To some, this is surprising. Presumably, as a nuclear fusion engineer, I’d be scientifically literate enough to dismiss UFOs and nonsense. To others, it’s unsurprising, in that both fusion energy and UFOs are sci-fi wish-fulfillment.

What I’ve found is that how someone evaluates the evidence for UFOs doesn’t actually determine their position on the subject. You can try to debunk videos or poke holes in testimony, but it’s not yet possible to come to a definitive conclusion one way or the other. The evidence is good enough to defy easy explanations but not quite convincing enough to support any theory with foolproof certainty.

In practice, whether someone believes ultimately comes down to their prior probability of UFOs visiting Earth.

There are three groups:

1. The skeptics, who believe that UFOs visiting Earth is far-fetched, give low probabilities of UFOs being real. For example, Neil DeGrasse Tyson claimed one in ten million.

2. The believers, who want to believe in UFOs, government cover-ups, and psionic beings, believe aliens are already here.

3. The underdetermined, who don’t think it’s possible to establish good priors around UFOs, can be swayed either way by the evidence.

I’m in the last category. I think it’s definitionally impossible to predict the behavior of “aliens,” so perceived weirdness shouldn’t be counted as evidence against them. UFOs don’t break the laws of physics as we understand them, and estimates of the probability of intelligent life in the galaxy suggest that our prior should be that it exists.

This leaves me without any good reason to think UFOs are not here, but also without a strong prior reason to think they are. Therefore, the available evidence, inadequate as it may be, should lean me to one side or the other.

The first question is whether UFOs have been seen at all. From what I can tell, as President Obama describes below, the evidence suggests that there are objects being detected that defy conventional explanation:

Before 2023, I thought these sightings were both weird and important, but we didn't know which explanation to favor.

Then, in 2023, David Grusch testified under oath that it was aliens. Because I didn’t have strong priors that it wasn't aliens, and Grusch was highly cleared and made claims under penalty of perjury, this updated my beliefs. Other credentialed witnesses also agreed, and so aliens became my leading theory.

There have been claims that Grusch is a member of a UFO religion or is reporting on a hazing ritual that has gotten out of hand, but there’s never been any documents or testimony to support these claims, just speculation. The memo about the supposed “Yankee Blue” hazing was never produced, and a FOIA request turned up nothing.

This leaves me with the following chain of reasoning:

1. There is no good reason to think UFOs are either very unlikely or very likely.

2. There is widespread evidence and testimony of flying objects moving in ways that defy explanation.

3. Credentialed people from within the US government say UFOs are aliens, with no evidence for alternative explanations.
   

Therefore, I weakly believe that UFOs are here and they are aliens.1

On Friday, May 8th, hundreds of files were published at war.gov/ufo1. Much of it was similar to what had already come out, infrared photos and videos of unusual but not obviously alien objects. However, the best bits of information do move the evidentiary needle. The highlights:

UFO Sightings by the Apollo 11 Crew

During the post-mission debrief, all the Apollo 11 astronauts agreed they saw a cylindrical UFO from their spacecraft:

This is interesting for a few reasons:

1. Neil Armstrong, Buzz Aldrin, and Michael Collins are not just any witnesses; they are among the most important people in human history. Beware of doubting their credibility.

2. The transcript is from an officially recorded debriefing conducted within a week of their return from space. All the astronauts are present and agree they saw something cylindrical in space, with no explanation.

3. The document has been kept hidden for 55 years. This could be understandable for highly classified programs, but Apollo is the highest-profile and most widely publicized technical endeavor of all time. This section of the transcripts contains nothing sensitive aside from the UFO testimony, so why has its release taken so long?

Triangle UFO in an Apollo 17 Photo

Apollo 17 was the last time a man was on the moon, and of all the Apollo UFO photos, it’s the most remarkable:

The three dots form a shape consistent with an equilateral-triangle-shaped object pointing slightly away from the camera.

That’s notable because a triangle shape is less likely to be a camera artifact. The dots in other Apollo UFO photos are random enough to be camera noise.2 However, it’s unlikely for some of the few bits of noise to happen to form the shape of a triangle.

Interestingly, triangles are a known UFO shape, and even have their own Wikipedia page. The UK government featured one as the cover photo of a now-declassified UAP memo from Dec 2000:

Similarly, why were these Apollo records hidden for so long? Historical photos with UFOs in them being hidden by the US government imply that they hide these sorts of things.

Many historians think this is causal.

I think this is causal, and suggest that he was a time traveler. Hear me out, it’s not good evidence, but still, I challenge you to name anyone better.

1. It Could’ve Been Really Bad Without Him

You might be saying, why didn’t he go back and kill Hitler? Isn’t that what time-travelers are supposed to do? Yeah, that’s what you do in this timeline, because Hitler is the worst guy we have. But in John Brown’s timeline, there could have been somebody even worse.

For example, if the US had never gotten into the Civil War, maybe it never would have ended slavery. Slavery just keeps going, and the US still becomes the world’s superpower, but the entire southern half owns slaves.

This would be really bad. It would mean millions more enslaved people in the US, but maybe other countries would never have abolition either, like Brazil (1888 ), China (1910), or Saudi Arabia (1962). Worse, the southern US may have even pressured places to maintain it for ideological reasons, much like the US did with Capitalism during the Cold War.

So now imagine you’re John Brown and you’ve got a time machine. Slavery still exists in much of the US. The country is in a terrible equilibrium, it’s spread to the rest of the world, and over time it’s become too calcified to dislodge.

Now there’s a chance to travel back to a time when the question was still unresolved and prevent centuries of suffering. A brave man would take it.

2. He Thought Like a Modern Person

There were many opponents to slavery, long before it came to a head during the Civil War. The difference was that it was a debate, or a political fight, or, if you were especially dedicated, a religious struggle that warranted non-violent resistance.

John Brown was the guy who said, “I need guns.”

I have only a short time to live – only one death to die, and I will die fighting for this cause. There will be no more peace in this land until slavery is done for.

Though people are, on average, far less violent today, most agree that slavery is worth fighting to end. On some level, we’re sad that there isn’t anything so obviously evil to really fight over, and instead have to settle for getting mad on the internet.

John Brown had what was considered an insane and fanatical position for his time, but is a mainstream position today: that slavery must end, violently if necessary.

3. He Was Right

Here’s the thing: it actually freaking worked.

His plan of arming slaves to start a rebellion didn’t work, and he was hanged, but his overarching goal of ending slavery in the US happened less than a decade later.

American slavery really was as bad as he thought, violence to end it really was justified, and his raid on Harpers Ferry really would bring it to an halt.

As expected from good time-travel, we’re in a post-Brown world and hardly even know it.

The time clock beeped in recognition, and he heard the magnetic lock on his left disengage. He bumped open the door with his hip and slid past as it opened, while simultaneously putting on his glove. He tried not to use his hands, well, really fingers, unless necessary. Those were his moneymakers, and moneymakers were hard to come by these days, especially for humans.

