CSPI Essay Contest Submission

I applied to CSPI’s policy reform for progress essay contest. For the proposal, I rewrote my land value tax plan, which was a little off topic since the contest was focused on policy reforms to accelerate technological development. Here is my submission:

CSPI Essay Contest: An Approach to Land Value Taxation, By Sam Harsimony

Note: Though this policy reform does not directly address technological development, I believe that land value taxes can promote CSPI’s goal of accelerating progress by displacing inefficient taxes, reducing speculative bubbles, and encouraging denser development.


Land value taxes (LVT’s) are an attractive approach to public finance. Their proponents argue that landowners should be taxed based on the rent their land would generate if it was undeveloped (i.e. the rent generated by the empty lot) since this value is generated primarily by nearby economic activity and not by the owner.

LVT’s are one of the few taxes broadly believed to increase efficiency rather than reduce it. On top of this, LVT’s can fund a large fraction of government spending, replacing less efficient taxes. Because of their nice properties, LVT’s are the only tax that I know of with an associated ideology.

So, why haven’t we implemented them? There are two problems. First, for LVT’s to be efficient, it is critical that they only tax the value of the empty land. Unfortunately, it can be hard to determine how much a piece of land is worth without its amendments. How much should you pay for a property when you subtract the home? Second, current homeowners overwhelmingly oppose higher property taxes and have significant political clout. 

Here, I propose two complementary reforms that can tackle these issues. First, I will suggest a solution to the land valuation problem using public auctions. Second, I will discuss how LVT’s can be used to lower homeowners tax bills, providing a path to popular support.

Land Valuation

Let’s examine the problem of obtaining accurate land valuations. 

Part of the difficulty with LVT’s is that the property market often bundles the sale of land ownership with the sale of amendments to the land (e.g. buildings). This makes it difficult to determine the value of the empty lot. Many solutions to this valuation problem require sophisticated mechanisms or large government bodies in order to administer the tax. These solutions are unsatisfying due to their complexity and political infeasibility. Instead, there is a simple way to obtain land valuations that doesn’t require a complicated scheme. 

Namely, the law can require that all land titles be sold in a public auction, separate from the sale of the home and other amendments to the land. Once the land itself is sold, the homeowner is free to sell their property to the new landowner or remove these amendments if the parties cannot agree to a fair price.

Legally, this would take the form of the following amendment to a state’s property law:

All land ownership in the state may only be sold or transferred at a public auction. The ownership of the land must be transferred to the winning bidder and the value of the winning bid must be reported to the Internal Revenue Service.

By requiring that land be auctioned, the public obtains a reasonable assessment of the land value. Competition between buyers makes it hard to underbid in order to avoid taxes, while public knowledge of nearby land values makes it unlikely that the bids will be too high.  

Even without taxing sale prices, the public auction of land rights allocates land to the people who will use it best. Public land valuations also provide citizens and governments with an accurate indicator of local economic health, which is crucial to decision making. 

Most importantly, access to land valuations enables states to implement a land value tax, which can provide an efficient source of revenue while eliminating rent-seeking.

Political Feasibility

Though this simple reform can tackle the problem of obtaining land valuations, to have real world impact, it must be politically viable in order to have a chance of being implemented. For LVT’s to be politically feasible, they must be beneficial to existing homeowners and make minimal changes to the law.

Rather than upend a state’s property tax law, local governments can lower homeowners tax bills and increase efficiency simply by substituting land taxes for property taxes. Specifically, I propose that states retain their existing property tax system, while replacing all instances of property values with land values within their tax laws. Since land values are always lower than property values, this reform would significantly lower homeowners property taxes. 

Legally, this would take the form of the following amendment to a state’s property tax law:

All references to “property value” used for tax purposes are understood to refer solely to the sale price of land rights bought at public auction and must exclude the value of any property or other amendments to the land.

Because of the lower tax bill and improved efficiency, I believe that this policy can garner broad political support. Once in place, states can exchange large reductions in other taxes for small increases in their land value tax rate, keeping revenue constant while increasing efficiency. 


By requiring the public sale of land rights and taxing these sales in a politically viable way, land value taxes can provide a socially beneficial source of revenue for governments. By displacing inefficient taxes, LVT’s can boost economic growth and allocate land to those who would use it to create value.

