To be clear, women’s work before World War II was more than just the dishes. If you look at the guidebooks published for housewives back then, you’ll see that they were expected to have quite a few skills that most households now generally outsourc to external businesses:

• Feeding the family. This was more than just cooking. They were expected to process foods from a much less processed state (much more butchery of meats and cleaning and processing of vegetable products, dairy products, baked goods), and then preserve foods for out-of-season consumption (pickling, preserving in jams/jellies, home canning, drying, and in some cultures smoking). Much of this work is now done by the industrial food processing industry so that we can buy cans or jars or boxes of the stuff that’s already processed or partially processed. Even our fresh foods have been cleaned and sorted and trimmed to mainly just the edible parts.
• Making and maintaining textiles. We see bits of this surviving into knitting and crocheting as hobbies, but back before the rise of cheap apparel it was important to be able to clean and repair clothes that we’d now just take to our local dry cleaner.
• Maintaining the house itself. Home improvement is masculine coded today, but a lot of the stuff that qualifies as home maintenance was traditionally the work of a homemaker. Plus things like heating the house required active involvement of keeping fires burning and fuel on hand.
• Making household consumables. Homemakers were making their own soap, their own candles, and all sorts of little tools.

The economic shifts that come from women leaving the home for the paid workforce are all over, and some of them are pretty pronounced. But it’s important to remember that women worked hard before they ever got paid for it. Life was toil.

It’s not actually a clear inverse relationship on the individual level, even if the data shows a correlation at the national level.

There are a few things happening that complicate the analysis at the individual level, too:

• Wealth/income are correlated with age, and 40 year olds tend to have both higher incomes and lower fertility rates than 25 year olds.
• Wealth also correlates with race, for better or for worse, and there have always been persistent differences in birth rates by race.
• The sample sizes aren’t big enough to show whether the very rich (95th+ percentile) actually reverse the trend, to where being richer is correlated with higher birth rates, where the curve ticks back upward at very high incomes.
• The correlation is actually the other direction when looking at the individual incomes in certain countries (Netherlands, Sweden, Norway), and the effect is stronger when looking at men and their incomes.

Other country level data also suggest that there are big cultural factors in birth rates as well.

All in all, the relationship between income and fertility is complicated, with lots of other factors at play.

Ed Zitron publishes a lot of pretty biased reporting.

The core thesis is sound, though:

• Anthropic and OpenAI’s revenue comes in from customers.
• The revenue does not translate into profits, because their capital expenditures investing in future capabilities is quite high, and because their operating expenses are also quite high, to where their revenue doesn’t even cover their ongoing compute cost.
• The actual money Anthropic and OpenAI have to spend at a loss comes from their investors.
• The customers are largely vulnerable to shocks and are themselves reliant on investor cash rather than a profitable business model of their own, and are essentially subsidizing a lot of the demand for the core services that Anthropic and OpenAI provide.

Taken together, the whole ecosystem is currently relying on a continued influx of cash from investment: investors taking equity in these companies, lenders/bondholders charging interest on borrowed money, otherwise profitable businesses like Google and Meta steering their other profits into investment into AI.

And so if there’s a shock to investment activity, such as if there’s a war in Iran causing an energy crisis and a global recession in real economic activity, that might translate into a cash crunch, as the investors pull back right at the time that the customers start defaulting on their payments. And if you remove the middleman startup businesses that pay Anthropic and OpenAI more than they receive from their own customers, the underlying “real customer” demand at the actual unsubsidized prices charged by Anthropic and OpenAI will plummet.

I’m not a tech guy, but I am a business/finance guy, and I’m not seeing where the analysis is wrong. The argument is always that there’s a runway to profitability, and they just need to take off before they run out of runway. And we can argue about whether they will or they won’t get to takeoff, but if the business is relying on more runway being built because they know for sure they don’t currently have enough runway to take off, that’s a shaky situation to be in. Even if everyone is clamoring to be their runway-building partner today.