As he walked down the office corridors, his shoes clicked on the floor as he went. He always wore hard-soled shoes. It felt respectable. Most of his coworkers saw themselves as slightly pathetic bio-robots, but he saw himself as one of the last real tradesmen. Sure, people still made furniture in their garages as hobby projects or sold hand-made trinkets on Etsy to each other in a sort of arts-and-crafts circlejerk, but he was a professional. What he did was actually demanded by society because no one, or should he say, no-thing could do it better than he could.

Walking past the cafeteria, he glanced at the glass-walled meeting rooms. He worked the second shift and was early, so it was only a bit past 2:30 pm. The 9-to-5 office drones were having their meetings, or in their offices pretending to pretend to work.

Because even when they were doing their actual “work” and not slop-scrolling, it wasn’t real. There was no work to do. At least no knowledge work. Everyone knew this deep down, which is why they would lose their minds whenever it was said out loud.

As George walked past Engineering, he saw Carl at his desk, who immediately spun in his chair to greet him.

“George! I thought that was you I heard in the hall.”

Carl was always happy to see him. He was a middle manager and liked the chance to connect with someone he saw as an old-school, salt-of-the-earth type. George spent his days talking to a computer like everyone else, but still, it was the closest you’d get to manual labor nowadays.

“Hey, Carl. Just saw corporate strutting around V3. Nice work.”

There was “work” because it was illegal not to have it. Engineers were a licensed profession, and only humans could get licenses. Same for lawyers, accountants, welders, plumbers, you name it. And those are just the ones that were protected even before AI.

Carl stood up, beaming, and strode over to George, going for the handshake. In a practiced move, George stepped in, letting Carl grab his wrist, and gave him a one-armed hug.

He held it for a moment, knowing Carl hadn’t expected it and wasn’t the hugging type. “Congrats, buddy,” he said softly, before releasing him.

George wasn’t the handshaking type.

He continued down the corridor, thinking that the proportion of gladhanding strivers had gotten worse since the old days. It made sense: those legally required workers were hyper-productive thanks to AI, so there were fewer of them. But there were as many managers as ever, because they fight for zero-sum resources, which is a human job if there ever was one. AIs are trained to be bad at that anyway.

When George first started, he briefly had an AI boss before the managers staged a coup to ensure human input, conveniently also ensuring their zero-sum fight could continue.

Of course, the naive true believers think AI-Human teams are “actually, the best of both worlds” or whatever, though in any contest, AIs trounce human-AI teams. But sure, Becky’s HR job is the exception.

When George got to the end of the corridor, he paused a few feet from the door to the shop floor. The multi-sensor, a black ball just above the door, scanned him for FOD (Foreign Object Debris). It was a non-contact scanner only. There was still something they couldn’t do.

A short beep signaled the all-clear, and George turned the door handle with his lower wrist as he bumped it open with his shoulder, entering the work area. Now he walked on sealed concrete, and his shoes made a dull thud rather than the clip-clop from the office.

He went along a snaking foot path, bounded by yellow lines on either side so you didn’t get run over by an AGV, though honestly nowadays it was pretty hard to get hit even if you tried (a few people had tried).

As he passed the interior window of the brightly lit break room, he saw Tim, who did his job well but was pretty much a zombie at this point. He had this thousand-yard stare and hardly spoke to anyone, except for occasionally blurting out stuttering, disjointed asides that would have been normal in the right context, but the context was always outbursts between long silences.

He hung a right and got to the workers’ lockers. Annoyingly, these were the old-style ones that could have a padlock attached, so you had to grip hard and pull upwards to open them.

He opened his locker and took off his gloves. They were driving gloves. Of course, no one drives anymore, but hey, nobody in a leather motorcycle jacket rode a motorcycle either. They looked cool, and the back was mostly open, which didn’t need protection anyway. He laid the gloves down carefully so they’d air out, as one of his nightmares was getting a staph infection or something on his fingers or under his nails.

At the sink behind him, he washed his hands and used a tiny bit of moisturizer to get the slight dewiness one needed. It felt a bit effeminate to use moisturizer, but he had managed to convince himself that it was more like caring for your tools, almost like a cobbler treating leather.

He made his way to his station, with the low murmur of the other workers around him.

“smoov...smoov...smoov, ruh-ruff! ... smoov...smoov.”

The station looked a bit like a blue La-Z-boy, but with a tray table on either side. A headset with a microphone was looped over the headrest.

George put on the headset, sat down, and fastened the seatbelt. The seatbelt wasn’t because the lazyboy was going anywhere (though it could, but he refused to engage in that level of Wall-E indignity), but for safety. It confirmed you were actually seated at your station, so that a robot arm didn’t inadvertently take your head off while serving you parts. It was definitely possible to get hit by the arms as they moved, which is why, since last year, they all had to wear seat belts.

“rufffff, rufff! rugghhff...” gurgled the kid behind him. Even in this job, there were nepo hires. The kid was fine at his job, really, but it was unsettling how much he spazzed out whenever there was an out-of-spec part. As much as he disliked it himself, he was glad at least the kid was buckled in. That said, from what he’d heard, the kid might actually be one of the better-adjusted of the post-slop generation.

George clicked the foot pedal by his right foot. The whir-whoosh sound of an arm could be heard as it got up to speed, and then the quick “whoom” as it rapidly came to a halt in front of him to serve up a part.

The part that arrived was a stainless steel ring, 3/4 of an inch tall and about 1-1/2 inches in diameter. George still thought in imperial units, save for one exception—the micron.

It was a bearing race. It kept the ball bearings in place. He knew the part; only the inner surface would matter. He slid his right middle finger, his best one, onto the inner diameter and felt around.

“smoov...smoov...smoov” the kid behind him murmured. He seemed happy enough now.

“Tool chatter,” George said into his headpiece microphone. The subtle humps he felt had a wavy, repeating pattern. That meant the part was vibrating from hitting a resonance. “There’s resonance, revert any RPM changes and recheck.” The CNC lathe was going too fast, or maybe too slow. Either way, it was turning at just the right speed to shake the crap out of the parts.

His job was human finger. And, unlike his colleagues, he was a finger connected to a neural net, in this case, his brain. He’d been a machinist in a past life, so he could actually diagnose issues with parts when there were problems with the surface finish.

The other ones, like the kid behind him, were just sensor meatbags. They gave a binary “smooth” if the part was good, or “rough” if it was bad, at which point it was kicked up to one of the former machinists, like George. He would probably inspect the kid’s “rugghfff” parts soon, once the system batched the related ones and had a few more sensor meatbags confirm the reading.

Human fingers were the one thing that machines couldn’t quite replicate. Yeah, there were surface roughness sensors, but they didn’t like irregular shapes, and couldn’t fit in tight spaces, and were slow, and couldn’t actually diagnose the issue; they just spat out an Ra value in microns. Heck, even back when George was a machinist, they were around, but everyone just used their finger. They say we can detect features down to 13 nanometers.

