth is a close second. Rocket launches must accelerate out of the “gravity well” of wherever their launch site is and then expend the energy needed to get to their target transfer orbit. Escaping the gravity well is very difficult on Earth’s surface. The Apollo program required very little fuel to get off the Moon’s surface because the gravity was so reduced (and the Moon has no atmosphere). Thus, the energy needed to get off of other rocky surfaces in the solar system is much less than Earth. Mars also orbits further out from the sun, which means you also need less energy to get to a transfer orbit to the asteroid belt or outer solar system. Thus, one economic benefit would be to offer Mars as a cheaper launching point for outer solar system exploration.

However, if you need to first bring an entire rocket and fuel to Mars before departing for your eventual target, you’re better off just leaving from Earth directly without dipping into the shallower, but still significant Martian gravity well. For this to be valuable, you’d need to be able to build the rocket on Mars, which seems much further off than the other ideas discussed here, or you’d have to create the fuel on Mars. This may be possible; SpaceX’ plan is to use the carbon dioxide in the Martian atmosphere and water ice on or near the surface to produce methane and oxygen as rocket fuel. One might also be able to find an asteroid or comet with water ice, carbon dioxide, or even methane, and bring it into orbit around Mars or the Moon, avoiding having to accelerate it into and out of Earth’s gravity well. Rocket manufacturing could then use a combination of terrestrial, martian, or lunar parts and fuel, supplemented by materials already in orbit, or even manufacturing plants in orbit. This could be sustainable in the long term. Asteroid mining would be extremely lucrative, with perhaps tens of billions of dollars in potential profit. If an asteroid mining company believed Mars, or the Moon could save them launch costs, they would likely purchase property, bring equipment and employees, and thus improve the Martian trade balance.

Natural Resources

Mars also has an abundance of natural resources in various compounds where they may be more common than on Earth. These include halogens, organic compounds, and deuterium. The extent and speed at which these will become useful to export from Mars are unclear, but they may eventually be useful as a source closer to the outer solar system than Earth, due to the reduced launch costs. Editing note: I decided to add this as separate category later after discussing this post.

On-Site Entertainment Production

This is another odd case, but it may be more immediately plausible than launching exploratory rockets from Mars. The low gravity of Mars or the Moon would allow for a different type of videography, with actors that actually weigh less able to jump large distances. This could be literal as for science fiction films that need to take place in low gravity, or perhaps more creative and unusual projects that endeavor to take advantage of this location in ways we have not yet conceived of. Again, however, the Moon may be a more accessible location than Mars, given the length and difficulty of travel.

Medicine

From what we know about the human body in lower gravity environments, the health effects are largely negative. There may be the possibility that osteoarthritis would be improved by living in a low-g environment, but anyone who has that problem would likely be old and might suffer from other problems. However, the severity of injuries from falls would likely be much less in low-g environments, and perhaps high blood pressure would be less of an issue. Nonetheless osteoporosis is known to get worse in zero-g, and so astronauts are forced to do weightlifting regimens in space to try and maintain their bone density. It’s possible that if the osteoporosis could be combated, people who suffer from osteoarthritis might live more comfortable retirements on Mars or the Moon. They could likely never return to Earth though, as their heart and muscles would have become significantly weaker and would likely give out upon return to Earth. We need more research on the human body under low-gravity conditions to see what the long term medical effects are.

Law and Society

This is much more likely to be a reason for positive immigration than for exports. People who currently live under legal systems on Earth may be interested in moving to a different legal system not currently available on Earth. They may even want to start their own society. That is not currently possible on Earth, as states claim almost all land on Earth. Moving to Mars due to its nonexistent or only slightly existent legal system would bring with it a variety or related goals and ideas, such as chance to shape the cultural future of Mars. One can see the related desires of immigrants who moved to the Americas from Europe in the 18th and 19th centuries (and earlier, I suppose). This would also include people who wanted to live on the edge of human exploration or contribute to the project of humanity reaching other celestial bodies, and perhaps eventually other stars.

