Never a good sign for journalistic objectivity when the reporter says "Wow"

 ... particularly when it's a story about robots. 

From Planet Money 

JUSTIN KRAMON, BYLINE: We're standing in a restaurant kitchen in Philadelphia's Chinatown, staring at their newest cook, Robby the robot wok-bot. The automated wok stands 6 feet tall and is mostly metal, kind of looks like a washing machine with no door. At its center is this basket. That's the wok where Robby cooks its dishes.

KENNY POON: He can make over 5,000 different dish.

ERIKA BERAS, BYLINE: Five thousand?

POON: Yes.

BERAS: Wow.

That's Kenny Poon, co-owner of this fast casual restaurant, InstaFoodz. He shows us how Robby works. Poon selects the dish he wants Robby to cook from its touchscreen menu - beef chow fun. Then Robby the robot tells Poon the human what precut raw ingredients to add to the hot spinning wok as it heats up and spins. Different tubes squirt in sauces and seasoning as the ingredients are tossed around. Poon says not only is Robby efficient, but his cooking is just as good as a human's.

KRAMON: Do you think you can tell the difference which one's made by the robot?

POON: I don't think so.

KRAMON: Because Robby's so easy to use, Poon says his labor costs have gone down.

POON: Now, I don't have to require a main chef.

KRAMON: How has it changed the staffing for the restaurant?

POON: It's easier. So now I don't need to ask them - what's your skill? - no more. All I need to ask them - what's your availability?

BERAS: That's right. He says the machine is so good he's shifting his focus from making sure he hires and trains staff that are skilled cooks to staff that just shows up.

There is more to this report, but as far as I can see, the only source of information for the part about the robot wok is from the owner of this gimmicky themed restaurant that probably depends on PR and novelty seekers for much, if not most, of its business, someone who has a huge incentive to spin this as big as he can, making it sound like the future of the culinary arts.

The 5,000-recipe claim certainly sounds impressive, but did anyone actually verify it? And even if it’s true, are these dishes that require a wide range of techniques and skills? Can the robot, for example, make Cantonese scrambled eggs (which, according to YouTube isn't that difficult but would be out of the standard stir-frying repertoire) , or is it limited to recipes that rely on a small set of actions just with different ingredients and cooking times?

What’s the cost? How much maintenance does it require? How labor-intensive is the operation?

To be fair, this isn’t some goofy, wildly impractical bipedal humanoid lumbering around the kitchen. At least from the pictures I found online, the design of the robot seems to be squarely focused on functionality. There’s no reason to believe it isn’t a real innovation that will pay for itself and might even become a common fixture in restaurants in the years to come. 

There might well be a real and genuinely interesting story here about the future of automation in the food industry, but reporting on that would require actual work—reasonable skepticism, critical thinking, and independent thought. Instead, the Planet Money team cranked out a badly reported, standard-narrative-template story based almost entirely on a single source who had an enormous incentive to tell them what they wanted to hear.

 

Probably even worse than you realized

Historian Bret Devereaux (who has been on our must read list for a long time) has posted the best military overview of the war I've seen so far. It's a long piece (around 7,500 words) that takes a deep dive into the strategic and tactical state of the conflict. I'm not even going to try to summarize the essay, but I did want to highlight a couple of passages.

Iran is a large country

It has a population just over 90 million (somewhat more than Germany, about the same as Turkey), and a land area over more than 600,000 square miles (more than four times the size of Germany). Put another way Iran is more than twice as large as Texas, with roughly three times the population. 

More relevantly for us, Iran is 3.5 times larger than Iraq and roughly twice the population. That’s a handy comparison because we know what it took to invade and then hold Iraq: coalition forces peaked at half a million deployed personnel during the invasion. Iran is bigger in every way and so would demand a larger army and thus an absolutely enormous investment of troops, money and fundamentally lives in order to subdue.

...

And you may then ask, here at the end: if I am saying that Iran is being hammered, that they are suffering huge costs, how can I also be suggesting that the United States is on some level losing? 

 

And the answer is simple: it is not possible for two sides to both win a war. But it is absolutely possible for both sides to lose; mutual ruin is an option. Every actor involved in this war – the United States, Iran, arguably Israel, the Gulf states, the rest of the energy-using world – is on net poorer, more vulnerable, more resource-precarious as a result.

For more historical context, I recommend Devereaux's post Strategic Airpower 101.

The Double Taco -- updated

Groundhog Day: Trump saw his shadow, so there’s going to be two to three more weeks of war.

(Trying to remember—what happened around 9:00 Eastern time?)

Let's check with Josh Marshall.

I think any press person who watched President Trump’s Iran cheer-up session speech on truth serum would have to concede that this was a speech he shouldn’t have given. He meandered. He looked bad and worn out. He had the requisite moments when his degenerate inner monologue creeps into the open: he said that free passage through the Strait of Hormuz is something for importer countries in Asia to deal with, that they should “grab and cherish” the Strait, as though it were some underage beauty pageant contestant Trump was hungering to assault. [He also announced the US was "hottest country anywhere in the world by far, with no inflation.’’ which was both off topic and not exactly reality based -- MP] "What is important is that in political and public opinion terms, there was nothing new or newsworthy in this speech. They didn’t even manage to accomplish this in the narrow and cynical sense of saying anything new that could be a fresh point of public discussion. It was a rambling set of unconvincing excuses no one with any real concern or anxiety about this war (the only real audience) would find convincing. Why are you complaining, he asks? This war hasn’t gone on nearly as long as World War II! LOL.

There’s no announcement coming. There is no plan.

— Kai Ryssdal (@kairyssdal.bsky.social) April 1, 2026 at 6:13 PM

The “taco trade” was always based on multiple fallacies: the assumption that Trump would stop doing this stupid/crazy thing before it did any damage; the assumption that he would not revert to the disastrous policy; and the assumption that the chickening out couldn’t go in the wrong direction.

We have seen at least one case of a “double taco,” where Trump was frightened away from a position that was costing him politically, only to be frightened back into it by fear of MAGA backlash. I realize events seem compressed in this administration—it’s difficult to believe it’s been just a little over one year—but it wasn’t that long ago that Trump, under pressure from the agricultural sectors in his base, publicly reversed his position on mass deportation of farm workers, only to flip-flop again almost immediately when Stephen Miller got to him.

There are numerous reasons to believe that we’ve just seen another double taco, this time with the administration coming very close to a humiliating but probably best-deal-he-was-going-to-get surrender, starting with the speech itself (Why request network time to accomplish nothing but spooking the markets?). There were indications that the Iranians were preparing for a ceasefire up until some point during Trump’s speech. Even the Israelis seemed to be preparing for an American withdrawal.

It was, as mentioned before, an ugly choice, but it would have alleviated most of Trump’s immediate problems—and, given that the markets had skyrocketed on the mere possibility, they unquestionably would have rewarded anything that ended the war.

Nonetheless, as best we can tell, shortly before the time came to pull off the Band-Aid, Trump again chickened out, probably spooked by the inevitable backlash to capitulation. Possibly—given what we know about the man—still enraged by that day’s Supreme Court proceedings.

Regardless, another month of war will bring serious pain points in terms of oil/LNG, helium, fertilizer, and significant stagflationary pressure on the economy—not to mention tremendous human suffering.

None of which can be tacoed away.