He took his hand back and clicked the pedal by his left foot. The arm gently took the part back, and when at a safe distance, sped away with another whir-whoosh. Another part arrived. One of the kid’s probably. The arm seemed slightly snappier than usual; management was hellbent on finding more productivity gains, despite all the low-hanging fruit already having been plucked.

It turned out to be a pretty slow day. A few more parts showed up, but he could tell they didn’t really need diagnoses. It was just the system keeping tabs on him.

He took a long lunch out to meet one of his fun-employed friends. “Fun-employed” wasn’t really an ironic phrase anymore: it actually kinda looked like fun now that the stigma was mostly gone. Still, he thought of himself as a working man and got the sense that others respected it. It was like being a homesteader or something, the kind of retro thing everyone likes to think they would do, but also they’d probably hate, and so admire anyone who pulls it off. He liked that.

The autotaxi dropped him off back at the plant. “Autotaxi” always seemed like a dumb name, since “auto” was already a word for “car.” Unfortunately, Tesla was able to keep the trademark on “robotaxi,” so if you ever said it, you’d look like a fanboy. He kinda was a fanboy, but also didn’t want to enrage Becky.

He clocked back in. Left pinky. Made his way to the La-Z-boy. Clicked the right pedal.

Nothing.

Weird, he’d been out at lunch for over an hour. Surely there was something in the queue by now? Not even a test part arrived.

The kid was gurgling. “Rouuggugghf! Rooouof…”

It sounded like a lot of out-of-spec parts were going to him. Maybe every other part at this point. Was there a problem on the line?

He looked left and right, down the length of the shop at the other meatbags. He strained to hear what they were saying over the sound of the arms and kid, and the ventilation that kept FOD and dust to a minimum.

They were hard to hear; most of them spoke in almost a whisper, as if the effort of speaking would disturb their flow. “Smoov...smoov...smoov” they all seemed to be saying, visibly pursing their lips, as if going for a kiss.

Smooth. They were all saying smooth. The others were getting the usual parts, but the system was feeding the kid mostly failed parts.

That was George’s job. He checked the failed parts. Was the kid a double check? But why didn’t they end up going up to him afterwards?

“RooouuGGHHfffhh” crooned the kid, jolting George out of his head.

He waited. Silence from the kid. Then, at last, a part arrived for George, another bearing race, like before.

He felt it. Wavy. He started, “Tool chatter. Resonance...” but before he could say the fix, the arm retracted gently but insistently. George started to sit up to stay with it, but his seat belt kept him down, and before he could unbuckle it, the arm had sped off.

Quiet. No arms around. Kid fidgeted wordlessly.

He heard an arm whoosh in and serve another part to the kid, who then let out a “RoOogHHffffEEe…” in a tapering groan as he touched it.

An arm arrived at George’s station. Almost in disbelief, he probed the inside of a bearing race with his middle finger. Feed-rate spiral—he felt tiny ridges from a mismatch between feed-rate and RPM. They had adjusted the RPM without his input.

It clicked. Like hard-soled shoes on the linoleum.

It was the kid. He was diagnosing. He probably didn’t even know it, but the machine did. All it needed was someone to translate the gurgles into English. The kid was the sensor. George had just been his dictionary. But now the Machine knew the language itself.

The arm waited impatiently in front of George, waiting for a diagnosis. He stared at it, mouth open, for a moment. Then he said, “streaking, tool wear.”

Nothing happened. He stayed quiet for what felt like ages, though probably only seconds had passed.

He tried again with “Feed-rate spir...” and the arm whooshed off before he could finish. It already knew. It just wanted to be sure.

George didn’t get any more parts that day.

Back at his locker, he put on his gloves, more slowly and carefully than usual. When he checked his phone, he saw he’d gotten an email. This wasn’t an email job, so each one was allowed to push a notification.

“Please come in thirty minutes early tomorrow for a brief meeting with your shift supervisor.”

An autotaxi took him home. He didn’t sleep much that night but was in early the following day, as they’d asked.

It was 2:25 pm. He took off his glove and clocked in.

Right middle finger.

In late 2004, the USS Princeton radar ship and the USS Nimitz carrier detected anomalous radar returns off the coast of San Diego. The targets would descend from 80,000 feet to 20,000 feet, loiter for a few hours, then ascend, often at speeds typical of missiles rather than aircraft. They were also spotted with the ship’s “big eyes,” which are large binoculars.

The USS Princeton had a new cutting-edge radar system, which they thought was malfunctioning. They reset the system, which took two weeks, but they still detected the strange objects.

On Nov. 14th, 2004, Cmdr. David Fravor and Lt. Cmdr. Alex Dietrich were flying a training mission in their two-seater F/A-18 fighters, and had a Weapons Systems Officer (WSO) in the back.

The jets were sent to investigate an anomalous target, which was now detected by both ships, as well as an E-2 Hawkeye radar plane and submarine.

When the pilots arrived, they saw a “tic-tac” shaped object, a smooth white cylinder about 46 feet long, moving erratically above the water. It was a clear, calm day, and below the object, the water was agitated into whitewater.

As the planes approached, the object started ascending in an arc, which the pilots followed. They first circled each other, but when Fravor banked inwards to intercept, the UFO sped off nearly instantly and disappeared.

About half a minute later, the USS Princeton reported that they had detected it again, 60 miles away.

The two jets returned to the USS Nimitz, and later, a single-seater F/A-18 flown by Lt. Cmdr. Chad Underwood was sent to investigate and caught it on video.

Links

• Department of Defense Report (released via Freedom of Information Act)

• Fravor’s and Dietrich’s 60 Minutes interview, and Fravor testifying under oath

• Statements by Underwood

• Corroborating interviews with ships’ crew: Cryptography Tech, Fire Controlman, Air Intercept Supervisor, Aviation Tech, Supply Officer, Powerplant Specialist, Aviation Fuels Tech.

Best Picture: Costa Rica Land Survey

In 1971, Sergio Loaiza took this photo while flying over Costa Rica to survey land for a hydroelectric project.

It was commissioned by National Geographic Costa Rica, and the original film now lies within the National Archives of Costa Rica. With the chain of custody and original negatives accounted for, this is perhaps the best UFO photo ever.

Best Video: CBP Aguadilla

The Aguadilla UAP video was captured by a US Customs and Border Protection plane after it had received reports of an object detected on radar. The most anomalous part is when the object appears to enter the water going 40+ mph and then resurface, which no known airborne object can do.

Best Testimony: Secretary of Defense

The acting US Secretary of Defense said he was denied access to information on UFOs. He wasn’t told that it didn’t exist, but rather that he lacked the authority to see:

Best Counterpoint: Yankee Blue

The Wall Street Journal reported that the talk of secret UFO reverse engineering started with a hazing ritual:

For decades, certain new commanders of the Air Force’s most classified programs, as part of their induction briefings, would be handed a piece of paper with a photo of what looked like a flying saucer. The craft was described as an antigravity maneuvering vehicle.