Ed Zitron has a piece out talking about how even if Anthropic and OpenAI are bringing in revenue, most of it is coming from startups in such a way that their revenue will evaporate if the underlying startups crash, especially because the underlying cost of operating remains high (and where improvements in the underlying tech go towards improving performance rather than making operations cheaper).

To me, it’s a pretty persuasive argument that failures at any one part of the web of businesses and funds involved (even if completely unrelated to the AI industry) will cause stress on all of them, because they’re so interrelated in a web of relationships.

Still interesting to see how it is implemented in neighborhoods and buildings that are over 150 years old. I think the Smithsonian museums in our capital are actually the most interesting examples, because many are old buildings whose historical character were preserved, but where wheelchair ramps, railings, and elevators were tastefully and functionally installed many decades or more than a century after the building was originally constructed.

And perhaps the best thing about the ADA is the sidewalk requirements. It doesn’t much matter why a sidewalk developed a raised crack when the ADA requires that it be fixed.

I’m not even disabled, but I’ve pushed baby strollers in different cities (including outside the US) enough to realize how nice it is to be in a city where all the sidewalks and public buildings are ADA compliant.

Are you talking about battery storage itself being about $126/MWhr? Yeah, that incorporated into the solar+battery LCOE, because solar itself is $31, battery is $126, and the weighted average of how much energy is expected to come directly out of the solar panels onto the grid (at $31) and how much is expected to be stored for later ($31 plus $126) averages out to $53, presumably because most demand matches the daytime solar curve and doesn’t need to be stored for later.

Which graph? I linked to a PDF with a ton of graphs.

It’s because fuel is such an insignificant percentage of the overall cost of building, operating, and maintaining nuclear power. Increasing the complexity of the reactor in order to make the fuel use more efficient is basically a nonstarter, economically.

Oh, I agree.

I was a big, big nuclear proponent 20 years ago. But seeing how Vogtle and VC Summer played out, and how that “cheaper” and more “scalable” AP1000 design put Westinghouse into bankruptcy, basically turned me off from the economics of nuclear power.

Oh, and because of how utility generation is paid for, ratepayers in Georgia will be paying for the Vogtle construction and cost overruns in their electric bills for the next 75 years, as the nuclear plant is shielded from competition by price regulators (state Public Utilities Commissions and the Federal Energy Regulatory Commission), so even if newer and better technology comes online, customers in 2080 will still be paying for 2020 technology.

The technology is still neat but I don’t believe there’s an economic future for civilian nuclear power generation. Not anymore.

Last week, the Nuclear Regulatory Commission just approved a new construction of a reactor for the first time in 10 years, to the Bill Gates backed Terra Power. Cool, except it’s projected to cost $4 billion and the government is expected to cover half the cost, to build a reactor with 345 MW of capacity.

In contrast, solar panels cost about $1 million per MW, so an equivalent amount of peak capacity from solar would cost about $345 million, or about 1/12 the price. Solar won’t run all day, but the nuclear plants will also continue to cost money to run after construction is complete.

Looking at the different LCOE estimates of each type of power generation shows that advanced nuclear is around $80/MWhr and solar+battery for all day demand tracking is about $53/MWhr.

Basically nuclear is only economically viable with government support at this point, and we should be asking whether we’d rather have the government support towards other forms of energy.

Lots of non-American banks don’t want to deal with the rules, so it’s easier for them to just say that they won’t open accounts for Americans.

It’s not discriminatory.

I mean it quite literally is discriminatory, but it’s legal and justified discrimination.

Does anyone else think 3% return isn’t that spectacular?

But that’s not the percentage return on investible assets. That’s the increase in his net worth in a year.

Think about the typical upper middle class retiree who might have a 401(k) worth $1 million and a paid off house worth $500,000. If they get a 10% return on their portfolio, their house price appreciates by 5%, and they get $10,000 in social security income while their spending rate is $100,000/year, their net worth would go up by $100k + $25k + $10k - $100k for a total of $35k, which is only 2.3% of their total net worth. Even though their investments did pretty well that year.