PS I hadn't seen this when I wrote this post, but I think it provides even more evidence that Trump was intending on close to a full peace-is-at-hand taco as late as Wednesday morning. 

I know I said this before, but this is the last one, I promise.

RAPIDAN: “.. Trump will struggle to declare victory with Iran controlling Hormuz, the regime largely intact, and still in control of .. highly enriched uranium. Absent a ceasefire with Iran, Israel would likely continue independent operations, making a clean US disengagement structurally difficult.”

[image or embed]

— Carl Quintanilla (@carlquintanilla.bsky.social) March 31, 2026 at 3:17 PM

In times that are this strange and eventful, I think it’s important to keep some kind of journal—something that lets us go back and see how things looked and felt in real time. One of the things that people will struggle to understand once the dust settles is the market psychology of the second Donald Trump administration.

I realize we’ve been talking about this for a long time and have probably written more about it than most of this blog’s readers would care to hear, but today really does merit a post.

Followed by this.

 

 

At first glance, the logic here seems to be that, faced with gasoline at $4 a gallon—apparently on track to hit five—Trump will “TACO” his way out of the war and all of the painful consequences that have come with it. That narrative is not entirely wrong (the Trump Pressure Index is reaching dangerous levels), but it is incomplete and doesn’t hold up on its own. Even the markets of 2026, led by the dumbest of dumb retail money, have got to realize that getting out of a multiplayer Middle East quagmire is going to be more difficult than simply declaring victory in a trade war and canceling a bunch of tariffs that weren’t legal to begin with.

I reached out to someone who’s been following the story more closely, including some of the chatter and trial balloons, and with that context it starts to make a certain kind of sense—not in a way that could be called rational, but more in the way that the Joker’s plan makes sense after he explains it. You can see the internal logic behind the craziness. You’d have to be deluded to believe this, but at least the delusions are coherent.

Among the various contradictory statements and implications coming out of the White House, you can find some which, once you get past the belligerent language and chest-puffing, amount to essentially an unconditional surrender to Iran. We let them have the Strait of Hormuz, we unfreeze their assets, and we leave them far stronger than they were before we attacked.

It might even convince the Iranians of the importance of becoming a nuclear power as soon as possible. 

David Goldman, Aaron Blake writing for CNN fill in some of the details:

The Dow, S&P 500 and the Nasdaq just had their best day since May 2025, roaring higher Tuesday in large part because of a report (and semi-confirmation) that the White House is considering an end to America’s involvement in the Iran war without reopening the Strait of Hormuz. CNN later confirmed Trump and his administration increasingly believe that they can’t promise to reopen the strait as a prerequisite to declaring an end to hostilities with Iran.

That would be an extraordinary outcome: The war seems nowhere close to being over and, even if it were, the global economic ramifications of Iran continuing to block the critical waterway would be long-lasting – measured in years, not weeks or months.

Oil trades on a global market, and US crude and gas prices will remain high as long as the Strait of Hormuz is closed – no matter how much “drill, baby, drill” President Donald Trump proclaims.

You’d think that’d be bad news for markets.

Nevertheless, the Dow rose by more than 1,000 points, or 2.4% Tuesday. The S&P 500 was up 2.8%; and the Nasdaq, which had entered a correction last week, was 3.8% higher.

The reason: FOMO from a TACO, the Wall Street acronym “Trump Always Chickens Out.” Trump has repeatedly reversed course on some of his most economically significant policies and proposals, giving markets whiplash and leaving traders with significant losses if they had the wrong end of a bet.

“They’re waking up every morning, going to sleep every night, rubbing their hands together, thinking, ‘This is great. All I got to do is be on the right end of the giant roller coaster, and everything’s going to be fine,’” said Dan Alpert, managing partner of Westwood Capital.

...

To demonstrate how fidgety markets were Tuesday, while rumors of the statement surfaced on social media, it wasn’t until late afternoon that Iranian news agencies finally reported the statement — which ultimately only echoed what Pezeshkian has been saying for weeks.

“(Today’s market move) is not justified by the news,” said Art Hogan, chief market strategist for B. Riley Financial. “This is the market telling you it was coiled up for any kind of good news.”

That last sentence captures it nicely. 

It is next to impossible to argue that the markets are rationally pricing in the likelihood of the different scenarios and what the implications of each would be. Of the three main choices—peace, staying the course, or escalation—the best is probably the first and the worst is certainly the last, but all three are ugly, economically and otherwise. That said, if you ignore the possibility of the war staying the same or getting worse, and you overlook the economic consequences of an emboldened Iran holding the world’s economy hostage anytime it chooses, then, if you try really hard, you might understand the mentality of today’s Wall Street investor.

It will also be interesting how MAGA handles the most humiliating surrender in US history, but we've established their grasp of military history is not strong. 

McCormick: "We were horrible in Vietnam until we did Rolling Thunder Two, then we won. As soon as we do half-measures, we lose. The faster we get this over the better. If we seize Kharg island, it could be done almost flawlessly. If we have enough firepower, it would be very easy to defend."

[image or embed]

— Aaron Rupar (@atrupar.com) March 31, 2026 at 2:20 PM

"Just when I thought I was out, they pull me back in!"

 I know we've been banging this drum for a long time and I realize that I keep promising to move on, but the disconnect between market swings and the latest news and analysis continues to fascinate me.

[That and I love the mental picture of investors as golden retrievers.] 

 

Golden retriever runs 15 feet after owner mimes throwing tennis ball

[image or embed]

— Tim Carvell (@timcarvell.bsky.social) March 30, 2026 at 8:13 AM

 

 From Josh Marshall:

The U.S. is talking variously about degrading Iranian missile, drone and nuclear capacities. But if you look closely at words and especially actions the real aim appears to be to force Iran to let the U.S. out of the war with something it can call a win. “Say we won and stop fucking with the Strait and we’re all set,” the administration is basically saying. The problem is that if this scenario is basically accurate the U.S. is escalating with nothing it can call a “win” that isn’t 100% at the discretion of Iran, which now seems even more under the control of the Islamic Revolutionary Guards Corps than before the war, to offer. So what if the U.S. does a limited ground operation and Iran says, Nope, we’re still not giving you your win. What then? Full-scale invasion? As I’ve written, military planners and heads of state who are smart really want goals they can at least realistically try to achieve entirely on their own terms. So we want this piece of territory. Or we want to break this specific thing. In that case, you don’t need the other side to agree to anything. You can achieve your goals by force.

Eventually, you’ll want to make peace. But you can leave that to the other guys to worry about. You have what you want. But if your goal is entirely at the other guy’s discretion, you’ve got a big problem. And that really seems like what the U.S. is getting into now.

 

 From Patrick Boyle:

First, there’s the problem of the shut-in wells. Because the Strait of Hormuz has essentially been blocked for a month, Gulf producers were forced to stop pumping as their storage tanks hit their limits.

This is a much bigger problem than just pausing a production line. When you stop the flow of an oil well, the environment inside that well changes immediately. Without the constant heat and movement of flowing oil, the crude starts to settle and separate. It can become waxy and thick, physically clogging the tiny pores in the rock that the oil needs to travel through. In some cases, the surrounding groundwater can even seep into the oil layers, which can permanently drown a well’s productivity.

To make matters worse, many of these wells will have been closed using heavy mud or cement plugs, which are needed to keep them safe during the fighting. Those plugs will have to be painstakingly drilled out before any oil can move again. If you try to force the pressure back too quickly, you risk cracking the underground formations and ruining the field forever. It’s a delicate, multi-month restart process that doesn’t just happen overnight.