The officers were told that the program they were joining, dubbed Yankee Blue, was part of an effort to reverse-engineer the technology on the craft. They were told never to mention it again. Many never learned it was fake … The defense secretary’s office sent a memo out across the service in the spring of 2023 ordering the practice to stop immediately, but the damage was done.

So it was just a prank, which a 2023 memo put a stop to. And they should be able to produce this, unless, of course, only UFO claims require evidence.

Not only was no memo ever provided, but when a Freedom of Information Act request was submitted to the DoD, this was the response:

The Department of Defense itself says there are no records of anything along the lines of a “Yankee Blue” hazing ritual. This is different from a Glomar response. The fact that they are seemingly compelled to lie about this should raise flags.

Objections

I address various objections to UFOs:

1. Why do UFOs act weird and sometimes crash?

2. “Extraordinary claims require extraordinary evidence.”

3. The Dark Forest Hypothesis

4. The government can’t keep secrets

5. Grabby Aliens | Eternity in Six Hours | Sandberg, Drexler, and Ord

6. 9 other objections

Also:

1. Two papers on why life might not have originated on Earth

2. How the government used disinformation to suppress reports of anomalous health phenomena, until Trump blabbed about it.

Jhanas are absorption states, often deeply pleasurable, that sometimes happen during meditation. Depending on who you ask, they can be achieved within a week-long workshop, or they are rare states that are touched only after years of practice. They go back to the foundations of Buddhism, and were described by the Buddha himself, but they have also recently gained popularity in tech-adjacent circles in the West.

I won’t go into exactly what they are or what it takes to achieve them, but rather, should they be sought out?

The Case For

Seeking awareness, stillness, and an ability to concentrate are good, and insofar as working towards Jhanas accomplishes this, that is good.

Buddhism is essentially a monastic practice; there are behavioral expectations for lay people, or “householders,” but meditation is not among them. Most people in Buddhist countries, raised Buddhist, have never meditated.

Meditation is something for monks who have devoted their lives to the path of Enlightenment. For householders, your hope is that you’ll be a good person in this life and may be reincarnated as a monk in the next, when you will have a shot at Enlightenment.

The Western approach is to be a “householder” while integrating meditation into your life to enjoy its benefits. Some of these will be instrumental (better focus, less reactive, happier, etc.) and some will be spiritual (peace, self-knowledge, insight, etc.).

Jhana-seeking because it feels good is maximally instrumental, but if the only way to get there is to skillfully meditate, then you are learning something valuable. The Jhana-seeking might be the initial impetus, but if it gets you on the path, that is good.

It doesn’t seem worse than those who start meditating to help with anxiety, anger, or any other “bad” thing. After all, in Buddhism, it’s a known mistake to see pleasant and unpleasant sensations differently; you should neither cultivate desire nor aversion.

The Case Against

Craving, sensation seeking, and encouraging desire are bad, and insofar as working towards Jhanas accomplishes this, that is bad.

Imagine your leg falling asleep.

If you have meditated even a bit, this will have happened to you. If someone said, “I won’t meditate because my leg sometimes falls asleep,” you could say that they were misunderstanding meditation.

A key aspect of meditation is that you notice, but don’t necessarily obey or inhabit sensations and feelings that arise. While you should not disregard injurious sensations to meditate, much like life, certain discomforts are part of the experience.

Both choosing not to meditate because your leg falls asleep and choosing to meditate because you’ll get Jhanas are similarly missing the point. Both are side effects of the practice that should neither be craved nor feared. Practicing craving or aversion is anti-meditation.

What Works For You

However, one of the most interesting aspects of Buddhism is the belief that you should find what resonates with you, rather than merely following traditions, scripture, or clever arguments:

Everything that happens once can never happen again. But everything that happens twice will surely happen a third time. — Paulo Coelho, The Alchemist

In software engineering, the zero, one, infinity (ZOI) rule means that you should allow zero instances, one instance, or any number of instances of something. For example, this post is allowed to have zero embedded tables (Substack lacks this feature), one subtitle, and an infinite number of likes (hint, hint).

This is a reflection of a pattern in real life, the ZOM principle, which is that there are zero, one, or many of a thing, not exactly two.

Examples of Zero, One, Many (ZOM):

• Reproductive partners: (none) asexual reproduction, (one) sexual reproduction, (many) some slime molds and fungi (10² - 10⁴)

• Dialects in a language: (none) language is dead e.g. Sanskrit, (one) single dominant dialect e.g. Icelandic, (many) large number of dialects e.g. Arabic

• Number of hearts: (none) jellyfish and sponges, (one) vertebrates, (many) octopuses (3), hagfish (4)

• Limbs: (none) non-motile bacteria, (one) has a flagellum, (many) tetrapods (4), arthropods (6 - 500+)

• Nuclei per cell: (none) red blood cells, (one) most human cells, (many) skeletal muscle (10³ - 10⁴)

• Number of spouses: (none) celibate, (one) monogamous, (many) polygamous

Poisson Distribution

This is an effect of the Poisson distribution, which is how, for a rate of something happening λ, there will be k events over a given interval.

For example, if you’re detecting meteors, and the average rate of detections is two per hour, the rate λ is two. Sometimes there would be none (k = 0), and sometimes more (k = 5), but the average is 2.

Using a Monte Carlo simulation, you can plot how many meteors you’d see in a given hour, for various rates λ:

From this chart, you can see that there are regions where 0 instances dominate (i.e., a very low rate), and regions where 1 dominates the non-zero instances. There, however, is no rate for which 2 dominates. Even when the average rate is λ = 2 per hour, in hours where meteors are seen, you see exactly 2 only a third of the time.

The upshot is that in natural distributions, exactly two instances of something should not be common. If you’ve seen two of something, expect more to come.

Exceptions that Prove the Rule

The examples that seem to be exceptions end up being enlightening:

1. Number of Sexes: This seems like 1, 2, many, but sexual reproduction only makes sense in the context of multiple gametes. Either gamete never merges (asexual reproduction), merges once (sexual reproduction), or does the slime mold thing (many merges).

2. Official Languages in a Country: This can be zero (Mexico), one (China), or many (India). However, you also get countries like Canada and Belgium with two. This is a case of two regions with their own languages, tied together for historical and political reasons. It’s an unstable equilibrium, and both have had to deal with separatist issues.

3. Gods in a Religion: Zero (atheism, some Buddhism), one (monotheism), or many (polytheism). Zoroastrianism has two, but possibly because you need to pick one to worship, monotheism seems to be an attractor: one god ends up as the good one and the other as the bad one. This is what we now have among modern Parsis. Manichaeism is explicitly dualist, and maybe it’s not a coincidence that it no longer exists.

Two is not where things settle; it's the place they're passing through. If you find exactly two of something, either more are coming, or look closer — the two are really one.

Many have over the years; here’s a list of people who have come out from the US military. However, they’re considered cranks because their claims are too weird. This, of course, leads to the tautology where whistleblowers who make claims about UFOs are cranks, because anyone who makes claims about UFOs is a crank.

