Wednesday, January 27, 2021

Discontinuities and Discounting

Summary: Even with a zero rate of pure time preference, and persistent zero real interest rates, we should use discount rates of at least 1% for any real-world project or policy. Health, lives, or utility should be discounted because of the probability that a discontinuity in resources or policy or the state of the world will cause projected future effects to not happen. With policies that provide economic benefits, we must also increase the discount rate to reflect the fact that it will probably be more expensive to purchase lives, health, or utility as overall conditions improve.

This post is meant for economists, effective altruists, and others who are doing cost-benefit calculations over a long time horizon to decide what actions we should be taking now. I am trying to be as plain-language and non-technical as possible, which will involve simplifications of well-studied things. I do assume basic knowledge of cost-benefit analysis and time value of money calculations. Greaves 2017 is an excellent survey article for some of the relevant technical aspects, and this post builds on its foundation. 

Past use of Discount Rates
In the past, cost-benefit analysis was mainly used for investment projects, and was mainly concerned with money flows. Discount rates were based on the agent's opportunity cost of money, which was usually based on a market rate. There were no hard questions or philosophical considerations beyond "Don't waste money." Market rates were never close enough to zero, and time horizons were never long enough, for the considerations in this post to be relevant.

However, in recent years, people have started to use cost-benefit analysis as a decision analysis tool for many more kinds of policies, including those that do not use money at all but involve trade-offs between lives or health or utility today and in the future. This has forced a rethink of how to apply discount rates.

Pure Time Preference
In experiments, tests, and surveys, people always prefer things now to things later even when there is no possibility of trade and investment. This has been called 'pure time preference'. There are many good arguments that society or a government should have a zero rate of pure time preference i.e. it should care just as much about a life today as a life in 100 years.

I don't believe that there's really any such thing as a pure rate of time preference. I think that people discount the future based on uncertainty that the expected benefits will ever actually arrive. Of course, usually they may do this subconsciously, out of evolved heuristics. The monkeys who assumed things would reliably be around in the future died off, and the monkeys that grabbed what they could at each moment reproduced. We observe throughout history that whenever the economic and political situation becomes more stable, people's discount rate goes down and they start caring more about the future. Attempts at measuring time preference typically fail to adequately correct for these uncertainties.

I suspect, but am too lazy to prove, that hyperbolic discounting is actually based on a subconscious assumption of some fixed probability that the person is lying to you about ever delivering future benefits, plus a variable probability over time that they will be unable to even if they intend to. Hyperbolic discounting is entirely rational in any situation other than our weird world with giant reliable financial systems and strong enforcement of contracts, and people's instincts haven't caught up. (If I am right, an experiment that deliberately made the promise more trustworthy would see a lower discount rate, and I would love to see someone try to falsify this, even if I am proven wrong.)

Real Interest Rates
Everything will be discussed in terms of real values, not nominal ones, because tracking inflation is an unnecessary complication.

It makes sense to discount the future based on how much our money would grow if we just put it in the bank. There's no sense implementing a policy that has a 2% return if it costs us 3% to borrow money. For almost all of human history, governments paid a positive real interest rate to borrow. Now they don't, and it may be decades or more before they do again. Similarly, no individual can or will earn a positive real risk-free return on their money for the foreseeable future. A full discussion of this is beyond the scope of this post, but for our purposes it lets us zero out a complication. I'll make the simplifying assumption that governments will keep borrowing more until their borrowing costs increase and they pay zero real interest rates on all their debt, so we can just assume no interest.

Increased Cost of Utility
We don't care about money; we care about what money buys us. This could be lives, or health, or utility. The exchange rate between money and these things changes over time. When society is poor, or unequal, a little money goes a long way. As things get better, it is more expensive to do good.

Discussions of discounting often involve the Ramsey Rule, which is based on increases in wealth and the decreasing marginal utility of consumption, but I think that is both too technically intimidating, and also too simplistic because it just uses averages and does not take inequality into account.  

If people are using money at all intelligently to purchase utility, they'll be using it to increase the utility of the people who are worst off. This means that if inequality goes down in the future, or if the social safety net improves, the cost of purchasing utility will go up, probably significantly more than the overall economic growth rate. 

Our discount rate for a policy that has economic effects will be, at minimum, the expected annual rate of increase in the cost of purchasing utility. Ideally this would be estimated by forming a historical data set on the cost per QALY achieved by the 10% most effective charities, or by tracking the percentage of the population in various states of extreme distress or subject to easily preventable deaths, but as a kludge we can just double the real per capita rate of economic growth.

The primary message of this post is that we must discount based on the fact that everything we do can only affect the future with a probability rather than a certainty. There are many things that could cause our actions to have no effect, and each of them causes us to increase the discount rate. I call these discontinuities: anything that causes a break between what we hoped to happen and what actually happened.

Any competent cost-benefit analysis takes into account known unknowns, listing various scenarios and averaging them for an expected value. But I've never seen one that explicitly accounts for the annual tail risks of unknown unknowns, the discontinuities that could scramble plans and predictions by making big changes in the world. These things can usually be safely ignored on a short time horizon, but over long periods they matter a lot more.