The second hurdle is the naval gauntlet. While the U.S. and Israel have spent a month bombing Iran, they’ve achieved almost no substantive gains in loosening Iran’s chokehold on the strait. This is why the Pentagon is currently sending in 10,000 additional troops, including units trained to seize and hold land. Their mission appears to be prying the strait open by force — physically taking the islands and coastal slivers where Iran hides the drones and missiles that are keeping the American Navy at a distance.

Even after those launch sites are secured, the waterway remains, at the very least, a suspected minefield. Whether or not Iran has actually laid thousands of mines, the mere risk of them creates what experts call a psychological blockade, which is just as effective as a physical one.

Reopening the strait safely requires a careful naval operation: first hunting down any remaining threats like speedboats and drones that have been harassing shipping, followed by the slow, painstaking process of sweeping the water for mines. Only then can the final phase — providing continuous naval escorts for commercial tankers — actually begin.

Finally, there’s the logistical logjam. Traffic through the strait has dropped by 97% this month. The few ships that are moving are often “dark fleet” vessels or those willing to pay the $2 million safe-passage fee that Iran is currently demanding. Insurance premiums for the region haven’t just spiked — in many cases, policies have been canceled entirely.

Ship captains and their crews are not going to steam back into a recently cleared war zone the moment a Truth Social post goes out. They’ll wait for a sustained all-clear from naval authorities and for insurance markets to normalize.

When you combine all of these issues, the Oxford Economics timeline of the strait remaining largely impassable until May starts to look very realistic.

Introducing the Trump Pressure Index

From Axios: 

How it works: Maximilian Uleer, a strategist at Deutsche Bank, came up with a "pressure index" that considers the one-month change in Trump's approval ratings, stock market performance and whether people and bond markets are expecting higher inflation. (Thanks to the Financial Times' Robert Armstrong for highlighting this index.) 

... 

The big picture: Most Wall Street analysts are expecting that Trump will do what it takes to end the war and get the Strait of Hormuz reopened to bring gas prices back down before the midterm elections.

Zoom in: Dips in approval ratings have led to other capitulations from Trump over the past year, a separate note from the bank points out.

 

With Axios, it's always difficult to tell whether they are accurately reporting on the foolishness of conventional wisdom, or just being a part of it. 

The “most Wall Street analysts” framing here definitely adds to the confusion. Does the writer see the fundamental flaw in the “do what it takes” reasoning?

We have plenty of examples by this point of Donald Trump changing policy or pulling back from some disastrous decision because of investor rebellion and potential political fallout. We do not, however, have any evidence that he can change the laws of engineering and physics, control the minds of foreign leaders, or travel in time. Nor have we seen any evidence that the man is capable of growing emotionally or taking an immense hit to his dignity and ego for the good of the country.

(Josh Marshal walks through the diplomatic quagmire in the Bluesky thread.) 

Without those things on the table, there is no “what it takes” here. Wells are being capped. Refineries are shutting down. Mines are probably being laid. Infrastructure is being bombed. The war has expanded beyond just the initial three players. The notion that this can all suddenly go away like a bad plotline, explained away as a dream sequence, is simply not in the cards.

All of this connects back to our long-running thread on investor psychology in “Trump 2” and, in a sense, brings us back full circle. Our first take was that investors were in denial, and that may be where we end up leaving it. 

It's been a while since we've had a good, laugh-out-loud Patrick Boyle video. Unfortunately, this ain't it. It is, however, one of the best overviews you'll see of the situation.  

  

I've started a new drinking game where I take a shot whenever a tech bro uses the term "Kardashev scale."

(Though to be honest, it's less of a game and more of a coping mechanism.)

There’s a great quote from an actual authority on the history of space exploration, Dwayne Day.

Over the decades, many people—most notably Carl Sagan—have noted that space enthusiasm shares many characteristics with religion. People have a set of beliefs that seem perfectly logical and reasonable to them, but which they have great difficulty explaining convincingly to those who do not share the beliefs, or have an alternative set of beliefs. They also tend to not recognize the logical fallacies in their belief systems.

Of course, it’s possible to take this analogy too far. But it also has a great deal of explanatory value. One common attribute of many religions is their ability to incorporate superstitions or iconography or traditions. Space activism does this as well. There are fetishes—imbuing certain technologies with virtually supernatural abilities—and also what might be best called incantations, or things that people say almost out of unconscious habit. The belief in helium-3 mining is a great example of a myth that has been incorporated into the larger enthusiasm for human spaceflight, a magical incantation that is murmured, but rarely actually discussed.

Fortunately, NASA as an institution has been immune from the helium-3 incantation, even when human missions to the Moon were actual policy. This is probably because the agency has its own antibodies that have effectively fought it. At the very least, if a NASA official sought to invoke helium-3 for fusion reactors in a major speech or policy document it would have to be vetted with other government agencies like the Department of Energy, and would quickly be quashed. NASA’s scientists and engineers know that helium-3 is not a justification for a human lunar program, and the continued mention of helium-3 in popular articles about the Moon—or by non-American space officials—is not going to influence whether the United States sends people there or not. But you can guarantee that talk of helium-3 will flare up again whenever the discussion turns to returning humans to the Moon, but never producing much in the way of heat or light.

If you try to follow the techno-optimists, particularly regarding the commercialization of space, there are certain terms and phrases that you will encounter with mind-numbing frequency:

Kardashev scale
Making a civilization interplanetary
Abundance and the end of scarcity
Regolith
In situ resources
And yes, helium-3, just to name a few

Sometimes these words do feel like incantations—efforts to invoke some mystical force. Other times, they simply reveal a lack of knowledge or imagination. People writing these essays and business proposals don’t actually know that much about the field, so they mindlessly repeat what they’ve heard others say.

Much of the time, again especially regarding space, I think it’s simply grasping at a very small number of straws available. The sad truth which none of them want to face is that, beyond low Earth orbit, there is little potential for a space economy in the foreseeable future, and effectively none for manned spaceflight.

This explains the weird persistence of space tourism proposals. When you get beyond the just-barely-outer-space of Virgin Galactic day trips, travel to even the closest destinations is long, uncomfortable, and somewhat dangerous. The Moon offers a nice view of Earth, but it’s barren and and exposed to high levels of radiation. The trip to Mars would be even worse. This is not something that any technological breakthrough currently on the horizon is likely to improve. There’s simply no future for that industry.

For the foreseeable future, the only viable model for manned spaceflight is government-subsidized, and the only rationale, in an age of increasingly sophisticated robotics, is national pride. Anything else ignores the laws of physics and economics in equal measure. The True Believers don’t want to hear that, but it remains an inescapable fact.

Yes, I know I said we'd drop the subject, but that was before I knew the Onion was going to pick it up.

[Click here for our last post in this thread]

You should really click through on this one. theonion.com/markets-surg...

[image or embed]

— Tim Onion (@bencollins.bsky.social) March 24, 2026 at 11:06 AM

 I suppose it was only a matter of time.