Now that we have smartphones, why aren’t there videos everywhere?

They are everywhere! They’re just not that impressive because 1-3x optical zoom camera phones will not photograph airborne objects well. As an experiment, try to take a picture of a plane or helicopter with your smartphone and see how it turns out.

Why are they always seen by the US Navy?

They’re not always seen by the US Navy; they’re always disclosed by it. You can’t submit a Freedom of Information Act request to the Chinese or Russian Navies.

The speed of light is too slow

There are two ways to get around this:
1) Von Neumann probes - i.e. self-replicating probes. Cosmological time scales are more than enough to colonize the galaxy or even the universe. Humans did the same thing, colonizing the Earth while moving at only ~10 miles/year.

2) Wormholes or warp bubbles, which aren’t forbidden by physics.

Intelligent life in the universe is too rare

Life seems to lead to intelligent life, as evidenced by the fact that two very distant animal families, primates and octopuses, are intelligent. Their most recent common ancestor was a worm less than a few millimeters long.

Where are the radio waves?

With our current radio telescopes, we couldn’t detect another Earth at the nearest star. SETI is looking for a directional beam, not incidental radio traffic.

Why are UFOs an American Phenomenon?

You may have seen the map below circulating around:

However, this is just the UFO sightings for one American organization, NUFORC. In actuality, there are government initiatives on UFOs in Japan, France, Brazil, and China. One of the better-known videos of a UFO came from a Russian MiG.

Why Are They Showing Up Right Now?

It seems strange that UFOs would show up at this point in history. There are two possibilities:

1) Humans have only recently started broadcasting radio waves and detonating nuclear weapons; either one could have been a beacon that attracted attention. These signals would have had 80-100 years to reach other planets by now.

2) They’ve been visiting for a long time, but our sensors have only just caught up to detecting them regularly. We’ve only been industrial for a few hundred years, and digital for perhaps half a century.

Why would they come here?

Though there is probably other life in the universe, it doesn’t seem especially common. Humanity freaked out when we thought we found fossilized evidence of just microbes. Finding a whole new biosphere would be endlessly fascinating.

How Can I Find Some Myself?

Don’t use your phone; optical methods are bad for aircraft. Use passive radar; it’s not hard, and I’ll help.

(The mechanics of the complexity quantification are described in another paper, and it amounts to counting only the parts of the genome that can't be changed without breaking the organism.)

A plot of the complexity of various genomes vs. the time they first emerged shows an exponential curve. This makes sense, as it represents a continuous growth rate, and the paper calculates that life on Earth has averaged a ~1.8% increase in complexity every 10 million years.1

Below is a log plot of this:

On a log plot, the exponential function appears as a straight line. We can see that life on Earth aligns with the trend line, confirming that the expansion of complexity is exponential.

Interestingly, though, you can also project backward to see when the genome should have started. This requires the assumption that early organisms didn’t advance faster than later ones, but the authors argue that if anything, without template-based replication, early organisms should have advanced more slowly than they do today.

From the plot, you can see that life appears to have begun close to 10 billion years ago, with the paper reporting 9.7 ± 2.5 billion years. This is, at a minimum, about 3 billion years longer than Earth has existed.

Surprising. Even more surprising is that when Sven Jørgensen did a similar analysis, but using a form of energy density instead of genome complexity, he found that life’s expected origin was 5-6 billion years ago. This is still 1-2 billion years older than what’s possible on Earth.

Panspermia

These results support an earlier idea, panspermia, which is the theory that life originated elsewhere and was spread to Earth. Our planet is relatively young; three-quarters of planets in the universe are estimated to be older, by nearly 2 billion years on average.

The oldest planets in the galaxy are over 12 billion years old, long enough to have originated life, according to the extrapolations from the two papers.

So yes, there are planets old enough, but could life from those places ever reach Earth?

NASA tested just that, by exposing bacteria endospores to space for 1.5 years. Endospores are special, desiccated hibernation states that certain bacteria can enter when under stress and deprivation. If curious, below is a pleasingly retro 90-second clip about how bacteria become endospores:

NASA found that bacterial endospores were hardy to temperature fluctuations, vacuum, and cosmic radiation. They were only harmed by UV light. This implies that if protected from UV, for example, by being shadowed or buried, bacteria could survive the conditions of space.

But for how long? The NASA test was for 1.5 years, and most endospores survived. However, the oldest endospores ever revived were recovered from the abdomen of a bee, preserved in 25-40 million-year-old2 amber. In our region of the Milky Way, solar systems average only about 5 light-years apart, and Proxima Centauri, the nearest star to Earth, is about 4 light-years away. If endospores can also survive millions of years in space (and not just in amber), then they could manage the travel time between star systems.

Finally, panspermia requires a mechanism. This is proposed to be debris ejected from asteroid impacts, and simulations have shown particles from Earth could end up as far away as Jupiter. With enough time and enough impacts, this could end up in a different solar system.

Taken together, two independent extrapolations find that life appears to predate the Earth. This doesn’t prove panspermia, but otherwise, life would have had to arise and quickly develop immediately following Earth’s formation and cooling.

And at this point, I’m not sure which is the stranger story.

Though related, these are not the same. A paper that refutes the Fermi Paradox does not refute UFOs and sometimes actually supports the likelihood of UFOs.

(Other objections to UFOs covered in these posts.)

In fact, UFO sightings are themselves evidence against the Fermi Paradox. The paradox asks, “Why don’t we see aliens?”, and UFO sightings say, “You do, we’re right here!”

Here’s three commonly cited papers on the topic:

1. Dissolving the Fermi Paradox

This paper by Sandberg, Drexler, and Ord was reviewed on Slate Star Codex, and it delivers on its title: the Fermi Paradox largely dissolves. Using a Monte Carlo simulation to map out the alien-life scenarios that were discussed in other papers, they find that 30% of the time, you’d expect less than one civilization per galaxy.

This is a Poisson distribution, with intelligent life being very rare. Most galaxies have no intelligent life, but occasionally, due to improbably good conditions, you’d have one lone civilization. There would almost never be two in the same galaxy, so most observers wouldn’t see anyone else.1

The math checks out, so aliens are unlikely?

No. That means the Fermi Paradox is unlikely. In 30% of scenarios, there is no intelligent life in the typical galaxy, so it’s entirely possible we’re alone. Still, this leaves 70% of scenarios where there should be life to find.

• Don’t be surprised by no aliens; it’s 30% of scenarios.

• Don’t be surprised by aliens; it’s 70% of scenarios.

2. Eternity in six hours

Another paper by Sandberg, along with Armstrong, calculates that just six hours of a star’s energy is enough to send self-replicating probes across the known universe. Earth is a relatively young planet, so if probes were sent by an earlier-emerging civilization, they’d have had plenty of time to reach us.

That means if we haven’t seen any probes, there’s probably no other intelligent life in the universe. Again, the math does check out.

The paper claims: If there are no probes, there are no aliens.

The paper does not claim: There are no aliens, so there are no probes.

