There is always ample discussion about progressive tax rates in the United States. For those unfamiliar, income earned within certain ranges are taxed differently. Higher income is taxed higher. For Datawrapper, Luc Guillemot charted the rates for countries in Western Europe.

The x-axes represent income levels as a percentage of average income in each country. The y-axes represent the tax rate for the income level. The black bars show averages for the European Union. Belgium, with the steepest climb, increases taxes the most, whereas Hungary and Bulgaria use a flat rate across income levels.

For MIT Technology Review, James O’Donnell and Casey Crownhart ran numbers and interviewed experts go piece together a projection for how much energy AI will use. The takeaway is that it’s impossible to know with any certainty, because companies don’t disclose what they’re building.

The Lawrence Berkeley researchers offered a blunt critique of where things stand, saying that the information disclosed by tech companies, data center operators, utility companies, and hardware manufacturers is simply not enough to make reasonable projections about the unprecedented energy demands of this future or estimate the emissions it will create. They offered ways that companies could disclose more information without violating trade secrets, such as anonymized data-sharing arrangements, but their report acknowledged that the architects of this massive surge in AI data centers have thus far not been transparent, leaving them without the tools to make a plan.

“Along with limiting the scope of this report, this lack of transparency highlights that data center growth is occurring with little consideration for how best to integrate these emergent loads with the expansion of electricity generation/transmission or for broader community development,” they wrote. The authors also noted that only two other reports of this kind have been released in the last 20 years.

Hardware for AI uses a whole lot of energy while training on data from the internets, processing queries, and hallucinating surprising solutions. Alex de Vries-Gao, from the Institute for Environmental Studies in the Netherlands, published estimates for how much energy and compared to energy demand for countries.

Over the full year of 2025, a power demand of 5.3–9.4 GW could result in 46–82 TWh of electricity consumption (again, without further production output in 2025). This is comparable to the annual electricity consumption of countries such as Switzerland, Austria, and Finland (see Figure 2; Data S1, sheet 6). As the International Energy Agency estimated that all data centers combined (excluding crypto mining) consumed 415 TWh of electricity in 2024, specialized AI hardware could already be representing 11%–20% of these figures.

There are many assumptions behind the estimates, and they could be lower or higher depending on the unknowns, but most signs appear to point to steep increases.

We should probably plan for that. It doesn’t seem like this AI train is going to slow down any time soon. (via Wired)

Downloading survey microdata from public resources can be tricky. Sometimes the documentation is sparse, the tools are outdated, or the datasets are tucked away in obscure FTP subdirectories. This is annoying when you just want to work with the data.

Analyze Survey Data for Free, maintained by Anthony Damico, aims to streamline the download process via R. From a decade ago:

Governments spend billions of dollars each year surveying their populations. If you have a computer and some energy, you should be able to unlock it for free, with transparent, open-source software, using reproducible techniques. We’re in a golden era of public government data, but almost nobody knows how to mine it with technology designed for this millennium. I can change that, so I’m gonna. Help. Use it.

The site received an update to make downloading easier across 49 public datasets. Given the data takedowns these days, now seems like a good time to make quick use of the resource.

Roger Peng and Hilary Parker started the statistics and data science podcast Not So Standard Deviations almost a decade ago. It was one of the few podcasts I kept up with while I drove my kids to school. They posted their last episode last month.

Pouring one out for NSSD.

As we enter a time when people question the usefulness of vaccines, even though there are clear benefits, Neil Halloran revisits a time when vaccines did not exist. With a mix of charts, information graphics, and photographs, Halloran tells the story of the smallpox vaccine. High mortality rate and hundreds of millions dead transformed to zero.
Read More

The Congressional Budget Office published a report estimating effects on household income if the GOP’s budget reconciliation bill pushes through. The poor and middle class will effectively net less in income and the upper class, especially the top 0.1%, will net more. G. Elliott Morris describes it as paying more for less, which is a raw deal.

In other words, by decreasing the amount of tax revenue it gathers from the ultra-rich, decreasing the transfers it makes to the poor, and increasing its overall spending, Republicans are asking middle-income and poor families to shoulder a much larger share of the federal deficit — all while they get less from the government. They are asking you, in summary, to pay more for less.

The chart above is from CBO. It compares income for the lowest decile against the highest as a percentage. Check out Morris’s chart that shows the shift in dollars. The bars grow much taller on the high end with an absolute scale.

As the administration tries to block international students from attending Harvard University, NYT’s the Upshot charted the schools with the highest percentage of international students.

I don’t know anything about Illinois Tech, but whoa, over half of undergraduates and graduate students are from outside the U.S.

The administration is making it more difficult if not impossible for foreign students to attend college and universities in the United States. Catherine Rampell, for Washington Post Opinion, argues that doing so is increasing trade deficits when treating education as an export.

We also run a huge trade surplus in this sector, meaning that foreigners buy much more education from the United States than Americans buy from other countries. In the 2022-2023 school year, more than three times as many international students were enrolled in the United States as there were American students studying abroad. Translated to cash: Our education-services trade surplus is larger than the trade surplus in the entire completed civilian aircraft sector.

On top of that, the people who are able to study abroad are often hard-working and the brightest in their class. They provide American students with fresh perspectives.