I am not the first to propose solutions to this problem; I imagine that practical implementation of any land value tax will involve hard tradeoffs between political feasibility, economic efficiency, and legal complexity. But the first step towards finding new solutions is a simple, popular reform that can shake up our stagnant approach to distributing land.

Links #23

I recently worked with Manifold Markets to make Charity Prediction Markets a reality; now users can send their winnings to charity! Check out their launch post here.

Some cool companies: Cache DNA (DNA-based data storage) and Spin Launch (crazy)

Bryan Caplan on why the Tiebout model doesn’t work in real life. It seems the big issues are that local leaders salaries are unrelated to their performance and that localities can extract land rents.

Big List of Cause Candidates: January 2021–March 2022 update

Friends are under-clustered. Why?

The Economics of Fertility: A New Era

Applied positive meta-science

Tether-free photothermal deep-brain stimulation in freely behaving mice via wide-field illumination in the near-infrared-II window

All four of the key DNA building blocks have been found in meteorites. Gives us a clearer picture of the origin of life, leaning towards self-replicating collections of RNA or DNA strands.

Nitrogen and oxygen from the air can be used to produce NOx species, which can be converted into fertilizer. In fact, an early industrial fertilizer production did just that, but was out-competed by Haber-Bosch. Given cheap renewable energy, this could become a competitive process for fertilizer production. This research might benefit from the existing literature on the reverse reaction: selective catalytic reduction. Nitricity is working on a process like this. I did a short literature review on fixing nitrogen in this way, here are some relevant publications:

From the Birkeland–Eyde process towards energy-efficient plasma-based NOX synthesis: a techno-economic analysis

Reducing energy cost of NOx production in air plasmas

Plasma assisted nitrogen oxide production from air: Using pulsed powered gliding arc reactor for a containerized plant

Sustainable NOx production from air in pulsed plasma: elucidating the chemistry behind the low energy consumption

Experimental Study of Nitrogen Oxides and Ozone Generation by Corona-Like Dielectric Barrier Discharge with Airflow in a Magnetic Field

NOx production in plasma reactors by pulsed spark discharges

Synthesis of Nitrogen Oxides in ICP/RF Plasma

Nitrogen Fixation as NOx Enabled by a Three-Level Coupled Rotating Electrodes Air Plasma at Atmospheric Pressure

Global Totalitarianism

I think that the creation of a global government is a bad idea, partially because a global totalitarian regime is one of a few important existential and suffering risks. So we should think about how to prevent global totalitarianism from forming.

Make World Government Redundant

Why do people want a single governing structure in the first place? Generally, proponents want a world government to enforce laws that produce global public goods. World peace and global redistribution are often cited as valuable institutions that world government could provide1.

But this assumes that global government is the best way to achieve these goals. Consider our highly fractured world, does it require global government to maintain international law, treaties, aid, and coordination between states? Of course not. It’s possible to produce most if not all of the value we would get from global government via international cooperation.

To this end, states need formal systems for funding public goods. For example, countries could finance global public goods using treaties similar to a wealth dominant assurance contract.

Even with a global government, world peace will be difficult to provide. Civil wars happen frequently within states, and there is no reason to believe that a state spanning the globe would prevent local skirmishes or top-down oppression.

However, the situation can be improved even in our fractured world. Mutual defense agreements, proxy wars, and guerrilla tactics have made war extremely costly to aggressors, leading to a decline in fighting across the globe. Building direct, peaceful ways to punish aggressors will continue to reduce violence.

There are also ways to centralize redistribution without requiring a global state. Wealthy countries and their citizens could easily fund basic income or other poverty reduction programs in developing countries.

I hope that by eliminating the necessity of world government, we can begin to temper the desire for it. To this end, the public must be convinced that a global state is not only useless, but dangerous as well.

Prevent the Causes of Totalitarianism

Typically, the growth of government power follows large negative events. Every danger creates totalitarians that claim to fight it, and voters give governments more power in times of fear.

Unfortunately, this concentration of power is slow to dissipate. How can we combat the ratchet of state power leading to less free nations?