There are no aliens, conditional on the absence of probes. Evidence for UFOs (i.e. alien probes) is an input to this paper, not an output.

3. Grabby Aliens

Robin Hanson’s grabby aliens paper doesn’t ask about the Fermi Paradox, but rather why intelligent life seems to have appeared on Earth so quickly. He says it’s because “grabby aliens” gobble up all the other planets before life can emerge on them. Earth had life early because that was the only time it was possible; later on, all the planets will have been consumed.

His support for the claim that life was quick to show up on Earth is that most stars are much longer-lived than the Sun:

If stars of all masses were equally habitable, most habitable years should lie in planets at the far-longer-lived small mass stars. After all, 95% of stars last longer than our sun, and some last roughly two thousand times longer.

But there’s a much simpler explanation: stars are not all equally habitable—small mass stars might not sustain life. It could be that small stars don’t support rocky planets with an atmosphere, or it could be that their low luminosity (<10% of the Sun’s) won’t allow complex life.

We orbit a star in the ~95th percentile for mass. If all stars are equally habitable, it is improbable that we aren’t on one of the far more numerous, low-mass, and long-lived stars. Our big Sun implies that larger stars are more likely to support life.

Grabby Aliens on the Fermi Paradox

The Grabby Aliens Hypothesis has to address the Fermi Paradox, because if the universe is being grabbed up, we should see it in our telescopes. Hanson answers that if it happens fast enough, i.e., close to the speed of light, we won’t see it coming:

If their expansion speed were within∼25% of lightspeed, a selection effect implies that we are less likely to see than to not see such volumes.

The paper’s position is not only that there are aliens, but they “now control 40-50% of the universe volume.”

Extraterrestrial contact is therefore soon expected, and indeed, the author believes they’re already here.

The Second War of Priors was between the British Empiricists, like Bacon and Hume, and the Continental Rationalists, like Descartes and Leibniz. The power shift to the Empirical side was the Scientific Revolution, though both would claim Newton as their own.1

The Third War of Priors involves the Bay Area Rationalists, who came to prominence after correctly predicting the importance of AI. Their opponents haven’t yet sufficiently coalesced to be named.

Bayes and Priors

I call these debates “Wars of Priors” because, fundamentally, they are about how one arrives at one’s prior likelihood.

With Bayesian reasoning, you start with a prior probability of something occurring, for example, seeing a polar bear in Texas. It’s not their typical habitat, so most people will have a low prior probability of this. If a child were to say they saw one around, you would still think there probably wasn’t a literal polar bear nearby.

However, if you’re in Texas, but at the Zoo, and that child says they saw a polar bear, you’re more likely to believe them. Your low prior probability of seeing a polar bear in Texas is updated by the information that you’re at the Zoo, and so the child’s testimony seems more likely.

In this way, Bayesianism updates priors with evidence to arrive at probabilities. It would be tempting to call it a successful synthesis of Empiricism and Rationalism, but because it doesn’t answer the question of where priors should come from, it’s only kicking the can down the road.

Rationalists believe you arrive at priors through a logical analysis; in a novel situation, map out the potentialities and reason what follows necessarily from the knowns.

Empiricists believe you arrive at priors through experience. When facing the unknown, search experience for appropriate comparisons and adjust confidence based on how well examples fit.

Shots Fired

The first shots in the Third War of Priors, possibly as friendly fire, were the “monkey’s paw” of prediction markets.

Rationalists reasoned that prediction markets would aggregate the wisdom of the crowds with skin in the game, and despite cultural guardrails on betting, would deliver valuable information about the world. The project was also important methodologically for Rationality, as it served as a proof of concept that human minds together can accurately predict the future. However, the markets haven’t been especially useful, and the largest impacts are new sports gambling and insider trading. Ironically, the Rationalists’ public bet on betting epistemology went sideways.

The AI Debate

The AI debate seems to have delivered one win to each side. The Rationalists reasoned early, with little available data, that AI would be important, and they were proven correct.

However, it’s seen as a win for Empiricism that LLMs, rather than being built, are grown through brute-force extraction of statistical patterns from huge amounts of data. Every attempt to build AI from first principles failed. So though Rationalists predict from logic, Claude speaks via empirical inferences that no one (including Anthropic) understands. This creates a tension in logically modeling a non-logical system.

In the most important debate, on whether AI will kill us all, Rationalists have a probability of doom, P(Doom), that ranges from 10-90%, with 30-50% being typical. These numbers usually come from thought experiments and extrapolation from past development.

Empiricists have lower probabilities, if they accept the probabilistic framing at all, and particularly chafe at ranges above 50%. They believe high-probability claims are rationalized vibes and forward-projected just-so stories, whose reference class could be Homo erectus, but also past failed predictions of technological apocalypses, such as overpopulation, peak oil, and nanobot grey goo.

The debate will be settled in the Empiricists’ favor if AI develops slowly, develops safely, or goes bad in finite ways that allow for learning. If the Rationalists are right, though, they won’t get to enjoy their victory, and neither will anyone else.

There’s a wide range of grant sizes, from a few thousand up to apparently $175k at least (though I think awarding that much is relatively rare). If you have an interesting project, it’s worth filling out the application. It’s not very long (<1500 words), reasonably interesting to write, and they get back to you very quickly. I believe Tyler personally interviews every awardee.

The first section is your personal story, and the instructions include:

We probably don't care much about your formal education, credentials, or awards, unless they're particularly germane to who you are or your idea.

I think you should believe them here; they’re not just signaling. I didn’t include much in the way of credentials, but did explain my motivations and, in particular, told a story of ever-growing ambition.

I’m not including my bio in this post, but for your benefit, below are my answers to the second and third parts of the application.

2. A Mainstream View that I Agree With:

Here I answered with something everyone implicitly believes or demonstrates, but won’t actually say out loud. I get the feeling that this type of answer goes over well if thought-through and novel.

Amazon.com is just great. This is mainstream in that it’s widely believed in practice (Amazon has among the highest customer satisfaction among retailers), but saying this at a Brooklyn party will likely give you pushback on workers’ rights and on killing mom-and-pop stores.

However, aside from the ample consumer surplus from cost savings, there are enormous positive externalities from reliable, one or two-day delivery of hundreds of millions of goods, particularly for a hardware developer. Hardware development, compared to software development, is notoriously slow, complex, and costly. This is largely because the inputs are physical goods rather than software packages and libraries. Software engineers can instantly download a tool when needed, often at no cost. In contrast, hardware engineers have to purchase items and wait for them to arrive before they can test anything.

R&D is a continuous cycle of testing and iteration, so the speed of development is roughly the reciprocal of this cycle time. Amazon, with fast delivery and a large catalog of technical goods, has shortened hardware cycle times from 1-2 weeks to 1-2 days. In practice, this means that hardware development starts to approach the pace and expense of software development.

3. Project Proposal:

Though UFOs are reported by the Navy, and top officials like the Secretary of State, quality data is sorely lacking.

