After a contract had been negotiated with the exact terms that Anthropic pushed to keep in place.

We don't have the actual contracts publicly. So I bet it has terms that let the government do what it wants ultimately by naming exceptions. That way OpenAI can claim it has some controls in place while the military has the freedom it needs practically.

I'm referring to the contract between Anthropic and the DoD from July 2025 in which both parties agreed to Anthropic's acceptable user policies which then later were seen as unacceptable despite the same admin having signed them.

Well, regulating doomsday devices is a reasonable thing to want. A reasonable regulation of such devices would call for proper safeguards and safety testing. I think Anthropic would have been fine with that.

Instead what happened is a one-off nationalist decree that solves none of the two concerns.

They didn't slam on the brakes though. They asked access to be limited to US citizens which ended up being hard to implement but is implementable and IMO addresses zero real concerns.

It's not even that. If Anthropic finds a way to variate citizenship the cat is back out of the bag. None of the AI-related worries I've ever heard about are addressed by limiting access to US citizens.

It's true that the initial tool response still has the same amount of tokens but it doesn't keep dragged along in the longer-lived top context.

Don't you resend after every turn, so splitting it avoids the n^2 token usage (granted it's cached so there's some optimal amount here)

Yes, exactly. You resend it on every turn (assuming no cache hits). This is why using the shorter-lived subagent to take in that context and only return the useful result back to the longer-lived context safes tokens.

That's assuming that this undermining of trust won't result in reduced willingness to share accurate data.

It's already a federal crime to lie to the census bureau, it's just not prosecuted for discretionary reasons.

This seems to be a new iteration of what IMO made frontend work somewhat painful for almost the entire time I've been building software. It used to take the form that people did something with html that it wasn't designed to do. That thing looked cool and so everyone wanted it. This lead to pain and the perception that the tool is inadequate. So we eventually got CSS. And it continued there. Someone figured out a way to get cool dropshadows and rounded corners. These were cumbersome to implement. And so on.

And that's ten minutes every time someone orders a dish with hollandaise because it really breaks when reheating as well. Given how much cost of labor is a factor it's easy to see why hardly any restaurant will serve real hollandaise. Perfect Baumol cost disease example. Maybe something like a Thermomix could solve the economic problem of hollandaise.

Isn't this because there otherwise wouldn't be allowed to fly into US airspace?

I mean for a flight that doesn't go to/from the US.

The US requirement is that passengers on flights to the USA have been processed in conformance with US regulations _and_ since that processing have not had any contact with passengers processed otherwise. It's not in itself a stupid rule but does make the US rules contagious, since either other airports re-build to keep the US-bound and other passengers segregated or they have to apply US rules to all.

This hit Auckland International badly: it had a lovely open atrium with a garden but the rules forced a forest of partitioning walls since passengers were transferring from smaller airports that couldn't quickly adopt the US rules.

I often fly from Milan Malpensa airport, and I’ve noticed there are two separate security areas: one for people flying to the US or Israel, and one for everyone else. I’d always wondered why this was the case, and now I get it.

"What should a company that launches 98% of the world's non government tonnage to space (80% if you include the Chinese government) be valued at ?"

The TAM of that is under $10 billion. So even owning that entire market shouldn't get you anywhere near a trillion. Then factor in the development cost of starship which has been going on forever and still hasn't even made it to orbit.

Even the IPO filing isn't claiming the value comes from the rockets but from data centers in space which seems questionable. The real cash cow right now is Starlink but they aren't leaning into that heavily because those numbers also indicate that revenue growth might be stalling out.

If you just look at the pure numbers the case is very weak. No reason to change the inclusion rules. In fact I'd argue they never should change the inclusion rules. Let the market find a price first. Index funds are supposed to be boring and track the market. Including any recent IPO just adds chaos beyond simply tracking the overall market. It's trivial to buy some SpaceX if one wants and unlike selling your entire fund holding either doesn't trigger a taxable event at all or it's a much smaller one if you cannot avoid it.

I am not looking at the "pure" numbers, and many investors do not, they look at potential. They look at what the possibilities are. If you believe that space will be a significant part of the economy in 20 years, and that SpaceX will be a large participant, the current numbers do not mean a lot, this is not a large established company in a static industry. In 20 years the market cap of just the US could very well be $200T, if space is only 10% of that, there is $20T of valuation for space, and that is just US.

Google when it ipo'd about 20 years ago had a market cap of $23B. It is now close to $5T. Even if it went over 10x overvalued at $230B when it ipo'd, it would still have been a good investment. That is because the internet became a large part of our economy, and Google is a major player.

If you really want to compare the two, Google had a market cap/revenue of $2B/$2.7B, SpaceX is currently $1.8T/$20B, or about 10x the ratio of Google back then.

Yeah, very pricey. Crazy ? Not sure. Do I like the valuation ? No. Would I buy more SpaceX than what will be in my equity ETF ? No. But I am not unhappy that I will own a piece when it comes out. Do I like the change in rules ? Absolutely not, but that is just the way it is. The market, over time, is, and always be a lot smarter than I am.

Your TAM for "space" growing in 20 years from $3 billion to 20T honestly seems like quite the stretch. Starship has been in development for about 10 years and it's still not even been to orbit. I have a hard time seeing that sector grow by a factor of 6,666X in 20 years. It's not been growing anywhere near that despite SpaceX's incredible innovations and price reductions.