I will generally be speaking in terms of an annual discount rate, but as a simple example, imagine that there are three different things that each cause a 10% cumulative chance, between now and the time period we want to affect, of the plan failing. The expected value of the benefits is roughly 70% of what they would be if the plan was guaranteed to do exactly what we projected. (Technically the expected payoff is 0.9^3 ≈ 73% chance of success, but that assumes risk neutrality. A more realistic assumption of risk aversion decreases the expected utility, so I'll cancel them out and just add up failure probabilities for simplicity.) So if the plan gains us 2 utility on a success, our expected value is 1.4 utility.

However, we still spent the resources, meaning that we lost the ability to do anything with them today. This means that, when doing a cost-benefit analysis, we should discount all future benefits by 30%. If this plan gives us at least 30% more benefits than spending money today, we should do it, and if it gives us less, we should not.

In this example, the failure probabilities were fixed, so it is no different than a normal expected-value scenario analysis. But in reality, the chances of these things increase as more time passes. Every year, there is a chance of something scrambling our plans, either directly or by scrambling the entire world.

To gain an intuition for this, imagine that you are a university professor, with a big endowment and research budget, in a random developed country in the year 1920, and that you are implementing a plan to make life better for people in the year 2020. What is the probability that your plan will survive? (Keep in mind that there's a good chance you are Jewish and live in a country in continental Europe.) Then imagine that you live in 1820, or 1720. Consider all of the economic, political, moral, and scientific shifts that have happened in those years. What is the probability that anything you do can even persist at all in a form you intended? Even then, what is the probability that your plan actually made the world better, in a way that the people of the future agree with and appreciate?

So, our annual discount rate must always include the sum total of all the annual probabilities of everything that could possibly happen to make the plan fail or become irrelevant.

As a concrete example, consider a climate policy that costs a billion dollars a year but would, if consistently implemented for a hundred years, prevent a trillion dollars worth of damages in year 100 of the policy. (This is simplistic, but many proposed policies have a similar payoff structure.) The break-even discount rate is between 3 and 4 percent: with a discount rate of 3%, it's a good idea, but with a discount rate of 4%, it's a bad idea.

When determining our discount rate to use, we must add up:
1) the probability that something else will solve climate change in the meantime, perhaps a cheap geoengineering solution. If that happens, we wasted the resources.   
2) the probability that a future administration will overturn the policy. If they do so, the future benefits will never materialize. 
3) the probability that the country ceases to exist (or becomes conquered by a different power that changes the policy). If that happens, we don't get the benefits.
4) the probability of any form of existential risk that would not be affected by the proposed policy.
5) the probability of any form of catastrophic risk that reduces the population enough to make the benefits irrelevant.

All of those estimates should be based on an outside view whenever possible. For example, for #3, look at how many nations in the last 200 years have collapsed, or undergone a revolution, or been conquered, and how often they tend to last before that happens. Given that data set, I'd guess there is a roughly 50% chance of the country ceasing to exist in its current form over the next 100 years, so the discount rate, for just point 3, should be about 0.5%. 

Estimates of x-risk add 0.15% for point 4, and probably another 0.15% for point 5. Point 2 probably adds another 1-2% to the discount rate, as does point 1 (I am a tech optimist). So the point estimate is a 3.8% discount rate.

Of course, this assumes that discontinuities result in zero benefit. If they still allow some benefit, the math gets more complicated, but we can approximate that by adjusting the discount rate downward for the policies where a discontinuity still gives some benefit. If there is a 0.5% chance of government change, but we assume a 50% chance that a revolution or invasion does not change the policy, then the adjusted discount rate is 0.25%.

A less controversial policy, with less chance of being made irrelevant by other means, might have a discontinuity-based discount rate as low as 1%, but I think this is as low as we can safely go. Chaos happens, and we are entering an uncertain age.

Because people don't deal well with very small probabilities, it may be best, when doing the estimation (which will require surveys of experts) to focus on 10- or 25- or even 100-year blocks of time, and then divide.

Then, if the benefit is an economic effect rather than a pure impact on health or lives or utility, we also have to add the expected annual increase in the price of utility, conditional on no discontinuity. (Technically the total discount rate we should use is (1+u)/(1-d) - 1, where u is the growth rate in utility cost and d is the annual chance of a discontinuity. This will always be slightly higher than d + u, which will matter in the long run, but since these things are estimated with high uncertainty the difference is well within the confidence interval so we might as well keep things simple.)

To make an analogy to finance and investing, the discontinuity probability is the risk premium, and the utility price increase is the risk-free rate. We should only invest if our return is higher than both of these put together. In finance, the risk premium of the safest investment is assumed to be zero, and everything else is compared to that, but in this case, there is nothing with a zero risk premium, because we are comparing a certain result today with an uncertain result in the future. Anything we do has risk.

As a second example, consider a private action, perhaps setting up a think tank to research ways to stop a catastrophe, and that nobody else is trying to solve the problem or ever will. The discount rate is based on:
1) the probability that the organization suffers value drift.
2) the probability that the organization is taken over by a malevolent or selfish actor.
3) the probability that an outside force seizes all of its resources.
4) the probability that the government, society, and/or economy collapse for some other reason, making it impossible to continue work, or that another catastrophe makes the work pointless.
All of these risks increase with time, so the further out the benefit is, the less likely it is that the money spent will accomplish anything.