Most directly, the problems that aid totalitarian governance should be dealt with. Risks like climate change, pandemics, nuclear war, and AI risk2 must be mitigated without the help of a strong state.

More subtly, the scope and importance of such problems should be communicated accurately, with a focus on solutions3. Totalitarians don’t need to wait for trouble to arise, often they can inflate the proportions of a problem in order to justify power (or invent a new problem entirely). These tricks can be combated with public knowledge about the scope of problems, and their resolution. Only by fighting our propensity towards fear can we tackle global issues without losing our sovereignty.

Institutions must be in place to prevent the growth of a global state even if popular opinion sides with totalitarians. A strong culture of freedom and legal protections for individual rights are a check on the growth of power.

There are a few important norms that can help prevent global totalitarianism. Free speech norms can counter propaganda and take the wind out of attempts to stir fear. Free movement can allow people to leave authoritarian regimes, halting the growth of bad governments4. Free access to space is a subset of free movement, and provides the ultimate exit right5.

Resist Authoritarian Regimes

In the worst case, it will be useful to develop tools to resist global totalitarian regimes.

Federated military power is the most direct way to do that. If diverse groups have the means to defend themselves, it makes it much harder for a global state to oppress people. Large inequalities in military power can lead to a single government dominating many others.

Strong cryptography is also critical. Communicating and coordinating without being watched allows individuals to fight propaganda and organize resistance efforts6. This cryptography must be robust to hostile environments where even the use of cryptography is treated with suspicion.

Resistance efforts can be bolstered by good knowledge about how to fight totalitarian regimes. Publicly accessible documents on how to coordinate resistance, sabotage governments, and use guerrilla tactics come to mind. This knowledge must be continually updated to account for modern surveillance and weaponry.


Fortunately, without the assistance of an AI, global government will be slow (or unlikely) to form. Even then, it is unlikely to turn into a totalitarian regime that will persist for a long time. Hopefully, by the time global government does pose a threat, most of humanity will have expanded beyond its reach, and those under its rule will have easy ways to leave.

Regardless, many of the objectives set out here are useful in their own right, even if we never need to defend against global totalitarianism.

  1. International taxation is another stated goal of world government which should be avoided. For those that want to tax mobile individuals and businesses, there are ways to tax their activities within a border that are equivalent to taxing them beyond it.
  2. AI risk is particularly important in this regard, since these systems can compliment authoritarian regimes, or become authoritarian themselves.
  3. Fearmongering regarding climate change seems to be subsiding, but pandemics will repeatedly be used to justify more power. Consistent economic growth can counteract this fearmongering as well.
  4. Freedom of movement can also reduce collusion between states.
  5. Notice that exit rights are part of the Universal Declaration of Human Rights and that access to space is part of the Outer Space Treaty. Strengthening our commitment to these ideals is key.
  6. Satellite internet seems useful in this regard since it is hard to take down and can be used for cryptography.

Perpetual Lottery

This is another installment in a series of weird ways to finance public goods.

Here, the basic idea is to fund public goods using a lottery mechanism, but borrow from future philanthropists to provide the prize money.

The One Round Case

Lets start with a simple version. We run a lottery to produce public goods, but get the prize money after-the-fact.

It works like this:

  1. People buy lottery tickets, and the revenue from these sales is spent on a particular public good.
  2. The prize money is collected 10 years later from purely voluntary contributions and the prize is awarded to a random ticket-holder.

It’s a small change from normal lottery funding. All we are doing is pushing the prize award out a few years. But already, this creates interesting feedback loops. For example, if the lottery funds a public good which makes the world vastly richer, then 10 years later, the prize money will be collected from a bunch of super rich people with a lot of money to throw around. If you expect the prize to be large, then there is an even greater incentive to buy lottery tickets (and fund the good). And if you hold lottery tickets, there is an incentive to make the world much richer. So this lottery can produce a kind of self-fulfilling prophesy.

Why would people of the future give a prize for something that has already been built? For the same reasons someone might want to fund a public good retroactively. By rewarding ticket holders, it establishes a reputation for future lotteries, increasing the number of public goods that get funded today.

The Perpetual Case

The previous approach wasn’t clever enough; let’s make it more complicated.