The issue is twofold: Respectable civilian organizations find the topic too stigmatized or unserious to devote resources to, while the military classifies relevant data by default (sensor data reveals capabilities).

Crowdsourcing UFO sightings data is a third pathway, and this has been attempted through mobile apps. However, a camera phone isn’t suited for distant, airborne objects, and a 2D image can only give apparent size and velocity. Further, they require line-of-sight, and the camera needs to be pointed at the object (aside from fisheye lenses). The result is generally a single, zoomed-out photo of objects of indeterminate size, distance, velocity, and acceleration.

Passive radar is the ideal solution. “Passive” meaning it uses existing terrestrial radio broadcasts as illumination, so no powerful radio transmitter or FCC license is needed. Like other radars, it uses the delay between the outgoing signal and its echoes to determine range, and the Doppler shift in the frequency to determine velocity. By directly measuring true range and velocity at 30+ miles, each radar node collects robust data across huge areas.

The nodes upload their data to an aggregation server, which generates a global map of radar hits. With enough overlapping nodes, individual objects could be corroborated by several, if not dozens, of detections. The velocity and position data could then be scrutinized for anomalous behavior, such as unusual speeds, turn rates, or accelerations.

Distributed aircraft detection networks are not new. If you’ve ever tracked a flight, you’ve used one. Flight tracking networks like FlightAware and Flightradar24 work by aggregating data uploaded by amateurs using home-built receivers collecting nearby planes’ ADS-B transponder broadcasts. The networks then sell the data via API to airlines and mobile apps. If you live in an underserved area, the networks will often even send you a detection kit for free to improve their coverage.

ADS-B flight tracking requires that aircraft broadcast their location, whereas passive radar can track anything with a sufficient radar cross-section, broadcasting or not.

A system like this is only recently possible. In 2004, a NATO researcher built a passive radar using FM radio, which required a computing cluster for the signal processing. Today, a Raspberry Pi 5 (released late 2023) is fast enough to work and enables full radar kits for under $500.

A grant would accelerate the development of a network for passive radar. I’ve been working on this full-time for the past year, and this will continue to be my primary project in the years ahead. With about $30,000 in grant funds raised and a similar amount personally contributed, most of the individual node development is finished. We’ve established a legal strategy, tested and down-selected the hardware, and polished the software for non-technical users.

4 prototype radar nodes are now live, and their data is being used to develop aggregation and tracking software. A live instance of a node’s output is here: radar3.retnode.com (red dots are radar detections, green dots are plane transponder overlays for validation).

Further funds will be used to accelerate the development of an aggregation server and a 20-node alpha test to known volunteers, beyond what my personal finances could otherwise afford. This will validate the concept, expose the system to external users, and enable scale testing.

I’ve not included the budget section, but suffice it to say, it was only modestly detailed.

If You’ve Made it This Far You Should Apply

You read 1000+ word posts on Substack. This is a positive signal you should listen to.

At the very least, try drafting a project proposal to sharpen your thinking around the topic. Then you can write the “mainstream view” answer because, honestly, it’s an interesting thought exercise. And then you might as well write a bio…

I understand this; it seems like the government is so huge and frequently incompetent that doing an adult thing, like keeping secrets indefinitely, just seems impossible. Surely some idiot will drunkenly spill the beans eventually.1

But security is better than it might seem because of the toupe fallacy: only poorly concealed secrets ever get noticed, so it appears that secrets are generally not well kept. However, when happenstance reveals programs that otherwise haven’t leaked, we get a glimpse of just how much is kept secret and how they manage it.

1. Incentives - The Snowden Files

In 2013, Edward Snowden flew to Hong Kong and released classified documents to journalists on the NSA’s global wiretapping program. From there, he flew to Moscow, lived in the Airport for a month, and was finally granted asylum in Russia. His now-wife relocated to be with him, and they’ve stayed ever since.

You’re probably familiar with Snowden’s story as the biggest whistleblower disclosure in American history. You might even think that this proves that the government can’t keep secrets.

What it more clearly demonstrates, though, is the enormous consequences of leaking classified information, and why it never came out earlier. Snowden had to permanently leave his home to live in a comparatively poor authoritarian country. He’s still there now, and if the political situation ever changes, he could well become a bargaining chip.

The US government has continued to pursue him, despite some of the programs being ruled illegal in United States v. Moalin, which shows the enormous costs the government is willing to inflict on even righteous whistleblowers.

Everyone who works around sensitive programs knows this, has been briefed extensively on it, and really tries not to do anything that even has a whiff of leaking, for fear of their lives being ruined.

2. Compartmentalization - Stealth Helicopters

In May of 2011, Sohaib Athar live-tweeted the raid that killed Osama Bin Laden:

He was not only announcing what turned out to be the location of Bin Laden, but also the existence of stealth helicopters:

Before one accidentally crashed in the raid, the existence of operational stealth helicopters was not publicly known, even to elected representatives, aside from a select few specifically cleared.

Depending on how you count, there could have been hundreds of thousands of people in on the secret before the crash, because that’s how many people are in a stealth helicopter’s supply chain.2

The way it never gets out, though, is through compartmentalization, meaning dividing the information and spreading it widely. Each entity works on a tiny portion of the whole, so it can’t identify the bigger picture. If all you can see is a fastener, coating, or shroud, you may not even be able to figure out what sort of machine it’s for.

3. Disinformation - Havana Syndrome

See my Havana Syndrome post for more on this.

Denying and obfuscating allows secrets to be kept, even when there’s evidence out in the open. This is the case with Havana Syndrome, a debilitating medical condition suffered by American diplomats (read: intelligence agents) stationed in Cuba.

It was denied for years as psychosomatic and phisologically implausible, until earlier this year, Trump blabbed to reporters that the US used directed-energy weapons to induce similar symptoms in Venezuelan soldiers.

You don’t necessarily need to keep things secret if you can keep them unbelievable.

When Secrecy Works Too Well - FOGBANK

FOGBANK is a highly classified material that is suspected to be an aerogel booster in the fission-fusion process of a thermonuclear warhead.

But if it’s so secret, why does it have a Wikipedia page? Well, the problem is that it was kept too secret.

It was first produced in the 70s and 80s, alongside the production of the warheads it would go into. Then, in the 90s, when the warheads were ready for refurbishment to extend their service life, it was realized that suitable FOGBANK material was unavailable. Not only that, but the secret of even how to make it was completely lost, with records destroyed, and personnel elsewhere, retired, or dead.

This led to severe delays and cost overruns, as the material had to be painstakingly reverse-engineered from whatever they had, and it all came out in a Government Accountability Office report.

Sometimes the government keeps a secret so well that even they can’t find it.