Even in the extreme case where the plan, once set in motion, cannot be disrupted, for example a rocket launch to nudge the trajectory of an asteroid that will hit the Earth in 200 years, we should discount. There are still discontinuities. Every year in the intervening time, there is a probability that people will develop a cheap asteroid deflection system, or that the asteroid will be consumed by mining, or that some other x-risk will make our effort irrelevant.

It should be clear by now that this is not an estimation problem for economic data sets and equations with greek letters. This is something we must think of by looking at historical trends in the appropriate reference class. Look at what has happened to similar plans or organizations in the past, and estimate the annual probability of a discontinuity of policies or resources between now and the time period we wish to affect. 

Scientific and Moral Uncertainty  
I'm less confident about this, and it doesn't affect the main result, but I suspect that epistemic uncertainty also increases our expected discount rate. Learning that the world doesn't work the way we thought it does is another kind of discontinuity. The annual probability that our theories are wrong, and that our policy is therefore mistaken, should be added to the discount rate.

Given how much our science and understanding of the world has changed and improved in the last hundred years, it seems reasonable to increase the discount rate by another 1% to account for epistemic humility and the potential realization that what we think we know is wrong. 

I think the annual discount makes sense, because long chains of causality introduce more chaos and more opportunities to be wrong. It's kind of like weather forecasting: we can be pretty confident in our ability to predict things a few days out, but after 10 days predictions are basically just swallowed up in chaos. 

Currently, with economic forecasting of government policies, it is rare to go out past 10 years, because of this kind of uncertainty. Assuming no large-scale change in economic conditions, we can predict the short-term effects of a tax or transfer policy fairly well. The Congressional Budget Office can know with decent precision what the effects of a tax cut will be over the next few years, and how that will affect labor supply and similar things. But there are a lot of second order effects and long-term adjustments that they don't know how to deal with, so as we project the changes of a policy more years out into the future we become less and less certain of anything. Typically they do this by increasing the error bars, and just stopping when they get too big.

But if we were doing our best to predict the effect of a large-scale change, because if we don't then people will treat all the side effects as though they did not exist, and while we are uncertain we know they are not zero, it might be more responsible to run the model the best we can, include its probabilistic output in the expected value calculation, and then add in an extra 1% or 2% discount rate on all future effects to account for the extra chaotic uncertainty.

None of this changes the fact that the value of preventing human extinction is huge, at least three orders of magnitude more than the entire annual world GDP. And that is just counting the VSL-based value that currently-existing humans place on preventing their own death; it does not include the value that future people place on existing.

In fact, this discounting strengthens the case for making serious sacrifices to stop x-risks. Anything else we might do to affect the far future is rendered irrelevant by the discounting, including the harmful effects of significantly slowing down economic growth. This is because there will be great unpredictability in human affairs as long as they still exist.

It can be argued that, ironically, the ethical implication of all this consequentialist math is to push people into a precautionary-principle-based deontology where we focus all of our actions on stopping whatever is likely to cause existential or catastrophic risk in the short term.  

However, if we cannot do anything about an x-risk, then we should be doing things to help people now rather than in the far future. The certainty of actually helping now is better than a tiny probability of a plan surviving and doing what we think it will, and this is reflected in the discounting.

Friday, January 22, 2021

Radical Tax Reform

I know that this is probably-impractical utopian thinking, but whenever I discuss related ideas in conversation, people ask me to write it up, so here goes:

The core idea is to get rid of all current taxes, and some existing government power and regulation, and replace them with externality taxes and progressive consumption taxes.

Externality Taxes: The Basics
All significant, measurable externalities should be taxed. Whenever possible, replace existing command-and-control regulations with a measure-and-tax system. Instead of people paying real economic costs for regulatory compliance, the money goes straight into the treasury as a transfer. This increases economic efficiency and will end up with less overall compliance costs for many businesses.

I've estimated that just the basic and obvious externality taxes will raise one to two trillion dollars a year in the USA. This gives us between a quarter and a half of the government's budget. In addition, there are some other externalities to tax:

Corporate Power is Taxed with a Progressive Corporate Income Tax
When considering only economic efficiency, i.e. avoiding the distortionary unfairness of double taxation, the ideal corporate income tax is zero. However, corporate size and market concentration are plausibly negative externalities that should be taxed.

All small businesses in competitive markets will pay no tax. Large companies pay exponentially increasing tax rates based on size and market concentration. If this is calibrated so that this is roughly revenue-neutral with the existing corporate income tax system, it would raise a quarter trillion dollars annually. 

I have no idea if a quarter trillion is an appropriate estimate of the annual harm done to our politics and society by concentrations of corporate power. I doubt anyone has estimated this in any sensible way. Probably the easiest way to do it would be a QALY survey. Fairly or not, many people hate big business, and would prefer to live in a world where it did not exist. Suppose that, on average, living in a nation with our amount of corporate power costs everyone one-thousandth of a QALY i.e. 1/20th as much discomfort as mild lower back pain. A third of a billion people, times a half a million dollars per QALY, times 1/1000, equals a sixth of a trillion dollars, which is pretty close to the current tax.