NEW YORK—In what came as a welcome shock to investors amid recent dips in the global economy, markets reportedly surged Tuesday after President Donald Trump wrote in a Truth Social Post that he’d had sex with an angel. “I AM PLEASED TO REPORT THAT OVER THE LAST TWO DAYS AN ANGEL HAS VISITED ME IN MY SLEEP AND I HAVE HAD VERY GOOD AND PRODUCTIVE SEX WITH IT,” read the lengthy, all-caps post, which with its claims that a heavenly being had done “INCREDIBLE THINGS TO [the president’s] PENIS” immediately sent the S&P 500 soaring 2.1%.

Allison Morrow ran through the latest iteration in her newsletter.

ICYMI: Over the weekend, President Donald Trump set a Monday night deadline for Iran to fully reopen the Strait of Hormuz, or else the US would start attacking Iranian power plants. But around 7:30 am ET Monday, two hours before US financial markets opened, Trump postponed that deadline by five days, citing “VERY GOOD AND PRODUCTIVE CONVERSATIONS” with Tehran toward a “COMPLETE AND TOTAL RESOLUTION” of fighting. (All-caps style his, of course.).

 

Immediately, US stock futures soared and oil prices fell. The Dow, which had been on the verge of “correction” territory last week — nearly 10% off its most recent peak — briefly surged more than 1,000 points. It ended the day 630 points, or 1.4%, higher. The broader S&P 500 gained 1.2% and the Nasdaq rose 1.4%.

 

Now, because you’re all a bunch of Readers of News, you might be thinking: Come on, finance folks, are you really falling for this again? This is the same Trump who, just two weeks ago, told CBS that “the war is very complete,” prompting a similar stock rally and pullback in oil. And the same Trump who so frequently says things and then walks them back that there’s an entire trading strategy — called the TACO trade, for “Trump always chickens out” — based on the behavior.

 

So are you guys really buying it again?

 

The answer: It’s complicated.

 

The abrupt surge was not so much a sign that Wall Street believes Trump.

 

Rather, investors saw Trump’s statement — published just as many bleary-eyed New York traders would be bellying up to their Bloomberg terminals to start their work day — as a kind of reassurance that the president’s aversion to poor market numbers will ultimately keep him from acting on his more extreme threats.

 

“There’s no real fundamental reality to any of this trading — it’s just trading Trump,” Daniel Alpert, managing partner of Westwood Capital, told me.

 

And trading Trump is a lot of trading on vibes rather than, say, the truth. Is it true that Trump had talks about a resolution of fighting this week? Iran says no; Trump says yes. Most market participants have no way of knowing who’s right.

 

...

 

“If you’re trading markets, as opposed to investing, you’re focused on what other people think, what they’re likely to do,” Alpert said. “You may not think it’s good, and you may not even be sure it’s a lie, you just think that everybody else is going to think it’s good, and so you pile in with your order … When you’ve made enough money, you dump it, and that’s it, end of trade.”

Here's the thing about a Keynesian beauty contest: you can get to a point where you can't tell the smart money from the dumb money, and when that happens, I start to worry about how well markets keep doing things like allocating resources or protecting passive investors or not driving the economy over a cliff. 

Today in 11-year-old hyperloop news

I was feeling a bit nostalgic, so I thought we could look back at the world before "the fifth mode" changed everything. (Make sure to check out the post script. It may be the best part.)

From Forbes:

Hyperloop Is Real: Meet The Startups Selling Supersonic Travel

Bruce Upbin

---

The majestic Senate majority leader suite in the U.S. Capitol was still Harry Reid's in September when he eagerly scooched his leather chair across the Oriental rug to gaze at something that, he was told, would change transportation forever.

Former SpaceX engineer Brogan BamBrogan (yes, that's his legal name) pulled out his iPad for a preview. Two business partners, the half-billionaire venture capitalist Shervin Pishevar and former White House deputy chief of staff Jim Messina, carefully studied the powerful senator's reaction. Even Mark Twain, a onetime riverboat pilot whose portrait hung over Reid's desk, eyed the proceedings warily.

"What's that?" asked Reid, sitting up, animatedly pointing at the iPad. BamBrogan's home screen showed a photo of a desert plain with dazed and dusty half-dressed people wandering around at sunrise.

"Er, that's Burning Man," the engineer responded, then clued in the 75-year-old politician to the techno-hippie carnival that takes place pre-Labor Day in the Black Rock Desert of Reid's home state of Nevada.

BamBrogan's formal presentation was even wilder, a vision for efficiently moving people or cargo all over the Southwest, to start, and the world, eventually, at rates approaching the speed of sound.

At the end of the 60-minute pitch Reid sat back and smiled. That's when Pishevar leaned in, asking the senator to introduce him to a Nevada businessman who owned a 150-mile right of way from Vegas to California for a high-speed train. Reid said he would, and they shook on it. And thus fell another obstacle in the group's fast-moving efforts to actualize what until recently had seemed not much more than geek fantasy: the hyperloop.

You remember the hyperloop, don't you? It's that far-out idea billionaire industrialist Elon Musk proposed in a 58-page white paper in August 2013 for a vacuum-tube transport network that could hurtle passengers from San Francisco to Los Angeles at 760 miles an hour. Laughed off as science fiction, it is as of today an actual industry with three legitimate groups pushing it forward, including Hyperloop Technologies, the team in Harry Reid's office. They emerge from "stealth" mode with this article, armed with an $8.5 million war chest and plans for a $80 million round later this year. "We have the team, the tools and the technology," says BamBrogan. "We can do this." The 21st-century space race is on.

[Quick aside: (Regular readers, feel free to skip this paragraph—you’ve heard it all before.) Elon Musk did not propose the technology being discussed here in his 2013 white paper. What he suggested was a high-speed train running on an air cushion in a near vacuum—a system so laughably bad that even these guys wouldn’t touch it. They did, however, keep the name. -- MP] 

It's hard to overstate how early this all is. There are dozens of engineering and logistical challenges that need solving, from earthquake-proofing to rights-of-way to alleviating the barf factor that comes with flying through a tube at transonic speeds.

[Quick aside II: If you were listing the actual "challenges" in order of difficulty, none of these would make the top twenty. It's almost as if the author was downplaying the real reasons that this would never rise beyond the level of Dubai tourist attraction, and probably not even manage that. -- MP] 

Yet it's equally hard to overstate how dramatically the hyperloop could change the world. The first four modes of modern transportation--boats, trains, motor vehicles and airplanes--brought progress and prosperity. They also brought pollution, congestion, delay and death. The hyperloop, which Musk dubs "the fifth mode," would be as fast as a plane, cheaper than a train and continuously available in any weather while emitting no carbon from the tailpipe. If people could get from Los Angeles to Las Vegas in 20 minutes, or New York to Philly in 10, cities become metro stops and borders evaporate, along with housing price imbalances and overcrowding. 

It goes on into considerable length after that though I can't recommend reading further.

In case you were wondering what became of the author of the piece.  

 

Another angle on housing costs

This is the long absent Joseph.

I was talking to Mark and one element of the emerging crisis of affordably housing we discussed was eviction law and how it discouraged some forms of housing. That same day, the Washington Post published on opinion case on the Veronica Hegens case, which I thought makes an interesting addition to the debate. 

The facts are stark:

In 2023, Hegens decided to rent out her rowhouse in Southwest D.C. so she could care for her elderly father. For two years, she rented to Phillip Graham, a participant in a D.C. housing voucher program. He died in September.