We can chain together the lotteries from the previous section like so:

  1. People buy lottery tickets, and the revenue from these sales is spent on a particular public good.
  2. The prize for the lottery is awarded 10 years later, when a second lottery is being run. A fraction of the revenue from the second lottery is used as the prize for the first lottery.
  3. The process repeats forever, using a portion of next years lottery revenue to fund the prize from the previous year with most of the revenue being spent on public goods.

This is neat for a couple of reasons. First, unlike normal lotteries, you never need to gather prize money; just sell more tickets. Second, buying tickets both funds new public goods, and retroactively funds public goods (by rewarding those who funded public goods with a larger lottery prize). Third, because the prize depends on future economic growth, it can help catalyze the creation of valuable public goods.

There are lots of ways to adjust this approach to better suit the particular public goods people want to fund. The lottery money could be spent by an elected committee, some portion of the revenue could be used to fund future lotteries, and tickets could be awarded to individuals to incentivize the production of public goods.

Notice that this system depends pretty heavily on steady growth, since this drives larger prizes in the future, which jump-starts giving today.

NFT Lotteries

I warned you I would blather on about public goods funding long after it stopped being useful. Here, I present an implementation of public goods funding which leverages the surprising profitability of NFT’s.

First, lets talk about lotteries as a public goods funding mechanism. It turns out that if you offer a lottery for a fixed amount and spend the proceeds from ticket sales on a public good, the lottery will provide more funding for the good than voluntary donations!

In fact, instead of donating directly, a philanthropist can do more good by using their donations as a lottery prize. This is nice because public goods are typically underfunded, so improving upon the amount people voluntarily donate is valuable.

Even better, if the lottery prize is large enough, the funding for the public good will approach the optimal amount.

Lotteries are simple, well known, individually rational, efficient, sybil-proof, and improve on voluntary contributions. These factors make them my current favorite public goods funding mechanism.

But like all things, we can improve upon them by adding an NFT somewhere.

In this case, instead of using actual money as the lottery prize, we can offer an NFT as the prize. Some of them are worth a lot of money!

Why is this beneficial? NFT’s are a valuable prize, but cost very little to create. Imagine trying to fund a lottery using $10,000. You can either directly use your $10,000 as the lottery prize, or you could commission an NFT which might have an auction value of millions of dollars. By using an NFT as a lottery prize, you can multiply the value of your funds going to the public good. Even better, NFT’s themselves are a public good, as anyone can view and enjoy the art (even if they don’t own the NFT).

In general, subsidizing public goods using something more valuable than it cost to make can multiply the impact of voluntary contributions. Art, diamonds, and prestige are all more valuable than they cost to produce, and would work well here. What else?

Links #22

Taxation and National Defense

Fusion without Fissiles: Superbombs and Wilderness Orion

Increasing agricultural productivity across Sub-Saharan Africa is one of the most important problems this century

Charitable donations and the infrastructure supporting them (H/T Austin Chen)

The Overedge Catalog: New Types of Research Organizations

Interesting software: MakeSpace (browser), Natto (spatial JS development), Flowsheets (reactive spreadsheets), and Enso (visual data manipulation).

Seeds of Science is interesting, making the publishing of scientific ideas more blog-like and lowering the barriers to entry. This could be the beginning of a pipeline from “crazy idea” to “published paper” where each step of the way is managed by knowledgeable people in the field.

A company working on scientific publishing: Authoria

15 rules for blogging

Lots of AI progress recently with DALL-E 2, PaLM, and the new scaling laws. But I want to highlight this paper which uses natural language to allow different language models to coordinate and talk with each other. This seems like a pretty big advance, more important than simply scaling up previous work. Transformers are now foundational models that can do a lot of different things, arguably the next paradigm in AI. I expect to see many more clever insights into how to use them.

Related: Machine Learning for Hardware Design. The natural next steps here are to produce optimize chip which designs other chips and produce a chip optimized for the transformer architecture. After this, I wonder if ML could improve the fabrication of chips themselves.

Replicating and extending the grabby aliens model

Idea Review

I write down many of the ideas that float through my head. This habit has made me more creative which is useful in other areas of my life. But after 3 years of doing this, I had accumulated roughly 4000 ideas. It seemed silly to leave them unused; shouldn’t I do something with all of these thoughts?