Other Externalities
Basically any problem you have with our society can be thought of as an externality to be taxed. The tax system, rather than the criminal justice system or any other power of government, should be the first tool to reach for to try to fix things. 

Advertising to children can be taxed. Advertising to anyone can be taxed. Political lobbying and large donations can be taxed. Actions that break up or damage families can be taxed. Conspicuous displays of wealth can be taxed. Harming the cultural foundations of our civilization can be taxed. Companies failing to hire in an equitable way can be taxed. Social media hate mobs can be taxed. 

I don't know if it would be best to have each separate tax require an act of Congress, or to give the executive branch broad enabling powers to identify externalities. Probably it would be safest for each one to require separate, narrow, enabling legislation. Congresspeople would propose tax-enabling legislation in response to constituent pressure. The legislation should include a rough estimate of the damage caused by the externality, and exactly what actions will be taxed and at what rate. Then we can have a productive discussion revolving mostly around checking the math and thinking about practicalities and side effects rather than shouting at each other. Once the tax is in place, people can either pay it or change their behavior.

Progressive Consumption Tax

Even if we identified all harms that could be profitably taxed without violating rights or causing too many bad side effects, it probably wouldn't be enough money to fund the government. We need another source of money that is as non-distortionary and politically sustainable as possible.

Whenever you tax something, you destroy it. Income taxes discourage productive work and investment. Wealth taxes, if only placed on wealth that the state can see, generate an economy-destroying incentive to hide money rather than invest it. Consumption taxes, on the other hand, have far fewer problems, and possibly some benefits, like discouraging extravagant lifestyles that fuel envy and social division.

In order to allow progressivity, and to avoid double-taxation of retired people and people who have saved, and to more easily transition from the current system, we implement it as an alteration of the income tax system. You would get taxed based on the money you withdraw from the bank rather than the money that goes from your employer into your bank account. Bank accounts that existed at the time of the tax changeover are exempt, under the assumption that you already paid income tax on them, and all of your income from that date forward is deposited in a new account.

Instead of your employer withholding money from your salary and submitting tax information to the government, your bank does, at the time you withdraw or spend it. (Assuming direct deposit. If you have an actual check you cash, it gets reported by the employer as normal and you are assumed to consume all of it, but you get a tax refund for depositing money in an investment account) All brokerage accounts etc. are considered 'in the bank' and untaxed. Charitable contributions are untaxed. Aside from that, the current tax brackets would be roughly the same, except we add lots of increased tax brackets for people who spend more money. Anyone who lives a lavish millionaire lifestyle pays 70%+ marginal tax rates on that spending, and seven-figure and higher lifestyles can be taxed at 80%+ At the top end, you'd lose almost 90% from each withdrawal, so you pay the government almost ten times as much for your consumption as it costs.

Some people will complain that this is unfair, because rich people can 'evade' almost all of their taxes by living a frugal middle-class lifestyle while their money sits around growing. But this is exactly the kind of behavior we want to encourage; it would increase investment and economic growth. We should think of the consumption tax as a kind of externality tax: Purchasing things takes them away from others, and plausibly harms them by decreasing their perception of relative status, but earning and storing money doesn't hurt anyone (assuming all the harms involved in what you did to earn the money are taxed properly as externalities).

Tax Liquid Wealth that Produces Power? 
We are probably already taxing lobbying and political donations as an externality, and of course they also get taxed as consumption. But potential power is still power, and it may be possible for large concentrations of wealth to exert improper influence, without paying any tax, by merely threatening to spend a lot on lobbying. If this turns out to be a problem, we can have a wealth tax on large pools of liquid assets. Make sure people can escape this by locking their money up in long-term investment projects, or by signing some kind of binding pledge to never spend money on lobbying or influence.

That is basically it. We've switched away from taxes that give bad incentives and toward taxes that give good incentives. In most cases, 'tax evasion' is stopping doing things that harm people. The next two points are details on stopping nonproductive tax evasion under this new system:

Include all Perks and Benefits as Income
In order to prevent powerful executives from evading taxes by disguising consumption as business expenses, we can count absolutely everything they get from their employer as income. The cost of all benefits is taxable income. In addition, something like exclusive use of a company car is income equal to the rental cost of the car. Use of an office counts as income equal to its rental cost.

All business travel is income equal to what it would cost an outsider to participate, e.g. a trip to Davos counts as income equal to the airfare plus the hotel cost plus all expenses plus whatever it would cost an ordinary person to bribe or buy their way in. If the event is so exclusive that there is no observable market for its tickets, this cost is assumed to be in the tens of millions.

In order to enforce this in an incentive-compatible way, the government collects taxes from companies on everything they spend on employees at a rate a little higher than the highest possible marginal tax rate of the recipient. (If a company has any employee who gets over a hundred of million dollars including benefits, it pays a tax on any benefit they get equal to ten times the total cost of the benefit.) Then individuals declare the value of all benefits and perks they received, in order to get a tax refund on them. 