Hegens would have had no trouble finding a new tenant, except that she’s stuck with Graham’s live-in girlfriend, Gwen Broadie, who wasn’t on the lease but considers herself entitled to D.C.’s extremely generous tenant protections. When Hegens changed the locks in the fall after the woman refused to pay rent, Broadie took her to court. A judge granted a temporary restraining order allowing the squatter to stay.

Almost six months later, Hegens still hasn’t collected any of the $13,000 Broadie owes in overdue rent. As a result, she’s struggling to pay her other bills and has received a foreclosure packet from the bank. When a pipe burst, and Hegens couldn’t afford to fix it, the city government reportedly fined her more than $1,200.

To be fair, this version is the most extreme painting of the facts. Some more balance can be found here, including some pretty decent evidence that Hegens knew about Broadie before the sudden disaster. It is also the case that there is a vulnerable person here -- who has lost both her partner and her housing at once. I will say, that there is something to the idea that if you think that people should have this length of protection (it looks like Broadie will leave in June) then it may make sense to have some funds to cover these expenses. Otherwise you make small landlords reluctant to rent to people who might have issues. 

But where I think that this has the potential to really cause trouble is with the housing model of "boarding house" or "apartment hotel". Housing environments with many shared spaces require the ability to enforce rules. While these rules are often petty or annoying (think of people's deep love of Homeowners Associations), they can also make it hard for people who defy the rules to make the neighborhood worse for everyone. 

In the case of a house, things like fines are backed by assets. In the case of shared living, it's eviction that underlies fines and warnings. Sure, dirty dishes in the common kitchen are no big deal but if you create a norm that they never get cleaned then you can end up with all sorts of hygiene problems as well as an unusable kitchen space. Shared spaces are always tricky. 

Mark pointed out that this started as a comedy line on the show Silicon Valley, except it was funny because it was so true that some of the California eviction laws are a bit nutty.

I don't have a great solution here because it is complicated all around. Providing funds for landlords for tenants fighting eviction risks collusion. Fast eviction risks sympathetic people being treated badly by landlords. The current system puts a lot of legal risk on landlords and makes the risk of being a small businessperson a lot higher. This isn't one for simple solutions, and that is good. The best policy is rarely made by looking at one extreme. 

"[I]sn't necessarily a systemic catastrophe" is not the most comforting of phrases.

 As Neil Young once said about one of his albums, don't listen to this in the morning. It'll ruin your whole day. 

Patrick Boyle is back with a video explainer and it's not one of his funny ones. It is, however, one of the be overviews you'll find of the private credit situation and why so many smart people are so worried. 

The as-good-as-you're-going-to-get news?

If a fund loses 40% of its value, it’s a tragedy for the investor. But as long as it doesn’t trigger a run on the banks, it isn’t necessarily a systemic catastrophe.

The irony of the democratization of finance is that the average saver has been invited to the table just as the exits are being locked.

 The bad news? Pretty much everything else.

[Transcript cleaned by ChatGPT] 

Then there’s the sector-labeling trick. A recent investigation by Bloomberg News found that this isn’t just a few isolated incidents — it’s a systemic practice. By analyzing thousands of filings, Bloomberg identified at least 250 different loans worth more than $9 billion where software companies were being creatively relabeled. A struggling tech firm might suddenly be classified as a food-products or logistics business, allowing the fund to hide its true exposure to the tech sector and avoid triggering alarm bells about concentration risk.

Lenders also use liability-management exercises to keep the wheels turning on bad loans. Instead of admitting that a borrower is in trouble, they might allow them to stop paying cash interest and instead add that unpaid interest to the total loan balance — a practice known as payment-in-kind, or PIK debt. It’s a system of “mark-to-magic,” where the only thing being managed is the investor’s perception of risk.

But as we’re seeing with the orderly spiral in the BDC market, perception eventually has to meet reality.

 ...

The real concern for many regulators today isn’t the banks — it’s the multi-trillion-dollar insurance industry. Life insurers, particularly those controlled by private equity firms, have become some of the biggest buyers of private credit.

Insurance companies are heavily regulated. They’re required to hold a specific amount of capital against their investments to ensure they can pay out claims. Regulators generally view a direct stake in a risky private credit fund as an equity-like risk, which carries a high capital charge of around 30%.

To bypass these rules, the industry has developed a magic trick called a rated note feeder. In this structure, a special-purpose vehicle sits between the insurer and the credit fund. The vehicle issues bonds or notes, which are then graded by a specialist rating agency.

This bit of repackaging allows the insurer to treat a stake in a risky credit fund as if it were a top-rated corporate bond. By doing this, they can slash their capital requirements from 30% to as low as 10%.

The Financial Times recently described these as “black box” products. Insurers are being flooded with pitches for these feeders, often from newer, smaller managers who don’t yet have an established track record. In many cases, they say the rating agencies are grading what is essentially a blank sheet — rating the manager’s reputation rather than the actual loans, because those loans haven’t even been made yet.

One insurance executive noted that buying these notes is akin to giving a loan to a manager while having no idea what’s going on inside the actual portfolio. Insurers are essentially trading visibility and safety for yield, using financial engineering to hide the risk from regulators.

This is what Bill Dudley means when he warns about a slow-motion crisis. Because these losses are hidden from view and the liabilities are long-term, the danger isn’t a sudden explosion. It’s that by the time the situation becomes apparent, it’ll be far too late to fix the balance sheet.

 On the bright side, al t least we don't have to worry about any other potential economic crises... 

 The $3.5 Trillion Crisis No One Is Talking About

Yes, I am about to unironically post a Joe Rogan interview with Ben Affleck.

I have to admit, I’ve always been a bit mystified by Ben Affleck, movie star. Not that he’s a bad actor by any means, but his performances tend to leave me a little underwhelmed, and I often find there’s something off-putting about his interviews.

That said, he is clearly no idiot, displaying probably more talent behind the camera than in front of it. While I don’t want to perpetuate the generally unfair stereotype of the stupid actor, I’ll admit I tend to be more impressed, in terms of intellect, by those who write and direct—both of which Affleck does very well.

(An almost complete digression here, but I have to work in that great quote by Mike Nichols, which I’m about 90% sure came from a William Goldman book. When asked if he had met any truly intelligent actors, he answered, “Anthony Perkins is brilliant, and Richard Burton has something.”)

Affleck is also a sharp businessman, all of which makes him a pretty good example of “smart establishment Hollywood.” Simply by virtue of who he is, his opinion on AI is of interest here, perhaps more so because it is more clear-eyed and insightful than that of easily 98% of the people who are paid to have opinions on the subject.
 
I should probably warn everyone that the video is sponsored by a gambling site, but I doubt the readership of this particular blog is their target market.

 

  

After this interview, Affleck faced some accusations of hypocrisy when it came out that he had quietly started an AI company, which he has since sold to Netflix for potentially a great deal of money (the deal is contingent on various metrics). Personally, I don’t see the issue here. His company appears to be focused solely on using generative AI as a post-production tool, focusing on areas like editing and visual effects. Two of the examples he gave were background replacements and incorrect lighting, all perfectly legitimate uses of the technology.

I read the Financial Times because of its sharp insights, witty prose, and the fact that they have writers with names like Robin Wigglesworth

More on the approaching SpaceX IPO scam. The definitive explainer remains the Patrick Boyle video that we discussed Monday, but if you prefer a legacy media newspaper article, the Financial Times has also been on top of this story.