There are of course limits here as well. Any existing Martian colony would have its own law and rules that might be just as restrictive as Earth, especially given the harsh conditions and lack of natural resources on Mars. To create a legal system and society on Mars, any group would require a massive amount of resources, equipment, and specialized knowledge, essentially creating a whole new colony. Could such a move possibly be cheaper than simply establishing a place on Earth? It’s hard to say. The problem on Earth is not just money and resources, but confronting states who have access to military power.

Capital Flows

It should be pointed out for completeness’ sake that of course Mars could potentially export nothing, but still expand their economy as Martians did productive work for each other. Then could borrow money from Earth, invest it, and Earthers would see a positive return. This would counter the negative pressure on the Martian exchange rate from exports. If Martian investments were failing though, then Martian currency would likely become pretty worthless to Earthers, and they’d likely stop selling goods to Mars. We should thus note a financial crisis could be quite problematic!

Conclusion

Mars has a variety of goods it can export, although the only immediately available ones are likely Science, Society, and perhaps space rock souvenirs. Others may eventually become economically useful, but will likely take some time.

Picture: Public Domain Image, Global mosaic of 102 Viking 1 Orbiter images of Mars taken on orbit 1,334, 22 February 1980. NASA

Prediction Markets: What are they and why are they useful?

Given the importance of the topic to this blog, I thought it best to create a discussion of prediction markets here that I can refer back to. In the process of researching this topic, I found some other resources of varying quality which I will be referring out to as necessary. The best analysis is Paul Sztorc’s Prediction Market Sequence, which is unfortunately kind of buried under a subsection of the Bitcoin Hivemind website, and is also a combined ~75 pages (!). I would definitely recommend it to anyone who wants more details, but if you only want to read 3000 words, this blog post will cover a lot of the main ideas.

What is a prediction market?

Prediction markets are exchanges where you can buy and sell “derivatives” (also referred to as shares, contracts, options, depending on context) on whether a given event will occur. They can be simple yes or no questions, or they can be more complicated, which we will cover later. The market price reflects what people believe to be the probability of the event occurring. A simple example would be a prediction market on whether Donald Trump will win the 2020 election. Suppose each contract resolves to $1. If you buy a “yes” option at 40% or 40¢ and then Trump wins, you’d be able to redeem it after the event at 100% or $1. If he loses, the share value would be 0% or $0. Other prediction markets can use any pricing scheme reflecting 0-100%.

PredictIt.org, a well known prediction market site, uses shares that resolve to $1 or $0 with trading occurring in between those prices until the event occurs. A 70¢ price would indicate a market belief of 70% in the event occurring. This is similar to betting markets on external events (like betting on the outcome of sports events) and also futures markets, where additional contracts can be created as long as there is interest in trading them, unlike stocks or bonds which are issued by an organization in limited quantities.

For more info, check out the first three sections of Paul Sztorc’s first PM paper.

What’s the point of a prediction market, is it just a way to gamble?

They can be used for gambling, but that is not their most interesting use case. Prediction markets aggregate information, and so they are sometimes called information markets. They offer a profit incentive for someone to share information publicly or for you to investigate something to try and uncover information. For example, in 2016, there wasn’t a great deal of presidential election polling in Wisconsin and other Midwest states. You might have been able to leverage information you found out about Trump’s support among Obama voters in those areas to place bets on Trump doing better in the electoral college. While talking heads on TV get paid for giving opinions, they don’t get paid for correct ones; prediction markets change that, allowing for a direct reward mechanism for correct predictions. Prediction markets thus improve the quality of predictions while simultaneously cutting through the self aggrandizing opinionators who have nothing on the line.

Aggregation of information is a concept so broadly applicable, it’s hard to convey the potential impact of its widespread use. At the very least, we would have a significant improvement in forecasting due to the profit incentive of those with useful information as well as those who can find out information and profit off its discovery. In the best case, we can aggregate all human knowledge into singular probabilistic forecasts about the future, providing real world application and feedback of empirical methods for everyone’s benefit.

If prediction markets are so great, why aren’t they being used already?