For example, if consuming all your income (after including the benefits) would put you in the 50% tax bracket, and you got $100k worth of perks and benefits, then the company paid a bit more than $100k on your benefits (at a 50% consumption tax, you need to withdraw $200k to purchase $100k of consumption). If your actual consumption only takes you to the 40% bracket, you'd get a tax refund of a bit more than $33k. (with 40% consumption tax, you need to withdraw $166k to end up with $100k, so the proper tax would have been $66k.)

The more benefits you report, the higher your refund is, so you have an incentive to report everything you possibly can, even at the highest bracket, because the company was forced to pay a bit more. If employees collectively report more in benefits than the company reported and paid taxes on, the company pays the difference plus an additional fine.

This would result in total compensation becoming more transparent, and taxed properly, and would probably quickly settle down into everyone being paid in mostly cash, which would end a lot of current economic inefficiencies like health insurance being tied to employment.

Include Imputed Consumption Value of Non-investment Assets  
In order to discourage certain kinds of tax evasion, and to properly measure lifestyles, we also need to count as income the consumption value that you get from owning things like yachts and mansions. Most or all of this would apply only to people in the new lavish-lifestyle tax brackets; it isn't sensible or desirable to try to collect these taxes from people in lower brackets.

The imputed consumption from any non-investment asset you own is added to your income. For most things, this can be the rental market value. If you own a home, the government uses real-estate data to look up how much it would cost to rent a similar home, and increases your income by that amount. Same for cars, yachts, vacation property, etcetera. 

In cases where there are no rental markets, we do our best to estimate its auction value and assume that the consumption value is a fixed percentage of that. For example, someone hanging a Picasso on their wall, or who owns a million dollars' worth of jewelry, would pay a percentage of its value each year.

In order to avoid double-taxation, you subtract from the property value the amount that you paid for it. So if you paid $300k for your house, your imputed income is reduced by the rental value of a $300k house. If your house still has that value, you pay no additional taxes. If you bought it before the changeover, it is assumed that you paid appropriate taxes, and if you bought it after the changeover, you already got taxed on the money that left your bank. But if the house value has gone up since you bought it, you pay based on the difference.

This system taxes speculative profits rather than productive investment. Purchasing something that later rises in value is taxed, in a way that buying stock in a productive enterprise is not. The current system gets things exactly backwards, giving people strong incentives for unproductive real estate and art speculation.

Monday, October 12, 2020

Worthless Replicators

Futurists often think that self-replicating machines will change everything. But It's possible to imagine a world where self-replicating technological systems are common, but nobody bothers to use them because they are economically and militarily worthless. In a world with relatively strong property rights and relatively low transportation costs, this could be the expected state of affairs.

The key insight here is that a civilization (meaning a combination of machines, ideas, and people) already is a self-replicating technological system, and every part of that system will always be able to operate more efficiently, through specialization and economies of scale, than anything at the same tech level that must also be small, portable, self-contained, and/or limited to local resources.

It's pretty much a guarantee that no matter what level of technology you have, the output of a portable fabber system is always going to be worse than the output of a factory. Mass production of a single thing by a large facility designed to do that thing will always be more efficient than production from a device that must be small and capable of flexible output. This is true no matter what tech level you have; any tech that makes a fabber work better can also make a factory work better.

For example, in our world, it's easy to see that the output of a printing company is always going to be better, and cheaper per page before delivery costs, then something that you print out of your home printer. Sure, the output of your home printer might be better and cheaper than the output of a printing company in the 1950's. But someone in the 1950's who correctly predicted the capacity of a home printer, and then predicted that nobody would ever buy a printed book or magazine because the home printer could print things on demand more cheaply, would be mistaken. The factory tech also advanced.

It is certainly possible to have a world where all of the components of a replicator can easily come out of a factory, but the output of that replicator (be it useful things or other replicators) would be much more expensive, in terms of resources and time, than the factory version.

It should also be easy to see that any weapon or military system with the design constraint "Must be capable of self-replicating" will be far less cost-efficient than any weapon of a similar tech level that does not have that design constraint. A military power that builds a lot of self-replicating weapons will lose to a power that spends similar amounts of resources on building non-replicating weapons.

People seem to think that resources no longer matter once you have self-replication. But this is not true at all. Any resource that can be consumed by a replicator can also be fed into a factory, and the factory will do a much better job of converting that resource into wealth or power or whatever else you care about.

People also assume that self-replicators will dominate everything because of their capacity for exponential growth. But that implicitly assumes that the adversary is doing nothing with that time. All well-functioning economies grow exponentially whenever they have access to resources, and they do so at a rate based on their productive efficiency. Sure, if you drop a self-replicating weapon in the arctic tundra, then in 100 years you might have billions of drone weapons. But an adversary civilization that spent 100 years converting a similar mass of arctic tundra into weapons using factories will have far more of them (and their weapons will all be 100 years more technologically advanced than yours, and your drone control system is 100 years old and probably easily hacked by modern tech).