I recently spent a lot of time reviewing this writing. My primary goal was to save thoughts that seemed particularly useful, so I kept ideas that I wanted to explore more, write about, or pursue directly.

As expected, very few made the cut. I decided to save 137 ideas, or 3.5% overall. I probably kept too many since I’m biased towards my own ideas; when I read other lists of ideas, roughly 1% strike me as important. In the future, I will try to reserve only 1%.

Looking through my writing, I was stuck by how similar my old thoughts are to today. Though my focus changes significantly from year to year, some things keep coming back to me. In some sense this reflects a lack of creativity; I keep rehashing the same points in my writing. But it also seems like certain lines of thought take a really long time to develop; I had to keep coming back to the same point in order to develop it fully.

Some other things I noticed while sifting through old thoughts:

  • I more frequently saved recent ideas. This is some combination of knowing more and being more biased towards things I thought about recently.
  • As a corollary, reviewing an idea long after I had written it was valuable since I was able to dispassionately assess whether it was important.
  • Over the years, my thinking goes through phases where a certain topic comes up frequently and then dies out (though some things have stuck with me).
  • This process also sparked many new ideas. Some bounced around in my head for a few days before popping up as something new and interesting.

For a while, I felt like there was something missing with my idea generation process. I was producing a lot of ideas but not actually prioritizing or acting on the things I had written. I think this was the final piece, and, in the future, I will make it an annual habit.

With this addition, my personal idea generating process is mostly fleshed out. A few tweaks here and there will improve the process, but investments in other areas of the marketplace of ideas will be more valuable.

The Age of Altruism

Trends in Giving

I see several trends that could dramatically change the scale and character of charitable giving.

Consider the explosion of online finance. Companies like PayPal, Stripe, MPesa, and Wave have reduced the frictions of moving money over the internet (cryptocurrency will only accelerate this progress). It’s easier than ever to send money across the globe and engage in sophisticated financial arrangements.

I have also speculated that more money will be spent on charity as people get richer, the population grows, and inequality rises. This spending will only accelerate as expectations of the wealthy increase.

This rise in charitable giving is enhanced by the fact that the internet raises awareness of global injustices and makes personal choices more public. Both increase the driving force for giving to charity.

The rise of the Effective Altruist movement complements these trends by providing a rationale for giving which has become popular amongst a core of thoughtful, influential individuals. EA also provides a new awareness of big problems by emphasizing issues like existential risk and global poverty over more local issues.

Easy movement of money, rising global wealth, and increasing interest in charity should combine to push more money into the charitable ecosystem. New approaches to public goods funding will multiply these contributions, leading to a dramatic increase in charitable spending.

The Age of Altruism

Bringing these factors together, I think the “charitable industry” will grow to encompass roughly 1% of global GDP, moving $1-2 trillion/year by 2030. This torrent of money will change charitable giving, bringing about an “age of altruism”.

This is important because increases in size can lead to huge changes in character. Like other fields, I expect giving industries will have significant returns to scale. The massive amount of money directed towards charities will open up new approaches to philanthropy, making it fundamentally different.

Already, the way charities move money is changing. Effective Altruist organizations are experimenting with different ways to spend money such as donor lotteries, regranting, and S-process funding. There is also a growing need for a more streamlined grant application process, and I expect to see innovation in this space. In general, there are a lot of opportunities to improve the financial infrastructure that supports charities.

But the most interesting part for me is the opportunity to create new social structures for collective giving like charitable banks, charity prediction markets, and charitable social media. Imagine how different our world would be if a significant fraction of our finances and social activities were oriented around doing good.

Potential Problems

These are encouraging trends, but it’s important not to get complacent. The transition to an age of altruism should be carefully managed.

For example, I have some concerns about charitable social media. If donations are directed towards causes with high signalling value, we could end up in a equilibrium where everyone donates too much or focuses on ineffective causes. It would be especially damaging if philanthropy became politically polarized, leading to spending on zero sum competitions. We need to consider how emphasizing charitable giving over personal happiness can be taken too far.

