Links #18

Stop emissions to stop warming, more money more babies, and more.

I switched to using BlueSky for tweet-length writing, follow me here!

1.

Stop emissions, stop warming. A great post by Andrew Dessler that answers a question I’ve had for a long time: what happens to global temperatures when net emissions reach zero? Climate scientists once thought there would be momentum and the Earth would continue to warm. But that’s not the case. Because the ocean and land are constantly drawing CO2 from the atmosphere, net zero means that the CO2 concentration begins to fall leading to constant global temperature.

The situation is different if we maintain a constant CO2 concentration. A molecule of CO2 stays in the atmosphere for centuries, warming the planet for a long time. On top of that, it takes a long time for the ocean and atmosphere to reach thermal equilibrium. If I’m understanding correctly, the equilibrium temperature at our current CO2 concentration is 10 C warmer than baseline. But this would require thousands of years at constant CO2 concentration.

That’s very good news! We just have to reach net zero to pause climate change. Dessler thinks we have the technology to do so, as long as we have the political will.

2.

Lyman Stone has a good post on demographic decline, arguing that western countries pushed the idea that development required a move to low fertility. This “developmental idealism” lead many countries to push for low fertility via policy and top-down cultural change. In addition, falling child mortality led people to lower their fertility desires, and desires are the strongest predictor of completed fertility¹.

What I found most interesting were posts he linked that argue that fertility and income are positively related in many subgroups. He explains why this is so important:

The income-fertility relationship is important because if rising income inevitably causes falling fertility, then there’s no way to achieve higher fertility and modern living standards. If the only way to have higher fertility is to become Amish, probably not many people will make that choice, and societal fertility is doomed to keep falling. But if—as I believe to be the case—there is no necessary and inevitable negative link between fertility and income, and if in many cases income and fertility may be positively connected, then pronatal policies … are likely worth trying. If the income-fertility relationship is not inherently negative, then there’s a chance we can make a world where people have the families they want to have, while also enjoying modern standards of living. But if the income-fertility relationship is inherently negative, then there’s no such hope.

Recently, the relationship between growth (and thus incomes) and fertility became positive (albeit small).

Cultural norms shape whether higher income leads to higher fertility. Not every culture and subgroup exhibits this relationship. But if those norms can be built and maintained, then economic growth can create a virtuous cycle of rising incomes and rising fertility which in turn boosts long term growth.

What I would like to learn more about is whether relative or absolute income is more important for fertility. Do people have kids when they are relatively wealthy or when the cost of having kids is a smaller portion of their disposable income?

Some other posts by Stone:

No Ring, No Baby: How Marriage Trends Impact Fertility and more recently, this piece on marriage and fertility. It seems that marriage subsidies would probably raise fertility.

Higher Rent, Fewer Babies? Housing Costs and Fertility Decline. Building houses where people have the space to raise kids raises fertility.

3.

The Future of Education. Math Academy has some impressive claims of accelerating student learning in math using AI to determine how much a student knows and select tasks with spaced repetition to help them retain things. It sounds like this mostly works for self motivated students and I would guess that 30%-50% of all students would do well if this was implemented broadly. Other students would need regular check-ins to prod them to complete lessons.

Ethan Mollick highlights a study of AI tutoring in Nigeria:

If this actually scales (big if) then kids in poor countries can get high-quality education for the cost of satellite internet. Next question is whether this kind of education actually translates to productivity and welfare gains.

Online education has only had limited success so far, but it’s worth the effort. It would completely change the education system:

1. Kids from any background could receive a quality education for cheap. Talented students can get opportunities regardless of their family background.

2. Since kids can use chatbots to do their homework, we can do away with homework. This is good because homework doesn’t improve learning outcomes and uses up a lot if time. Instead, in-person work and exams will save time and increase fairness.

3. Since students working independently in class, disruptive students will distract their peers less. Teachers can focus their attention on disruptive and under-performing students.

4. Home values will decouple from local school quality.

5. There will be more flexibility to change curricula to achieve better learning outcomes.

I expect if everyone used online education we would need more teachers (but with less training). It seems like a lot of students need a guiding hand to stay on top of their studies. Because different students will be using different software and get different levels of support, standardized exams will be necessary to verify that a student has mastered a particular subject. This combination of online learning and verifiable exams could replace the formal credential system for some students.