On both long and short time scales, factory output dominates. In the short run, they can pump out thousands of things while you are on the first few generations of replication. And in the long run, their increased efficiency and technological advancement will allow them to exponentially increase their productive capacity at a faster rate than your replicators. Your replicators may be doubling every month, but their economy is doubling output every week. You are building things to a more restrictive design constraint, while they can use that tech to crank out specialized factories that efficiently build weapons and factory components.

Of course, all this is before travel costs. If you want to turn an asteroid into a colony, moving the factory to the asteroid is going to be much cheaper than moving the asteroid to Earth and then shooting the factory output back into space. But even in this situation, there is little economic incentive for self-replication. A machine (or system of machines and people) that turns an asteroid into a colony will always have much better performance if it does not also need to be able to create a new colony-creation machine in the process.

So even in the case of a spacefaring civilization with access to self-replication technology, it seems like there's very little economic or military incentive to drop a replicator in an asteroid belt and have it convert the belt into something. No matter what you're trying to accomplish, deploying something built in a large modern efficient factory would always be able to do it better, faster, and cheaper.

The only real motivation for using completely self-replicating systems is independence. If you want to be liberated from the large efficient systems that enable modern civilization, then a self-replicating system loaded with all the important knowledge of civilization can do it for you. But your quality of life, per resources spent, will be far lower than any citizen of that civilization. It will also be significantly lower than anyone who wants to live off the grid but does not insist that their off-grid house and fabber be capable of self-replicating.

Some people will choose to do this, but they will only exist as long as a large civilization actively protects them. They will be utterly incapable of resisting any kind of military aggression, and they are sitting on useful resources. Anything that is capable of sustaining them is also capable of generating wealth for the larger more efficient civilization. So they will never have any real independence in a geopolitical sense, and will always be dependent on a sovereign power for protection.

It will be a performative delusion of independence, probably nestled safely in the territory of a major power while paying property taxes. And this is only possible if the property taxes are a tiny fraction of the economic worth of the resources; if taxes are higher and based on market rates, then the controller of the replicator will never be able to afford them, because the large-scale systems are capable of generating much more wealth from those resources than the smaller systems whose efficiency is constrained by a need to self-replicate.

So in whatever sci-fi setting you like, be it plausible hard sci-fi or something like the worlds of Star Trek or Star Wars, self-replicating systems could be kind of like bitcoin. There's a bunch of random tech geeks that swear it's the next thing, but nobody's ever actually able to find a practical case for it. Occasionally their community makes a useful tech breakthrough, which is immediately absorbed into existing systems and used to make them better.

Some rando could think they're wealthy or powerful because they control a Jupiter-sized mass of mobile self-replicating probes, but traders don't buy anything from them because other people are selling it more cheaply, and any military power's reaction is "Whatever. I could destroy that in an instant but I just don't feel like it right now"

Wednesday, May 13, 2020

Estimate of Potential Externality-tax Revenues in the USA

A colleague asked me to estimate how much money the USA could raise by taxing externalities equal to the damage they cause. This is the kind of thing that should be done by a team of competent experts, but apparently nobody with resources has even bothered to ask the question, so I took a shot at it:

CBO's list of budget options estimates about $100B per year from a carbon tax set at the estimated social cost of carbon. In order to be sensible (i.e. non-distortionary and having a chance of passing), you would also need a 'carbon tariff' i.e. taxing all imports on their estimated carbon emissions. Ideally at a slightly higher rate for political reasons and to compensate for enforcement or estimation errors. Imports are 15% of GDP, so assume $120B from carbon tax and tariff.

After that it's all napkin math. We know the harms of various externalities, so we can calculate their social cost, but we don't know how much of that we could capture with a tax. 

Air pollution kills about 100k Americans per year. I'll assume that 50% to 90% of this would get engineered out in response to the tax, and the rest hangs around to be taxed. With 10k to 50k deaths remaining and taxed at the $10M VSL, you raise $100-500 billion.

The economic costs of alcohol are about $250 billion. This does not count the monetized costs of the 2 million lost QALYs. (I had to do the QALY calculation by multiplying their 'years lived with a disability' by the QALY cost of 0.3, assuming an average of mild and moderate QALY loss. At $500k a pop, the monetized QALY cost of alcohol is about $1 trillion a year ($1.3 trillion total cost). With that level of social harm, the question then becomes how much money you can possibly extract from the alcohol market without the market going to organized crime. Annual alcohol sales are a quarter trillion, and maybe you could double or triple the prices before drinkers go to drug gangs instead (or quit), so that is $250-$500 billion in possible taxes.

Bad diet causes a loss of about 10M DALYs a year in the USA (source, use the 'plot option'). Again, I assume that tax-induced behavior change and reformulation would eliminate 50-90% of the harm, leaving 1-5M DALYs lost, or $500-2500 billion in potential taxes. Then assume that a quarter to a half of this is eventually lost to gray-market evasion. (buying a hunk of pig with cash from a local farmer)

Plug all that into your favorite Monte Carlo software:
The 90% CI for taxes raised on unhealthy food $250-1000 billion. The 90% CI for all my health-harm estimates, added up, is $800-1700 billion. Add in the carbon tax, and another 100-200 billion from other kinds of externalities, and you get about one to two trillion per year. 

