Sunday Supplement: Small Bucks for Rockets and Claude Cheaper Than Grad Students

What's the news?

Climate Science

I've been covering the global warming/climate change controversy since the University of East Anglia emails came out — in fact, I was the first to write about it for general readership in the U.S., and would have been first worldwide, except my editors here couldn't believe it, either.

Related: Hacker Releases Data Implicating CRU in Global Warming Fraud (Updated); Violating the Social Contract of Science; Fast Facts About Climategate; Pielke Sr.: Climategate Emails Just a Small Sample of a Broad Issue (PJM Exclusive).

(Good God, that was 27 years ago!)

Bjorn Lomborg, who started asking about the consequences and costs of environmental policy back in 2001 in The Skeptical Environmentalist: Measuring the Real State of the World, came out with a couple of interesting fact checks in the last few days on X.

First on the list was this month's UK foreign secretary, Ed Miliband, who declared: "The wildfires in Europe remind us that the climate crisis is happening now."

On 30 July, UK Foreign Secretary said Europe’s wildfires "remind us that the climate crisis is happening now" NASA’s satellite record, 1 January to that day: Europe’s burned area was 3rd lowest The fires were real. They were not evidence that Europe was burning more than ever https://t.co/XlELmcjPxj

— Bjorn Lomborg (@BjornLomborg) October 2, 2026

If so, it's working by reducing wildfires. And even that is in the face of widespread arson.

Lomborg, who, as I said, has been interested in the consequences of overheated (heh heh) climate policy for decades, also looked at the results of a net-zero goal on energy prices in Germany. The consequences: prices up 222% since 2000.

Expensive German climate policy: Germany faces “structural decline” and is in “free fall”, with price increases “across the economy” Real German electricity price 222% of price in 2000 (households+industry) You can see all the references here: https://t.co/mIW3Clwq1o https://t.co/X9BdPF6YKP

— Bjorn Lomborg (@BjornLomborg) October 3, 2026

Lighten up, Francis.

Alaska's largest glacier retreated more than 30 miles between 1780 and 1860. LIkely due to emissions from Thomas Jefferson's Sport Utility Horse (SUH) https://t.co/r1lHZJhK8r https://t.co/7P3C47H1ZD

— Tony Heller 🇺🇸 🇯🇵 (@TonyClimate) October 4, 2026

Space Economics

I've included a video about SpaceX building Starbase Louisiana in the videos of interest. There is a lot about what SpaceX is paying to build 10, count'em, 10 new launchpads in Louisiana.

Larry Goldberg ran an interesting article on "What Washington Paid to Reach Orbit." He has a full article attached to the tweet, but in short, an Artemis launch of the Space Launch System — if one ever happens — is projected to cost $4.1 billion. The Falcon Heavy launch of the Europa Clipper mission cost $178 million.

https://t.co/Gy99IOoeg8

— Larry Goldberg (@TeslaLarry) October 2, 2026

From the article:

For most of the space age, the US government bought its launches from a small group of established contractors, and for years even the Pentagon's own auditors had trouble saying what a single launch actually cost. Over the last fifteen years that has changed so dramatically that I think it's worth laying the numbers side by side. 

Related: SpaceX 'Failure' Makes Big Bank

Physics (and AI)

Claude did a computation of scattering amplitudes, and got a result beating the previous record. I'll include the tweet because I only vaguely understand the result, but it involves computations in a mathematical model of quantum field theory involving high dimensionality and super-symmetry. I'd be curious what Sabine thinks of the model, but as a computer scientist, what is interesting is that by letting Claude grind at the model for $2000 worth of tokens, they got a new result.

New on the Science Blog: Yes, Claude can do Nine Loops. Theoretical physicists predict how particles behave using formulas called scattering amplitudes. These are notoriously hard to compute, so researchers work with layers of increasingly fine corrections called “loops”—each added loop makes the answer more precise but takes exponentially more computation. Most calculations stop at two or three loops. Eight loops was the previous record in a simplified model physicists use as a testing ground (planar N=4 super-Yang-Mills), set by SLAC's Lance Dixon and collaborators. Last month, physicist and science writer @4gravitons issued a challenge: could an AI push past eight loops in this model, using only the compute budget an academic could reasonably access? Given a single prompt describing the nine-loop problem, Claude ran largely unsupervised for days in Claude Science and solved it using methods developed by Dixon and his colleagues, at a total cost of a few thousand dollars. Dixon independently verified the result, and von Hippel wrote about the experience for our blog. Read more: https://t.co/CS2f2qoIhJ

— Anthropic (@AnthropicAI) September 25, 2026

Back in 1963, Hodgkin and Huxley got a Nobel Prize for a room full of grad students with a desk calculator spending a whole winter computing how a single nerve cell propagates a signal. Now, by the time I was working in the medical school, Hodgkin/Huxley would run on a 1980s PC. I promise you, that's a lot easier.

Closing

Related: Well-Designed Systems Self-Correct

So that's the Sunday Supplement for this week. I'm really still figuring this out, trying new things, so make sure to comment and let me know about this format. In case you missed it (snif) here's Friday's Sky Candy, and of course there will be a new Sky Candy next Friday, and a Sunday Supplement next Sunday.

YouTubes of Interest

SpaceX Starbase Louisiana (24 minutes)

Forgotten Eighth Continent (2:25)

Atomkraft? Ja, bitte! (16:50)