What could end up being the biggest bagholder exercise of all time — the Operation Overlord of jamming retail investors with an overpriced IPO — is opening up on several early fronts.

 As Bloomberg puts it:

(Bloomberg) -- S&P Dow Jones Indices LLC is considering changes to rules governing how companies join the S&P 500 Index, a move that would potentially fast-track SpaceX’s entry after its IPO, people familiar with the matter said.

The rule change could mean that billionaire Elon Musk’s space transportation and satellite company would see a wave of billions of dollars in forced buying. Funds that track the index must buy newly added stocks, and roughly $24 trillion is tied to the S&P 500, according to Bloomberg Intelligence.

 The S&P 500 news comes on the heels of Nasdaq doing everything it can to help Musk screw over its index fund investors. 

From Reuters: 

 Under Nasdaq's "Fast Entry" rule, versions of which are being considered by some other indexes, a newly listed company would be eligible for accelerated inclusion on the Nasdaq ​100 in just under a month if its market capitalization ranks among the index's top 40 current members. SpaceX is seeking a valuation of around $1.75 trillion for the IPO, one ‌of the ⁠people said, which would make it the sixth-largest company by market value in the U.S., based on the latest share prices.

Back to Wigglesworth and the FT:

A quick reminder for those lucky enough not to have to follow this saga: A series of opaque fund-raisings and mergers have lifted the “value” of SpaceX to $1.25tn, and the company is said to be seeking to sell $50bn of shares in a 2026 IPO at a valuation of $1.75tn — despite estimated revenues of only $20bn-ish and probable losses now that it as absorbed the xAI “not-built-right” money furnace.

This is not a great look for Nasdaq, as some people have commented even more archly than us. [You should check out this link. An incredible read. -- MP] But at least Nasdaq has a primary listings business that would receive a huge boon from snatching a SpaceX IPO from its big rival, the New York Stock Exchange. Why on earth is S&P DJI also seemingly flirting with a rule-bending change to allow Elon Musk’s satellites-to-AI company a quick entry?

There are various entry requirements to ensure that the S&P 500 remains the gold standard of stock market benchmarks, and doesn’t allow just any random hot stock to jump in.

...

    However, the rules should be seen in totality. The requirements for a certain free-float, liquidity, and a 12-month stint as a public company are broadly designed to prevent immature, dicey and easily-distorted stocks from being dumped onto the public via investment funds that are measured against or track the S&P 500.

These rules are now more important than ever before. Back in the day, indices only sought to reflect the market. Today — with tens of trillions of dollars in passive investment strategies that slavishly follow them and active strategies that at least have to be “benchmark-aware” — people naturally seek to game benchmarks for their own purposes. 

Elon Musk is not a particularly bright man. His reputation as an intellect is largely a product of a press corps that was eager to create the myth of a real-life Tony Stark and of a society that can’t imagine rich people not being smart.

His tremendous success looks a lot like Donald Trump’s. Both men benefit from an unearned reputation for competence, a fanatical group of followers, a lack of shame, and tremendous good fortune, first coming to power in a time that tolerates blatant, large-scale scams and corruption.

Blah blah Ginger blah blah blah

The following quote comes from the distinguished computer scientist Yann LeCun, but I've heard others in the field say virtually the same thing. The trouble is it's wrong. Clearly, demonstrably wrong.

 We want [language] to be complicated because we think of it as uniquely human; it’s what makes us humans superior to other animals.

Not only is this an unsupported assertion presented as a self-evident fact, it is obviously contradicted by examples so widely known that all of us have seen them.

When presented with any signs of language use in animals or machines, the natural human tendency is to overestimate the underlying linguistic and reasoning processes. We've already talked about the Talking Tina effect, but a far more familiar example is that of dogs. These animals can learn to recognize specific words much in the way they can learn to recognize the sound of a can opener or of a leash being taken down from a hook.

Now ask yourself: which is far more likely to happen—will a dog owner overestimate or underestimate their pet’s level of comprehension? If these people wanted to think of language as “uniquely human,” they wouldn't be talking to their pets in full sentences and frequently insisting that the animals understand or even recognize more than a handful of words.

We'll call this the Ginger effect, referring to that great The Far Side cartoon.

The very act of anthropomorphizing undercuts our sense of superiority, and yet we do it all the damned time.

LeCun wrote this comment in 2012. It wasn't convincing then but the events of the years since have rendered it laughable. 

Not only have recent breakthroughs in Natural Language Processing confirmed the Ginger effect, they have taken it to startlingly high levels, often with disturbing implications. What we've learned recently is that not only are people ready, even eager, to accept the idea that a machine can use language, they also have a tendency to project upon these machines all sorts of human qualities such as intelligence, insight, empathy, and motivations.

Depending on your tolerance for anecdotal data, we have lots of well-documented cases of people forming relationships with chatbots that are so intense as to lead to severe depression, isolation, psychotic breaks, criminal acts, and even suicide. Admittedly, in absolute terms those numbers are still fairly small, but that's not the case with people using the technology as a substitute for personal and even romantic relationships. Those numbers are alarmingly high.

It will take years of psychological and sociological research to definitively say what’s going on here, but there seems to be little doubt that many of these people—possibly most—believe on some level that they are in a relationship with some degree of emotional reciprocation with a computer.

Scientists, journalists, and pundits have spent a couple of years now on largely unproductive speculation about whether LLMs have displayed intelligence or emotions, when we should instead be talking about the far more immediate questions: what are the best applications, and what are the most worrying unexpected consequences of a massive step forward in computers’ ability to use and process language?

MAHAspital

Pretty good Saturday Night Live sketch. Rather typically, it starts out very strong, then falls into familiar bad habits, putting a hat on a hat (a phrase possibly coined by Mike Nichols) and falling back on cultural references passing for jokes.

It’s a little ironic that SNL started by satirizing this very style of humor with Chevy Chase’s Gerald Ford impersonation, which consisted of no attempt to capture voice or mannerisms, just falling down repeatedly. The show was making fun of those tired old variety show bits where someone would say something like, “Hey, it’s Ed Sullivan,” and the actor would walk out with his shoulders hunched, repeating the line “really big shoe.”

These days, that same tired old bit is usually the payoff for a sketch.

Of course, it should be noted that the weird, deconstructionist phase of SNL was very brief, arguably not even lasting as long as the original cast. For the remaining 45 or so years, the show has been mainly interesting as a cultural phenomenon, managing some clever bits and really funny moments but of more interest as a showcase for new talent and for what it told us about the zeitgeist of a particular season.

For that reason, it’s heartening to see the satire aimed at deserving targets.



 

I don’t know if any of the writers on SNL today are British comedy fans, but the sketch bears a strong resemblance to this classic from That Mitchell and Webb Look. 

Homeopathic A&E

Patrick Boyle lays out the ugly details of Elon Musk’s ultimate scam (perhaps in both senses of the word).

From the video description:

SpaceX is targeting a $1.75 trillion valuation for what could be the largest IPO in history. In this video, we examine how Elon Musk is folding a money-burning AI startup and a struggling social media platform into a rocket company to justify a price tag that defies financial gravity. From the engineering absurdity of "orbital data centers" and lunar railguns to the structural manipulation of the Nasdaq 100, we explore how low-float strategies and "fast-track" index inclusion rules are being used to turn passive 401(k) investors into exit liquidity for insiders. We look at the gap between EBITDA "vibes" and GAAP reality and the pivot from Mars to the Moon.