They are being used in limited ways, although they face many challenges. One is legal; the US has made it illegal to gamble on the internet. That means no large scale prediction markets can function in the US. PredictIt.org, a well known politics prediction market, has limitations on the amounts individuals can stake in a market as well as the number of individuals allowed in a single market. This limits the liquidity and financial interest in prediction markets. There is also the fact that knowledge about them isn’t widely dispersed; it’s a niche topic for obscure blog posts and academic papers.

Even if banned publicly, private organizations have an incentive to use prediction markets to improve their forecasting. Why haven’t more done so?

Prediction markets don’t necessarily mesh well with human organizations; managers may not be optimizing for the best information, but rather the decision making policies that make them look good. Dan Ariely writes how resistant companies are to conduct experiments that would actually give them better information; experimenting on customers is seen as immoral even if that results in a company never improving the products it provides for fear of trying new things. In addition, as Bryan Caplan points out, markets themselves aren’t very popular generally, and the Senate shut down a Defense Department experiment on prediction markets because Ron Wyden found it “grotesque”, despite the obvious national security incentives to learn as much information about security threats as possible. Paul Sztorc discusses similar challenges prediction markets face in his first paper, linked earlier. Prediction markets challenge the status quo and conventional wisdom and thus may never be adopted in organizations with established hierarchies.

I believe that in addition to the points Sztorc makes, prediction markets are simply expensive to set up. You need a market infrastructure (likely a digital platform, secure database architectures, etc) but also a mechanism for resolution, a decision structure for what markets will be created, and of course money to seed the market (If you’d like to learn more about market seeding, this blog post dives into Hanson’s logarithmic market scoring rule). You also need people to know to go to your market to make predictions; if no one is going to make predictions or buy shares on your market, other people won’t be incentivized to either. This is a network effect that may be hard to overcome.

So why are we talking about them if they can’t exist in real life?

One reason is to increase the awareness of the potential of prediction market to the readers of this blog. But the big reason I’m talking about prediction markets right now (as well as Paul Sztorc) is because of the rise of cryptocurrencies. Many of the social challenges to prediction markets can be overcome by instituting prediction markets on decentralized platforms and trading the contracts in cryptocurrencies. If you need more background on cryptocurrencies, I have some previous blog posts.

Prediction market projects on distributed blockchains, like Bitcoin Hivemind and the recently launched Augur, cannot be blocked by traditional legal prohibitions. They also don’t need permission from the established experts in a field in order to make predictions. They are, however, more expensive than prediction markets like PredictIt or Betfair; they may require additional money to transact and specific cryptocurrency expertise to enter, and they risk uncertainty in how dispute resolution will occur in this new space. Nonetheless, many of these risks and costs may decrease over time as this technology becomes more familiar. Their location on a blockchain of course also means they are not just censorship-resistant, but effectively immortal.

Furthermore, prediction markets can do much more than the simple binary option market we discussed before.

What else can prediction markets do?

Ok, it starts to get complicated here. I’m mostly pulling from Paul Sztorc’s second paper on prediction markets (Unlocking the Power of Prediction Markets). Prediction markets can also create scaled markets instead of simple Yes/No binary markets. This is like trading a stock. You can buy Apple at 100 and sell at 200. The only difference is that the prediction market ceases after the event occurs, so instead of buying Trump at 40¢, you could buy Trump at 250 electoral votes in August 2016 and your share would mature at 304, since that is the total number he received when the electoral votes were counted in December. On such a trade, you’d receive a profit for guessing that Trump would receive more than 250, and the total profit would relate to how much more the ultimate “price” or quantity beat the price you paid.

The more interesting addition is the use of multiple dimensions. For example, you could create a market with 4 outcomes trading across two dimensions: Unemployment up year over year and whether Trump adds tariffs to Chinese imports. The resulting percentages should yield a relationship in how the world views these events; it’d be likely additional tariffs would increase the expected unemployment, but the prediction market would tell us if the expected effect were certain or uncertain. If we used a scaled dimension, we could even see the magnitude.