The unilaterialist’s curse is another potential issue. If a huge volume of money is directed to charitable causes, might some of that money be given to organizations that are actively harmful? What can we do to prevent this?

It’s also crucial avoid lock-in across the giving industry. For example, EA was criticized a few years ago for focusing on straightforward, incremental causes instead of institutional reform. This criticism was taken seriously and EA seems to have improved. But there is no guarantee that things will go well the next time. What other blind spots are lurking in the charitable industry? How can we ensure that our approach to philanthropy adapts as the situation demands?

To be clear, I am not very worried about the downsides of an age of altruism. It still seems like spending more money on charity will be good overall, but we shouldn’t assume that this is the case.

Regardless of the nature of this change, the age of altruism is coming. It’s time to work through the implications, and start building institutions that can bring out the best in these trends.

Links #21

This group is creating contests to promote science through science fiction. Its great to see more optimistic science fiction getting produced!

On The Ones Who Walk Away From Omelas

I really liked the concept in the TV show The One, where a genetic test to find true love has been invented. Show is mostly focused on drama but I like science fiction that invents new concepts.

Searching for Outliers

Camps is experimenting with new way to discuss ideas. Interesting!

20 Modern Heresies

This piece offers an interesting perspective on innovation, mainly that lack of creativity is an adaptive trait in children, and that innovation requires incremental advances from highly skilled individuals. I’m not sure I buy that conclusion, but if true it would necessitate more focus on mastery and less search for creative ideas.

Cause profile: Cognitive Enhancement Research

Non-Invasive Deep Brain Stimulation

Neurons are fickle. Electric fields are more reliable for information.

Dense neuronal reconstruction through X-ray holographic nano-tomography

This post extrapolates connectome mapping which suggests the human connectome might be mapped by 2050. This Metaculus question suggest whole brain emulation could happen by 2080. I think timelines for the creation of Em’s are shorter than most in the AI safety community expect.

New Scaling Laws For Large Language Models. Should this extend peoples timelines for AGI?

A footnote in this post by Basil Halperin:

The goal should not be to build an AI that has a decision algorithm which never is time inconsistent, and always cooperates in the prisoners‘ dilemma, and never generates negative externalities, and always contributes in the public goods game, and et cetera. The goal should be to build up a broader economic system with a set of rules and incentives such that, in each of these situations, the socially optimal action is privately optimal – for all agents both carbon-based and silicon, not just for one particular AI agent.

I completely agree! This was a subpoint in a previous post of mine, but I still haven’t seen this idea explained at length.

Robin Hanson discusses a new publication which improves upon the grabby aliens model.

High-throughput molecular recording can determine the identity and biological activity of sequences within single cells

Recording of cellular physiological histories along optically readable self-assembling protein chains

I was starting a literature review on producing hydrogen directly from seawater, but these papers convinced me that it is not particularly useful relative to pure water splitting. Desalinating the ocean water before electrolysis is cheap enough not to matter much:

Is direct seawater splitting economically meaningful?

Seawater electrolysis for hydrogen production: a solution looking for a problem?

AI Will Multiply

AI alignment work typically treats AI’s as single entities. While I agree that this is a good approximation, I think they will be better described as a highly coordinated population of agents. Rather than simply grow in size and acquire more resources, AI’s will find it in their best interests to split into many smaller agents.

One reason to multiply is to save resources. It may be more efficient to break into smaller, simpler agents to handle multiple tasks rather than use a single large agent to handle all tasks.

Having many copies can leverage certain efficiencies of scale. It also offers the AI a way to increase it’s capability without needing to worry about creating an aligned agent.

Splitting into smaller components also has the benefit of reducing overall risk. A single agent faces a much higher probability of extinction compared to a population of agents. Having many copies allows the AI to diversify its strategy.

Under certain circumstances, copies may also be able to carry out a Sybil attack that a singleton could not, giving the AI more influence in the world.

Copies can gather resources more effectively as well. When an AI needs to cover a large area but cannot adequately control actions at every point, it makes sense to split into independent agents. This is particularly true for space expeditions, where the speed of light makes communication too slow to be useful for making quick decisions.

For these reasons, making copies is a convergent instrumental subgoal.