Bezos is getting more than 3% return. He’s just spending a lot of it. Like on a $55 million wedding.

How do they get calculated?

This page has answers:

The CPI consists of a family of indexes that measure price change experienced by urban consumers. Specifically, the CPI measures the average change in price over time of a market basket of consumer goods and services. The market basket includes everything from food items to automobiles to rent. The CPI market basket is developed from detailed expenditure information provided by families and individuals on what they actually bought. There is a time lag between the expenditure survey and its use in the CPI. For example, CPI data in 2023 was based on data collected from the Consumer Expenditure Surveys (CE) for 2021. That year, over 20,000 consumer units from around the country provided information each quarter on their spending habits in the interview survey. To collect information on frequently purchased items, such as food and personal care products, approximately another 12,000 consumer units kept diaries listing all items they bought during a 2-week period that year. This expenditure information from weekly diaries and quarterly interviews determines the relative importance, or weight, of the item categories in the CPI index structure.

The CPI represents all goods and services purchased for consumption by the reference population (U or W). BLS has classified all expenditure items into more than 200 categories, arranged into eight major groups (food and beverages, housing, apparel, transportation, medical care, recreation, education and communication, and other goods and services). Included within these major groups are various government-charged user fees, such as water and sewerage charges, auto registration fees, and vehicle tolls.

If you want to see the current makeup of the basket of goods whose prices are tracked, and their weights in the index, here is Table 1 of the most recent report. And if you want to follow the price of a specific category over time, the Federal Reserve Bank of St. Louis keeps a really helpful interactive chart service for almost every public economic stat. Here is Table 1 of the CPI report.

It’s a lot of data collection on prices across a lot of transactions, and a lot of list prices, and a lot of locked in contract prices, to determine how much people are spending on different types of things, whether the quality of those things is changing over time, and what percentage of a typical household income gets spent on those types of things.

1 oz = 28.35 g

Maybe that 1/3 of a gram is shorting someone, but a single ounce is closer to 28g than to 30g.

Obviously depends on the specific item.

Here’s the national averages for:

Flour

Milk

Boneless chicken breast

Or an entire table of food items they’ve been tracking monthly prices on for years.

The Five Dollar Footlong was a promo created in 2003 when the normal price of a footlong was $6, by a single franchisee. By the time the promo went national, supported by the chain itself (and a national ad campaign), in 2008, that became a big enough deal to really move sales. And they watered it down at some point (by late 2010 when I was working next to a Subway and no other lunch options, I remember it only being a specific sandwich that rotated monthly, with all other footlongs regularly priced). And it was eventually discontinued in 2012.

It’s hard to pin this particular promo and call it totally representative of all pricing in the mid 2010s.

$5 in April 1980 dollars is the inflation adjusted equivalent of $20 in December 2025:

https://data.bls.gov/cgi-bin/cpicalc.pl?cost1=5.00&year1=198004&year2=202512

You make periodic deposits

We pay a premium, which is a monthly or weekly payment to the insurance company in the same amount each time.

Then, when we see a doctor, we have to pay a copay (a single payment in a fixed amount), coinsurance (payment of a particular percentage of the whole cost), and a deductible (either a per-visit or per-year amount where we have to pay ourselves before an insurance company pays). Together, these types of payments are known as member payments, member responsibility, or out of pocket payments, and they’re capped at a particular amount per year (at most $9,200 for an individual or $18,400 for a family).

It’s a complex system, and insurance is only a part of the problem. Plenty of countries have private insurance and don’t have these issues (Germany, Austria, Switzerland, Japan, South Korea). And many of the providers in the US (hospitals, doctors, clinics, labs) are scummy corporate profit-driven providers and try to enrich themselves at the expense of insurance (including government and nonprofit insurance), so there’s a lot of fraud and anti-fraud measures creating messy overhead and inefficiency.