This is pretty powerful. I highly recommend Sztorc’s paper on this, as even more complex markets could tell us more detailed info (and he has nice illustrations). However, we should note that more complex markets require more complex contracts embedded into the blockchain. Blockchains don’t scale well (so far), so complex contracts may result in expensive transaction fees, and the multiple bets you can place may mean overall liquidity is low. If liquidity is low, people may not feel like betting is worth their time, since large bets might not be possible, reducing the incentive to research hidden information.

What are some examples of prediction markets?

The simplest prediction markets are already in existence at places like Betfair and PredictIt. For those betting markets to be useful to us, those websites have to share our interest in betting on things. And while this is generally true for broad markets (like prediction markets for president or senate seats), there are very few places you can easily place a bet on something other than company or commodity stock prices, political races, or sports. Moreover, current prediction markets have various problems, like legality and limitations on bets.

So there are simple prediction markets, like binary options on whether an event will occur, that could be useful to those who are curious about the world. Examples might include current events, like North Korean nuclear tests before the end of next year. North Korea might even buy shares anonymously (especially if this is a cryptocurrency market) which could actually directly convey information from inside North Korea in a way we never have before, albeit by directly financing a dictatorship. We will talk more about the criminal uses of prediction markets later.

Paul Sztorc points out in “Extra-Predictive Applications of Prediction Markets” that prediction markets could offer an easier way to bet on company stocks. You wouldn’t need a brokerage account with a bank, just access to a cryptocurrency. Perhaps this is more difficult for less tech savvy users to get than a bank account, but perhaps not. Many foreigners or people without good access to banking might not be able to trade in stocks. This would allow them to have a portfolio. It’s Finance 101 that you should diversify your assets to save for the future and hedge against risk, yet many people without access to banking and financial markets have no way to diversify their assets. With a prediction market tied to the S&P 500 price, you could simply buy a contract and you’d immediately have an asset that tracks the general stock market. In fact, you could use something similar to track any publicly known stock or ETF.

This would also allow for insider trading similar to the North Korean nuclear weapons market. Interestingly, this is a strength of the market, causing any insider knowledge to be quickly dispersed to the outside world. That’s actually another fascinating application Sztorc discusses: whistleblowers.

Whistleblowers risk lawsuits, job loss, prison time, and their lives, and yet they are guaranteed nothing in return, even if successful. Can we do better?

Sztorc goes on to point out that whistleblowers could collect money for their knowledge, providing more of a safe haven for coming forward sooner.

There are other fascinating applications for public construction, which would allow public betting on whether a project would complete on time. This could be used to keep public officials who awarded the contracts accountable and called out directly when making corrupt deals. You could also make markets estimating what the expected level of depreciation are for various new models of cars to assist people in buying cars that will retain their value. Nominal GDP futures markets are an idea that has been discussed on this blog as a useful tool in monetary policy, providing feedback for setting interest rates.

Prediction markets don’t even have to be used just for information gathering, as Sztorc points out. They can be used as insurance as well. There could be a market for natural disasters which allows people to place bets on events that they don’t have any additional knowledge on, but simply want to be insured against poor outcomes. They can then collect the winnings of the bet as their insurance claim.

The possibilities for simple single event prediction markets are countless. But there are even more interesting examples when we get more complicated markets.

How can we get more complicated than North Korea using insider trading to make money off of their own weapons program through online cryptocurrency betting markets?

It’s time to explore multidimensional prediction markets and in particular conditional prediction markets. Robin Hanson has suggested conditional prediction markets for publicly traded companies. One dimension would be the company’s stock price and the other would be whether they fire their CEO. It would provide immediate input from the market on whether people believe the CEO is actually useful and adding value to a company. If the projected company value was similar to or higher if the CEO was fired, that would provide good justification for getting a new CEO.

I also like the idea of a conditional prediction market for candidate policies, conditional on them getting elected. This would provide feedback on whether anyone actually believed the candidates’ promises to deliver on various policies. One can imagine cultural norms arising where candidates had to provide proof that they had purchased significant amounts of shares that they would carry out the policy if elected. Political campaigns might then be more substantial, focusing on policies candidates had demonstrated they were actually committed to.