Probably the only way that all this could be politically acceptable is if it funded a UBI, basically making it revenue-neutral by giving all American adults their share of the tax. With 250 million adults, this gets you a $4-8k UBI per person.

Friday, January 10, 2020

Roaring 20s

I am confident about the upcoming decade. I think by the end of it, the current mess will be mostly over and we will be on track for a good sustainable semi-utopian future for humanity. I'll briefly list the reasons below. Even if you don't share my overall optimism, please at least consider these as good and interesting developments to watch out for.

Biotechnology: Things will start to arrive in a big way. By the end of the decade, we will see many new medical treatments, improved industrial processes, efficient carbon sinks, etc. Things will only just be getting started, relative to their potential, but a lot of good things will have already happened and people will understand that more good stuff is coming. 

Plant-based meat: Impossible Foods is shooting for 2022 as the year that high-quality plant-based burger patties become as cheap as the dead-cow version. I'll apply the 'Elon Musk Adjustment' to that goal and say it will probably happen between 2023 and 2027. Assuming they (or someone with similar tech) can scale up production, people will be able to conveniently replace most of their meat consumption, giving us a huge win for health, environment, and ethics.

Scientific Research: We have learned many lessons from the replication crisis. Data openness is now an established thing. By the end of the decade, most research will be freely available, with all source data online in ways that allow not just for replication and error-checking, but also massive meta-analyses that give us highly-powered studies to untangle a lot of hard questions.

Renewable Energy: With continued progress in solar and next-gen nuclear power, we should see grid parity in lots of places by the end of the decade. Between this and plant-based meat, our carbon footprint should start to get a lot smaller.

Space Travel. I am pretty confident that we will have a manned research outpost on Mars by the end of the decade, as well as the beginnings of a space-based economic infrastructure with asteroid mining etc. that will allow humanity to really start expanding beyond our home planet.

Politics: If you take a step back from the hype cycle of people trying to generate clicks by preying on your negative emotions, you will realize that the current mess shows just how resilient the system is. Most of the day-to-day details of governance, like food safety inspections, are ticking along in regular order despite the chaos and dysfunction that grabs the headlines. Just like with all the crap and violence in the 60s and 70s (which was much worse than the last two decades) by the end of the decade people should settle down and realize that we are all in this together and things are mostly okay.

Consumer Tech: Never underestimate the capacity of our civilization to create compelling consumer products. I won't make any predictions about specific things, but by the end of the decade, there will probably be at least one thing that generates value and convenience equivalent to the introduction of smartphones.

Logistics: Just as in the military, in the civilian economy the professionals talk logistics. By the end of the decade, much of the warehousing and delivery infrastructure will be extensively automated, making everything that can be delivered in a box cheaper and more convenient to purchase. This improved supply chain efficiency will have more of an immediate macroeconomic effect than anything (possibly everything) else on the list, significantly lowering consumer goods price inflation.

Basic Income: If you don't need the services of a credentialed professional, or space in a high-rent area, a good life is already incredibly cheap. Not counting rent and health insurance, I pay about $300 a month for a lifestyle that would seem shockingly rich and comfortable to anyone other than a modern professional. The basic essentials of life are only going to get cheaper in real terms. By the end of the decade, there will be a growing realization that wealthy countries can and should give their citizens a basic income of about $5,000 a year (i.e. spending about 10% of their GDP). Living only on this would require you to live in the boonies, and will only buy a 'minimal' standard of basic health care (i.e. the stuff that you could get 40 years ago, but of better quality), and you couldn't afford any expensive hobbies like drinking, but this would essentially end deprivation and exploitation.

Add all of this up, and we will be well on our way to a Star Trek kind of world, 100 years ahead of schedule (assuming we can skip the Eugenics Wars).

Wednesday, August 28, 2019


Art is a subset of the class of objects that are created by people and have an effect on the mind of other people. It may or may not have many other definitions and features, as people have argued over for millennia, but this is a true thing you can say about it. There are many mind-affecting things not created by people, but they are not art. The selection, curation, or framing of things not created by people might be art, but this is done by people. Furthermore, I believe everyone can agree that something that has no effect on the mind of any observer is not art.

So, no matter what other judgments you can make about art, you can always make judgments about it that you could apply to any other object in the class of things that are created by people and have an effect on the mind of other people.

The most relevant judgment is what effect it has on the mind of people. What are some general features of that effect, and is it a positive or negative effect?

One effect that it can have is to give people pleasure. Not all things that give people pleasure are art, and not all art gives people pleasure, but to the extent that art gives people pleasure, it can be judged by the standards of such pleasure giving objects. In many cases art, or a reproduction of it, gives a great deal of pleasure for a relatively low cost, so in that respect it is quite good.

Another effect of art is that it can teach people things. Not all things that teach people are art, and not all art teaches things, but to the extent that art teaches people things, it can be judged by the standards of things that teach. People can be taught things that are true or false, good or evil. It is legitimate for society to encourage the production of things that teach truth or good, and discourage the production of things that teach falsehood or evil. When dealing with art in this way, it is legitimate to focus on what is being taught and ignoring other aspects of it.