4.

A nice figure from this report on highway performance demystifies traffic for me (it’s a Laffer curve!). Highways achieve their highest throughput at pretty low speeds (~45 mph). Congestion pricing can help achieve this optimal throughput, which works in theory and in practice. After that, you can either add more lanes, subsidize busses, or build public transit to increase throughput.

As an aside, the congestion pricing tracker I linked above is a great way to illustrate the success of the policy. Such a public and accessible display of congestion pricing working in practice will lead to a lot more cities implementing it. Good!

Everything else

Hang on, are there ANY lost minerals? Julian Simon remains undefeated.

Medellín was able to cool the city by 2 C by planting trees (H/T Tom Whitwell).

Nice piece criticizing SO2 injection for slowing climate change. Essentially, it might be 2x more expensive than we thought. I disagree with the conclusions the author draws from this though.

Via Separations is a cool company. Using graphene oxide membranes to separate molecules rather than distill them which reduces energy use by 10x.

Cruise ships are seeing remarkable progress. How long until we have seasteading cities?

Full-waveform inversion imaging of the human brain. A 2020 paper on using ultrasound and “full-waveform inversion” techniques developed by the petrochemical industry to image the brain. Simulations suggest they could get sub-millimeter resolution! Every time I see people throwing more computation at a problem, I wonder what the limit is. Would an even better simulation of ultrasound propagating though the brain replace MRI? Could we do coarse connectomics with this?

This paper puts a shaped, ultrasonic transducer into a vein to image structures near the vein. Could we mount a similar system on a stentrode and image the brain?

The Unbearable Slowness of Being: Why do we live at 10 bits/s? As the paper points out, this fact limits the value of brain computer interfaces for healthy people.

Spatially Varying Nanophotonic Neural Networks

Old, underrated result from OpenAI back in 2017: quirks of floating point arithmetic make “linear” neural network layers nonlinear. It would be pretty cool if LLM’s could be implemented with just these simple layers.

Eryney Marrogi discusses how to automate biology and why gene drives are hard for eradicating malaria.

Ticker: Don’t Die of Heart Disease. Best compilation of advice about avoiding heart disease, the leading cause of death in developed countries.

Good discussion here on why we don’t use more probation with GPS trackers to prevent criminals from committing more crimes rather than using prisons. On paper, a tracker would seem to have similar incapacitation effects as incarceration given the near certainty of getting caught. It could be a more humane and lower-cost option.

In practice, criminals often prefer prison to probation because even small deviations from the rules (e.g. being 1 minute late to a meeting with your probation officer) can lead to more total prison time. There’s also technological issues with (accidental or intentional) malfunctions of the trackers. On top of this, GPS resolution (~100 m) is much larger than a typical home which can make it hard to determine if someone is obeying house arrest. But these seem like solvable problems!

Assessing assessors finds that property value assessors increase property values in response to a municipality needing more tax revenue (specifically for local school systems). This kind of issue motivates my interest in auctions for land valuation.

How Cities Become More Prosperous with Size. How do we get more agglomeration benefits? Increasing urbanization, creating gentle density that people actually want to live in, and increasing how far you can go in a given amount of time are important. Self driving cars are interesting because they increase the amount of time you’re willing to drive (and the number of people who can drive) since it’s more pleasant.

Claims that a demographic decline was the major reason for the fall of the Roman empire. I doubt we will get strong evidence for or against this hypothesis, and it would have little relevance to today, but interesting nonetheless. I don’t really understand why farmers would’ve stopped having kids, weren’t they useful on the farm?

Zero-Based Regulation proposes that existing regulations should be repeatedly justified and removed if they are no longer useful. Similar to my post on expiring laws.

Avital Balwit writes about a guy who thinks online ads are good. Jeff Kaufman and I agree.

How to like everything more. You can learn to like things more and find appreciation for just about everything and everyone. Simply learning more about something often makes it more endearing.

What Goes Without Saying. I was nodding along to all of this. Some people don’t share these intuitions and while I wish them well, I don’t want to work with them.

1. It seems that people overcompensate for high child mortality by having more kids, such that areas with high child mortality have higher population growth rates. Across countries and across time, higher child mortality implies higher population growth. But it’s hard to disentangle child mortality from lack of access to contraception or other counfounders.