Robin Hanson has also suggested the idea of Futarchy, where prediction markets are created to predict the outcome of various policies, conditional on their implementation. He proposes the concept of “voting values, but betting beliefs”, where legislative bodies would vote on what we would want as an outcome of some policy, and then commit to implementing whatever policy is favored by prediction markets. It’s an interesting idea although it mostly moves the debate in legislatures from “what policy should be undertaken?” to “what policy indicator are we trying to maximize?”. That may be a better debate, but if poorly chosen, the indicator could maximize something to the detriment of the polity. It’s probably still worth investigating on a narrower scale, and certainly the existence of the markets themselves could only improve the information available to policymakers.

Paul Sztorc also discusses some complex applications of multidimensional prediction markets in the Bitcoin space. Unless you are really interested in Bitcoin, they probably won’t catch your eye, so I’ll just mention them briefly, as they are technically impressive. The first is a way to allow Bitcoin investors to gain more information about possible hard forks (changes in the network and blockchain) and insure against them occurring or not occurring. Hard forks are turbulent times in a blockchain, but current discussion is just theoretical arguments of what might happen. Network effects could mean people like the idea but don’t switch to a new fork, but a prediction market could give immediate feedback and allow people to purchase shares of post-fork-Bitcoin while insured against the possibility the fork does not go through.

Also discussed is the idea of “stable” coins which would be futures markets on the exchange rate between Bitcoin and the US dollar, as well as issuing stock through a prediction market. The most interesting to me was the idea of efficiently funding public goods. The basic concept is that you create a prediction market asking whether a public good will be provided (like a lighthouse) and then people who want the public good buy shares of “no”. They won’t lose any money unless someone builds the lighthouse. People intending to build the lighthouse can buy lots of shares of yes, and, when it’s built, the lighthouse builders would reap the benefits of the shares, and those who wanted the public good would pay for it, only if it existed. You can also create lots of different possible “yes” shares, based on publicly available information, like “what letter will be painted on the side of the lighthouse?”, which would allow different teams to compete. For the full write up, check out the paper.

Sztorc concludes with generalizing this idea to “smart contracts” based on prediction markets. This is built on top of the idea that prediction markets are a robust way to determine not only the future, but how the future eventually resolves (i.e. there is a mechanism that ultimately is used to determine what happened in the 2016 election to determine whose shares are paid).

What about insider trading you were talking about earlier? Can you use prediction markets to do other illegal things too?

Mostly likely yes. Just like Bitcoin and other cryptocurrencies, prediction markets can be used for illegal transactions since they aren’t tracked well by the state, and they can be used to bet on unsavory things or incentivize criminal behavior. Insider trading is one of those activities. However, I tend to buy the libertarian argument against the illegal prohibition of insider trading. Crimes require a victim, and insider trading provides information, while its prohibition does little to stem the flow of information to those who have little concern for the law anyway.

However, stranger markets exist as well. Sztorc devotes most of his final paper to discussing prediction markets in deaths, which could be manipulated by assassins. Sztorc is skeptical this could work, as any markets that believe someone is highly unlikely to die are automatically creating an incentive for an assassin and so the market would self regulate back to equilibrium where the predicted death is both more likely and less profitable. I’m still somewhat concerned. The bottom line is that death markets provide an avenue for assassins to reward themselves that does not exist today. Perhaps such incentives would be small compared to the difficulty and bodily harm risked by assassins, but it’s unquestionable that not everything about prediction markets is purely good. That’s expected though; information is powerful and does not always lead to to purely good outcomes.

Conclusion

The benefits and potential of prediction markets to improve the world is vast. As Sztorc states: “Their greatest benefit lies in their unlimited ability to scale”. Conversations are one to one, classes are at best a few lecturers and a few hundred in the audience, while prediction markets can take input from everyone in the entire globe and condense that into comparable and understandable probabilities. Being able to retrieve the world’s knowledge by doing something as simple as checking prices online is nothing short of a revolution in information. With the advent of decentralized blockchains, prediction markets are here to stay; we should embrace their power and potential.