 

This is a slightly drier treatment than we’ve come to expect from Boyle. There are no laugh-out-loud lines or extended bits of exquisite Irish deadpan sarcasm. Boyle means business here, and his presentation is devastating.

Managing to be concise and yet in-depth, Boyle lays out the criminal absurdity of the entire thing: from the silliness of shooting data centers into orbit, to SpaceX’s suspect profitability, to the way Musk used the rubber-stamp board of directors to bail out his disastrous artificial intelligence company (currently losing a billion dollars a month), which he had previously used to bail out his even more disastrous purchase of Twitter (all of which echoed his first great scam and the basis of his fortune, leaving Tesla investors holding the bag for Solar City, but we really don't have time to dive that deep), to the various schemes and misrepresentations he is currently employing to get a trillion-dollar-plus market cap for SpaceX before the credit bubble and/or the artificial intelligence bubble pop.

The video is too pithy to summarize—just watch the damn thing—but I did want to highlight this one part.

Video transcript cleaned up by ChatGPT:

This all leads us to the question of how Elon Musk actually expects to get that $1.75 trillion price tag. The answer isn’t found in the physics of rockets, but instead in the physics of the stock market.

The most important thing to understand about this IPO is that SpaceX isn’t selling the whole company to the public. They’re likely only releasing a tiny float. The financial press is suggesting that just 5 to 10% of the total shares might be issued. By keeping the supply of stock artificially low while marketing it to every retail investor on the planet, this could create a supply squeeze before the opening bell even rings.

When everyone wants a piece of the future but there are only a few shares to go around, the price has only one way to go — and that’s up.

Craig Coben wrote about this low-float strategy a few years ago in the Financial Times, where he pointed out that a lot of recent tech IPOs had floated only a tiny number of shares — often between 7 and 10%, compared to the historical average of 20%. He used the example of Instacart, where only 8% of the shares were floated, and of those 60% went to cornerstone investors and 5% to friends and family. This left a tiny supply of shares to be traded, presumably to cultivate a sense of scarcity.

The real magic for SpaceX might happen with the index-inclusion trap. This isn’t just a lucky break for Elon Musk — it’s reported to be driven by a specific demand from him. According to Reuters, SpaceX made early inclusion in the NASDAQ-100 index a necessary condition for agreeing to list on the exchange. In response, the exchange is consulting on a new rule that would allow SpaceX to join the NASDAQ-100 after just 15 trading days, bypassing the year of price discovery required for every other company.

And it isn’t just the NASDAQ either. Bloomberg reports that S&P Dow Jones Indices is also considering historic rule changes to fast-track SpaceX into the S&P 500.

This is significant because there are roughly $24 trillion tied to that index. Maybe Elon can get on the phone to the people at Russell to see if he can get SpaceX classified as a value stock too.

If SpaceX is added to the S&P 500 and the NASDAQ-100, every passive investor will find themselves buying the stock at whatever price the supply squeeze has manufactured.

As veteran fund manager George Noble points out, the proposed five-times float multiplier used to determine its NASDAQ index weighting is “shameless structural manipulation.” It effectively turns every pension fund into exit liquidity for SpaceX insiders, where your retirement account is being forced to buy into a bubble designed by the seller.

As Noble argues, the rules are being rewritten to benefit IPO issuers and early-stage insiders, and your capital is the tool being used to enrich them.

We’ve seen this movie before. In December 2020, Tesla was added to the S&P 500 after a massive run-up in price. Since the day of inclusion, Tesla has underperformed the broader index by more than 20%. The passive investors who were forced to buy in at a huge valuation right before sales started drying up became the bag holders, while the early insiders took their victory laps.

SpaceX may be an extraordinary engineering company, but it’s being sold to the public as a financial miracle at a price that, to me, is very difficult to explain.

This could be Musk's ultimate scam in terms of scale -- the potential payout here makes the mind reel -- but also in terms being his last chance to cash in on this level. There are signs (spelled out by Boyle and already familiar to anyone who has been following the story) that things are starting to dry up, Add to that the likely scrutiny of congressional hearings if the Democrats retake one or both houses in November. Remember Musk apparently immigrated here illegally, was already forced to step down as chairman of Tesla due to fraud charges, is actively funding far right extremists around the world, has extensive ties to Epstein, and has an AI company best known for  creating nonconsensual, sexualized images of undressed women and children. Retail investors may be the dumbest of dumb money, but presumably even they aren't stupid enough to bet on an aerospace company that can't get any government contracts. 

When it comes to agentic AI, a Newhart Airline is the best is the best case scenario.

In a previous post, I discussed the idea of a steam airplane (an impressive technology that still represented a soon to be abandoned dead end) and a Newhart airline (a genuine breakthrough prematurely commercialized).

The Wright brothers' plane was the very opposite of a dead-end technology. The basic principles and design choices were all completely sound, and you can trace a fairly direct line from those first models to the passenger planes and military aircraft of two or three decades later.

That said, for all the excitement, no serious person looked at this and said this is commercially viable technology. As with Edison’s phonograph, which had also shocked the world 30 years earlier, while virtually everyone recognized this as a breakthrough, it was also clear that the technology would have to evolve considerably before it could be rolled out for widespread business or military applications.

On his seminal album The Button-Down Mind, Bob Newhart imagined a conversation between the Wright brothers and a post-war era corporation trying to monetize their breakthrough. The humor of the monologue came partly from the absurdity of trying to stack multiple passengers on the wing of the Wright Flyer or making a coast-to-coast trip taking off and landing every 105 ft, but much of it also came from the banality and shortsightedness of 60s-era corporate culture in the face of a stunning, world-altering step forward. It’s a comparison that’s, if anything, even sharper in the age of venture capitalism.

...

Are [LLMs]  Newhart’s airline—a viable and important technology that isn’t ready yet to support the commercial applications that people are trying to impose on it?

 

It is too early to say how LLM-based AI will play out, but I feel confident in saying that LLM-based agents are not ready for prime time. 
 

Julie Bort writing for TechCrunch: 

The now-viral X post from Meta AI security researcher Summer Yue reads, at first, like satire. She told her OpenClaw AI agent to check her overstuffed email inbox and suggest what to delete or archive.  

The agent proceeded to run amok. It started deleting all her email in a “speed run” while ignoring her commands from her phone telling it to stop. 

“I had to RUN to my Mac mini like I was defusing a bomb,” she wrote, posting images of the ignored stop prompts as receipts.  

... 

But Yue’s post serves as a warning. As others on X noted, if an AI security researcher could run into this problem, what hope do mere mortals have? 

“Were you intentionally testing its guardrails or did you make a rookie mistake?” a software developer asked her on X.  

“Rookie mistake tbh,” she replied. She had been testing her agent with a smaller “toy” inbox, as she called it, and it had been running well on less important email. It had earned her trust, so she thought she’d let it loose on the real thing. 

Yue believes that the large amount of data in her real inbox “triggered compaction,” she wrote. Compaction happens when the context window — the running record of everything the AI has been told and has done in a session — grows too large, causing the agent to begin summarizing, compressing, and managing the conversation.  

At that point, the AI may skip over instructions that the human considers quite important.  

...

The point of the tale is that agents aimed at knowledge workers, at their current stage of development, are risky. People who say they are using them successfully are cobbling together methods to protect themselves.