A third function of art, perhaps its main function, is to define and reinforce social boundaries between different groups of people. Not all things that define social group boundaries are art, and not all art serves to define a social group boundary, but to the extent that art is being used as a social group boundary, it can be judged in that manner. Because cohesive social groups are valuable to people, anything that helps create such cohesive social groups adds value to society.

However, there are better and worse ways to create and define social groups. The best way is for everyone to admit that their preferences are arbitrary, and their subculture is simply a collection of people united in enjoyment of an arbitrary thing that is neither better nor worse than other people's taste or fashion.

Things become more costly, in the sense of generating negative emotions and status anxiety, when people insist that one social group or subculture should be accorded higher status than another. In general, almost all arguments over what should be defined as art are actually arguments for one particular group of creators or consumers to be labeled as higher status than another group. The word 'art' denotes high status, so people spend a great deal of effort inventing impressive arguments for why the thing associated with their group is art, but the thing associated with somebody else's group is not art.

Things can become very bad when a subculture unites around a definition of art that is either deliberately ugly (meaning that is designed to not give pleasure to the typical person who views it), or teaches things that are false or evil. This can happen quite often because of countersignaling. In such a case, the broader society has a right to tell the subculture that they are simply wrong, and should be considered low status.

A very important confounding factor when discussing art is that people instinctively enjoy and appreciate a thing more when it is especially difficult to create. The difficulty of creation serves as a costly signal, and whenever a thing serves as a costly signal of quality, because it requires expensive ingredients or a great deal of training to produce, people will instinctively consider it high-status and wish to own it and affiliate their group with it.

However, this difficulty of production should always be counted as a cost rather than a benefit of art. It would be better for society if people got the benefits of art out of something that was much cheaper to produce, and the expense of production is very often nothing more than a costly and wasteful signalling game. People try to attract attention to their thing by paying for really high production values, but that simply makes everybody else's thing look relatively worse, without any net social benefit.

Art may have many other features, and it may have many other effects on human minds. However, most or all these features are either arbitrary, or impossible to measure with current technology and philosophical understanding. The main things that we can measure are how much people enjoy it, what it is teaching people, and what kind of social groups form around it.

Therefore, the ideal of art is something that has beneficial effects in all three of these categories. Good art produces pleasure, it teaches things that are true or good, and it unites people around common affiliation with it, ideally in a way that causes good group dynamics and does not cause the group to think less of outsiders. As with most other good things produced by people, we should find ways to produce it as cheaply as possible and in a great variety, and we should encourage people to enjoy whatever they like without feeling a social need to copy the behaviors of other people.

Wednesday, April 24, 2019

Bullshit Jobs are Organizational Signalling

Sometimes people have a good intuitive understanding that something is very wrong with the world, but they lack the analytical ability to explain why the thing is happening. An important skill, related to intellectual charity, is to learn to listen to these visionary heartfelt cries of pain even when the professed explanation or ideology of the author is clearly very confused.

This is the case with Bullshit Jobs. The author, an anthropologist, correctly identifies and gives voice to a very real and serious problem in our world: the fact that many salaried professionals spend their lives doing things that contribute nothing to social welfare. However, he is unable to understand exactly why this happens, and his various explanations range from laughably wrong (a vast evil conspiracy of rich right-wing people who are manipulating the moral fabric of society) to almost right (employees serve the needs of individual managers rather than the organization).

The correct answer is that organizations (and departments within organizations) face the same pressures to spend resources on costly signals and arms races that biological organisms do. Bullshit jobs emerge from the same game-theory dynamics that generate peacock tails, elk antlers, and other wasteful features in animals. Just like a peacock without a tail will not pass on his genes, a company without a marketing department will not sell its products. Just like an elk without antlers will not pass on his genes, a company without a legal department will not be able to keep its resources.

Sometimes the leaders of organizations know what is going on; they would prefer to not have the expense of a marketing or legal department but know that it is necessary for self-defense. However, organizations also face the same incentives for self-deception that Hanson discusses in Elephant in the Brain. People dislike admitting that they are doing things for selfish reasons, so they invent justifications.

It is rational for the organization to hire people to generate these costly signals, because they need to attract attention and resources, but these signals are only a way to take resources form someone else and do nothing to actually make the world a better place. The result is even more pointless and destructive than the the zero-sum games that individual humans play, because millions of human lives are being utterly wasted in the service of Moloch.

This is a very hard problem to solve. People often assume that the problem is due to capitalism and can be solved with central planning, but this is exactly wrong, because departments within government have even stronger incentives to generate and sustain bullshit jobs than market-disciplined organizations.

Ideally there would be some mechanism to identify and heavily tax bullshit jobs as the horrible externality they are. But this is very hard, because from the outside, it is quite difficult to tell the difference between valuable coordination activities and zero-sum signalling. Many managers and planners are extremely necessary and things would cease to function without them. Of course, when you are necessary like this, you have the ability and incentive to extract more resources, so most manager jobs end up as some combination of valuable coordination and value-destroying rent-seeking.

Still, at minimum, a competent society would enact massive taxes on marketing and legal services.