Podcast Recommendations June 2018

A while ago I had an idea to try and catalog everything I could about the libertarian ideas on the internet, what organizations exist, which writers are interesting, whose podcasts discuss what, and compile it into a giant document. It was ambitious, but I figured over time I could add to it until it had tons of useful information.

When the 2016 election happened, a bunch of people and organizations I had considered pretty libertarian and conservative free marketers suddenly embraced a populist with no understanding of free trade or individual liberty. Not every conservative I had listed became someone I didn’t want to recommend, and I didn’t particularly want to politicize the list based on Trump, yet I couldn’t leave many recommendations out there. There was way too much to update, and so I abandoned the project. I’ve kept the blogs list in the sidebar, but I figured it was time to revisit some podcast recommendations.

Podcasts have become an excellent demonstration of the wonders of an unfettered free market. There are podcasts on virtually any topic you might be interested in, because they are just audio files, they can be distributed in a variety of methods and platforms in a decentralized way, and many are profitable on advertising alone. This decentralization also means that there aren’t specific places you might go to find out about new podcasts, and I have personally learned about most podcasts through other people. Consequently, I wanted to pass on some of the ones I listen to. If you have some good recommendations, leave a comment here or on reddit, or tweet @postlibertarian. I’m always looking for new podcasts.

EconTalk – A great podcast covering all sorts of economic issues, this is probably my favorite libertarian/economics podcast. Hosted by Russ Roberts, Hoover Institute Fellow, and free market leaning economist, Russ does an excellent job interviewing people he agrees and disagrees with, offering skepticism, even of his own biases, and investigating claims along the way. It’s part of the Library of Economics and Liberty which also hosts EconLog, linked in the sidebar. It used to be about an hour, although nowadays he’s letting it go a bit longer.

Reason Podcast – This is podcast put out by Reason Magazine, which includes a Monday roundtable of all the senior editors at the magazine, including Katherine Mangu-Ward, Nick Gillespie, Matt Welch, and Peter Suderman, with occasional substitutions. I really enjoy it, and I recommend following them on Twitter as well. During the week, there are additional interviews, usually by Nick Gillespie, which I sometimes listen to. Also on the podcast is the audio for the monthly Soho Forum Debates, which are really interesting always debating topics of interest to libertarians.

The Fifth Column – An excellent libertarian news and analysis podcast with Matt Welch, Editor of Reason magazine, Kmele Foster, an entrepreneur and libertarian talking head, and Michael Moynahan, a libertarian-leaning columnist. It’s more news focused than EconTalk, and more crude, but I find it is entertaining and provocative.

Security Now – If you are interested in cryptography and information security, this is the weekly podcast that covers security news. It’s on the TWiT (This Week in Tech) Network, hosted by Steve Gibson and Leo Laporte. You can get a lot of the information by just reading through the show notes or the transcripts which are there on the website linked, and I’d recommend that as the podcast is pretty long. I just find there’s so much security news I don’t ever want to miss this podcast.

Rationally Speaking – Hosted by Julia Galef, this interview podcast covers a broad range of topics, but often centers around epistemology, science, and knowledge. Galef is a good interviewer and is able to engage with an eclectic and wide-ranging series of guests. I’m not sure how best to place this podcast besides in the “rationalist/LessWrong community”, but it’s absolutely not necessary to have any other interaction with that community to enjoy the podcast.

The 80000 Hours Podcast – 80000 Hours is an Effective Altruist organization focusing on how people can spend their career best benefiting humanity. The podcast, hosted by Rob Wiblin, explores various effective altruist interest areas, such as artificial intelligence, pandemic risk, animal welfare, or international development. It is worth listening to even if you aren’t interested in a career change, as they tend to go pretty in depth with each guest on exactly how an organization or person can have a large impact on the world.