One day, perhaps soon (by 2027? 2028?), they may be ready for widespread use. Goodness knows many of us would love help with email, grocery orders, and scheduling dentist appointments. But that day has not yet come. 

  And we haven't even gotten into prompt injection.

Old Sheldon

This is an extraordinary speech by Sheldon Whitehouse on the Senate floor a few days ago. It is largely free of rhetorical flourishes, is often dry, drags in places, and is long, but riveting nonetheless. I started it assuming that I would drop out after 10 or 15 minutes, but once I got into it, I was never really tempted to quit.

Whitehouse spent a little more than half his working life as a practicing attorney, and it shows. This is constructed like a highly effective closing argument after a long and complex trial, with every thread laid down for the jury leading to a damning conclusion.

Josh Marshall had a similarly positive reaction.

The speech runs almost an hour long. But it’s worth it. There’s so many details in the speech it defies easy summary. The best overview is to think of all the ways Donald Trump was and is connecting to the Russian government and the oligarch para-government. Whitehouse then shows that Jeff Epstein is right there at almost every point of contact. It’s a mix of old information, new investigating and a pretty close analysis of emails in the Epstein Files that wouldn’t really jump out at you on their own but become quite interesting when lined up with other outside information which places them in context.

But that’s not why I’ve linked to it here. Instead, I want to highlight the Russian connection with respect to two companies—Palantir Technologies and SpaceX—and three individuals: Peter Thiel, Elon Musk, and Kimbal Musk.

Kimbal is often left out of this conversation, but he does serve on the boards of both Tesla, Inc. and SpaceX and has extensive access to the companies. He is also, as previously mentioned, extraordinarily intertwined with the saga of Jeffrey Epstein.

Add to that Elon’s extensive involvement with radical white nationalist groups in Europe, and you have the kind of stories that would normally rule out a high-security clearance—yet these companies have access to some of the most sensitive national security data imaginable.

Seems like something someone should look into.  

Sen. Whitehouse to uncover connections between Trump, Russia, and Epstein.

Strategic Airpower

We've recommended military historian and blogger Bret Devereaux before -- he's an exceptional writer and pretty much anything he posts is worth you time -- but this detailed (over 9,000 words) overview of the limits of Airpower from WWI to Ukraine might be the most timely piece of analysis you'll read about the war in Iran, despite being published in 2022.

Excerpts from Strategic Airpower 101

Before we dive in, we need to define what makes certain uses of airpower strategic because strategic airpower isn’t the only kind. The reason for the definition will emerge pretty quickly when we talk about origins, but let’s get it out of the way here: strategic airpower is the use of attack by air (read: bombing) to achieve ‘strategic effects.’ Now that formal definition is a bit tautological, but it becomes clarifying when we talk about what we mean by strategic effects; these are effects that aim to alter enemy policy or win the war on their own.

Put another way, if you use aircraft to attack enemy units in support of a ground operation (like an invasion), that would be tactical airpower; the airpower is a tactic that aims to win a battle which is still primarily a ground (or naval) battle. We often call this kind of airpower ‘close air support’ but not all tactical airpower is CAS. If you instead use airpower to shape ground operations – for instance by attacking infrastructure (like bridges or railroads) or by bombing enemy units to force them to stay put (often by forcing them to move only at night) – that’s operational airpower. The most common form of this kind of airpower is ‘interdiction’ bombing, which aims to slow down enemy ground movements so that friendly units can out-maneuver them in larger-scale sweeping movements.

By contrast strategic airpower aims to produce effects at the strategic (that is, top-most) level on its own. Sometimes that is quite blunt: strategic airpower aims to win the war on its own without reference to ground forces, or at least advance the ball on winning a conflict or achieving a desired end-state (that is, the airpower may not be the only thing producing strategic effects). Of course strategic effects can go beyond ‘winning the war’ – coercing or deterring another power are both strategic effects as well, forcing the enemy to redefine their strategy. That said, as we’ll see, this initially very expansive definition of strategic airpower really narrows quite quickly. Aircraft cannot generally hold ground, administer territory, build trust, establish institutions, or consolidate gains, so using airpower rapidly becomes a question of ‘what to bomb’ because delivering firepower is what those aircraft can do.

...

Now before we move forward, I think we want to unpack that vision just a bit, because there are actually quite a few assumptions there. First, Douhet is assuming that there will be no way to locate or intercept the bombers in the vastness of the sky, that they will be able to accurately navigate to and strike their targets (which are, in the event, major cities) and be able to carry sufficient explosive payloads to destroy those targets. But the largest assumption of all is that the application of explosives to cities would lead to collapsing civilian morale and peace; it was a wholly untested assumption, which was about to become an extremely well-tested assumption. But for Douhet’s theory to work, all of those assumptions in the chain – lack of interception, effective delivery of munitions, sufficient munitions to deliver and bombing triggering morale collapse – needed to be true. In the event, none of them were. 

 ...

That of course isn’t how it turned out. While the Luftwaffe initially began with attacks against shipping, progressing to attacks on airbases and air production, beginning in August 1940 the Luftwaffe began escalating attacks on civilian areas (the degree to which that was intentional remains contested). The British responded with bombing raids against Berlin, at which point Hitler and Göring retaliated with an intensive campaign of urban bombing which would become known as the ‘blitz.’ Hitler would claim these attacks were reprisals (Vergeltungsangriffen, ‘revenge attacks’) for the British bombing Berlin, which was frankly pretty rich hypocrisy coming from the fellow who had terror-bombed the Poles in 1939.⁴ But that progression brings an interesting distinction here between intentional strategies of using bombing to collapse morale and the reversion to civilian bombing as pure punishment. As we’ll see, it is a predictable human response when an effort is failing to attempt to punish the opponent for the temerity of not losing; this behavior is especially pronounced in personalistic dictatorships but certainly not restricted to them. Naturally bombing against civilian targets, since its introduction, has often been the means of this sort of punishment response; more broadly this kind of thing fits into the error of ‘emotive strategy,’ which we’ve discussed before.

... 

But perhaps most ominously for the theory, the Blitz didn’t seem to have meaningfully dampened British morale. Indeed, to the contrary – and get ready to hear this phrase a lot – being bombed hardened civilian will to resist. This hardly discredited the theory though, least of all among the British (or the soon-to-be-in-the-war Americans) who promptly decided to test it themselves.

...

Overall then, the promise of strategic airpower, that it could win wars entirely or primarily from the skies, turns out so far to have been largely a mirage; in about 80 years of testing the theory, strategic bombing has yet to produce a clear example where it worked as intended. Instead, strategic airpower must be one of the most thoroughly tested doctrines in modern warfare and it has failed nearly every test. In particular, Douhet’s supposition that strategic bombing of civilian centers could force a favorable end to a conflict without the need to occupy territory or engage in significant ground warfare appears to be entirely unsupportable.¹⁰ Nuclear weapons do not seem, so far, to have actually changed this; nuclear deterrence does not aim at ‘will’ in the Clausewitzian sense (drink!) but rather on altering the calculus of leaders and politicians through the threat of annihilation. In the event of an actual conflict, the public’s desire not to be nuked – which would be the key target in a Douhet-style morale bombing campaign – appears to factor very little into actual decision-making. No one checks the polls before intentionally embarking on nuclear war or in the minutes a leader might have to deliberate on ordering a second-strike.