The Economist Editor’s Picks – Sometimes I’m looking for a more international view of world affairs and The Economist is a lot better than many mainstream news sources. Of course, they publish way too much to read every week, so the editor’s picks podcast selects four top articles to check out. They’re usually pretty short so they’re easy to consume.

This post may end up becoming its own page, as I’ll certainly be able to update a page just for podcast recommendations.

Artificial General 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. At the very least, I’d like to signal that it should be more socially acceptable to discuss this problem.

First is the section on how I approached thinking about existential risk. My train of thought is a follow up to Efficient Advocacy. Also worth reading: Electoral Reform Fantasies.

Background

Political fights, especially culture war battles that President Trump seems so fond of, are loud, obnoxious, and tend 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. Trump seems especially good at finding meaningless conflicts to divide people, like NFL players’ actions during the national anthem or tweeting about Lavar Ball’s son being arrested in China.

Theorizing about how to combat this problem, I started making a list of what might be impactful-but-popular (or at least not unpopular) policies that would make up an idealized congressional agenda: nominal GDP futures markets, ending federal prohibition of marijuana, upgrading Social Security Numbers to be more secure, reforming bail. However, there is a big difference between “not unpopular”, “popular”, and “prioritized”. I’m pretty sure nominal GDP futures markets would have a pretty positive effect on Federal Reserve policy, and I can’t think of any political opposition to it, but almost no one is talking about it. Marijuana legalization is pretty popular across most voters, but it’s not a priority, especially for this congress. So what do you focus on? Educating more people about nominal GDP futures markets so they know such a solution exists? Convincing more people to prioritize marijuana legalization?

The nagging problem is that effective altruist groups like GiveWell have taken a research based approach to identify at what the best ways are to use our money and time to improve the world. For example, the cost of distributing anti-mosquito bed nets is extremely low, resulting in an average life saved from malaria at a cost in the thousands of dollars. The result is that we now know our actions have a significant opportunity cost; if a few thousand dollars worth of work or donations doesn’t obviously have as good an impact as literally saving someone’s life, we need a really good argument as to why we should do that activity as opposed to contributing to GiveWell’s top charities.

One way to make a case as to why there are other things worth spending money on besides GiveWell’s top charities, is to take a long term outlook, trying to effect a large change that would impact a large amount of people in the future. For example, improving institutions in various developing countries would help those populations become richer. Another approach would be to improve the global economy, which would both allow for more investment in technology as well as push investment into developing countries looking for returns. Certainly long term approaches are more risky compared to direct impact charities that improve outcomes as soon as possible, but long term approaches can’t be abandoned either.

Existential Risk

So what about the extreme long term? What about existential risk? 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 humans alive today (making it worse than a catastrophe that only kills half the humans), 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. Another way to state this is: assume it’s the year 2300, and humans no longer exist in the universe. What is the most likely cause of our destruction?

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). The chances here are thus pretty low.

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.

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, Google’s Pixel phone can create software based portrait photos via machine learning rather than needing multiple lenses. 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). 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.

Data on Accuracy

The Bigger Picture

Uncharted Territory

Existential Risk

So What Do We Do?

TL;DR:

Economically Efficient

Technical Improvement Proposals

No Block Reward

Inefficiencies Due to Fixed Bitcoin Protocol Constants (Block Size)

Proof-of-Stake

“Useful Work”

Carbon Emissions, Regulatory Arbitrage, and Renewal Energy

Final Notes

Science

Tourism

Martian Souvenirs

Reduced Launch Costs

Natural Resources

On-Site Entertainment Production

Medicine

Law and Society

Capital Flows

Conclusion

What is a prediction market?

What’s the point of a prediction market, is it just a way to gamble?

If prediction markets are so great, why aren’t they being used already?

So why are we talking about them if they can’t exist in real life?

What else can prediction markets do?

What are some examples of prediction markets?

How can we get more complicated than North Korea using insider trading to make money off of their own weapons program through online cryptocurrency betting markets?

What about insider trading you were talking about earlier? Can you use prediction markets to do other illegal things too?

Conclusion

Background

Existential Risk

Technological Existential Risk

Artificial General Intelligence

Further Reading