So which numbers to crunch? As I described in my post on the Baby Money Index, there are only two measures that matter:

• Income - Because, barring one exception, it correlates with, and thus captures, everything good (health, education, happiness etc.),

• Fertility Rate - Because it’s the one exception, being anti-correlated with income

Therefore, a weighted index of income and fertility is a metric of all good things. That is the Baby Money Index, which is income multiplied by fertility squared. Below are the rankings of America’s civilizations:

America’s Top Quartile:

1. Dakota-oming

North Dakota comes in first at 338 kiloChads; its got a low cost of living, religion, and oil, along with Midwestern values (which will show up again). For those unfamiliar, that means an obligation to the local, rather than the global, prioritizing family and community, and working hard as a matter of character rather than ambition.

South Dakota embodies these values even more so, with the nation’s top fertility rate at 2.0 children per woman, but because its economic offering is regulatory arbitrage rather than oil, it has lower incomes than its northern neighbor. Wyoming is where the culture begins transitioning from communitarian Midwestern to individualistic Western, and is rather irreligious, so it rates lower than the Dakotas, but still in the country’s top quarter.

2. The Corn Corridor

This one was a surprise to me, meaning I learned something. Nebraska is killing it at 337 kiloChads, just a hair behind North Dakota. Its lesser corn-brothers, Kansas, Iowa, and Minnesota, all similarly stack up.

This is genuinely impressive because these places don’t have an obvious geographic advantage, unless corn subsidies really are that lucrative. They combine Midwestern values, a strong service sector (Omaha is quietly a financial hub), and a low cost of living.

3. Alaska

Being a petrostate really helps.

4. Texas

The place that started the “American Civilization” discourse. Much of its success seems to be just from being better run, though Latin American immigration and the energy sector certainly help. Rugged individualism, translating to property rights, translating to legalized homebuilding, keeps housing costs low, and even in the hot market of Austin, rents have fallen.

(Hint: Cheap, abundant housing is good for the economy and for fertility).

6. Mormons

I could have said “Utah,” but we all know it’s the Mormons. It’s a pro-natalist culture, has BYU as a subsidized education and marriage driver, and a high-trust culture with a decent tech sector. They’ve done very well for themselves despite the state being mostly a desert, and much of it seems down to culture, which may be exportable.

7. NYC

Throwing a bone to the coastal elites, the New York/New Jersey metro appears to be actually good, and Delaware is contiguous and finance-y, so it is grouped in. They do better than competitor California, which is both a bit poorer and less fertile.1

The Stragglers:

New Portlandia

The lowest-scoring states are Vermont, Rhode Island, Maine, and Oregon, with New Hampshire in 6th place. It’s basically non-BosWash New England plus Portland, which all seem spiritually similar, and I will dub New Portlandia. The vibe is good for incomes, but bad for having babies.

The Rustier Belt

The particularly hard-hit Rust Belt states of West Virginia, Michigan, and Pennsylvania underperform due to middling fertility rates combined with low incomes (plus fentanyl).

Deepish South

Mississippi and South Carolina have slightly above average fertility, but that doesn’t make up for their especially low incomes. Most of the rest of the South doesn’t do much better, including Florida, which I was surprised to find in the bottom half. I assume you are all now thinking of this Albion’s Seed book review.2

The Big Empty

Rounding out the bottom quarter, New Mexico, Montana, and Nevada really don’t seem to have much in common structurally, though they belong together aesthetically.

I found the results unexpected, so now I am going to update my state stereotypes.

Alice: $5 million! For a frontier AI lab, are you joking? That’s nowhere near enough, even if you accept Deepseek’s claimed numbers; their all-in costs are at least an order of magnitude more than that!

Bob: Sorry, I meant a frontier AI safety lab and am raising $5 million. Will you help?

Alice: Oh! Yes, of course, here is $5 million dollars. You are now a significant player in the field.

I troll that AI safety has not been tried because all approaches make at least one implicit assumption: that, in some way, AI safety orgs are not like the next-closest entity — AI capabilities companies.

1) Assumption: AI Capabilities = AI Safety

The reason I say AI Safety is closest to AI Capabilities is that it is widely claimed that they are the same thing.

Anthropic is apparently actually an AI safety company, which is a relief to hear. I also found on OpenAI’s website that they assure us they are open:

So, I asked Claude to help me sort this out, but also about Millard Filmore:

Yes, Anthropic does both safety and capabilities, and I applaud them for that. However, what it is selling is capabilities with a strong focus on safety. Much like Volvo was a safety-focused car company, the thing it sold was cars, not car safety. You could not go to a Volvo dealership and buy some car safety, only an actual car.1

In contrast, you can go to a TÜV Rheinland office and buy some car safety, in the form of testing, inspections, and certifications. Symmetrically, though, they don’t sell cars themselves.

“X” and “X safety” are not the same things. In no other industry would the claim that a company is really both be taken seriously. Instead, there are “safety” or “audit” departments within a company that makes its revenue from another product. You cannot serve two masters.

2) Assumption: Entities Can Audit Themselves

Even if you believe that you can, in fact, serve two masters, you also have to believe that one can simultaneously be master and servant.

Entities are not trusted to audit themselves; all public companies are required by the SEC to have an external independent auditor. Companies have the choice of auditors amongst those registered, though in practice they generally go with one of the Big Four accounting firms, whose product is audit and consulting, not consumer services.

Currently, there is wishful thinking that Anthropic will continue to police itself, because the fact is that no one else has the resources to do so. There are evaluation organizations, but it’s an open question if safety, per se, can be done without matched resources.

3) Assumption: AI Safety Does Not Require Scale

The big lesson from AI’s dramatic early gains was the importance of scale. It may no longer be true that “Scale in All You Need,” but scale is still a necessary, if not sufficient, condition to be leading-edge.

It is currently assumed that AI safety approaches do not follow this same trend. As with AI historically, it’s thought that hard problems yield to small teams of brilliant people thinking carefully, rather than to capital and compute. Small amounts of funding are distributed among ~68 organizations, so no one has enough to pursue a scale approach, leaving the “small and brilliant” as the only option.2

The upshot is that any solution that requires scale and also must be performed by a third party cannot currently exist. Perhaps there are solutions without these requirements, but we should be honest about what the landscape now makes possible.

It’ll Be Just Another Capabilities Company?

In short, meme:

Honestly, maybe yes.

That said, now the downside risk is low. There are already many capabilities companies, so the marginal impact of yet another is small. This is different from when AI safety concerns inadvertently resulted in the first AI capabilities companies, but now that ship has sailed.

This issue could just be downstream of believing that a company can simultaneous produce “X” and “X Safety” though. Most people don’t think METR is going to morph into a capabilities company because it’s trying to do something clearly different. An org setting out to be an audit firm will have different incentives than one also building mass-market products.

An Example: Classifiers

As an example, let’s consider classifiers, or AIs that classify inputs and outputs. I’m not claiming that classifiers will solve the alignment problem; I’m just giving an example of an approach where scale may be required and therefore cannot currently be tried.

There’s evidence that LLMs won’t be able to regulate other LLMs. There’s also evidence that LLMs within the same family are especially bad at regulating each other.

If we wanted to take the AI’s regulating AI’s approach, it may be that:

1. We need classifiers that don’t come out of the labs they are intended to regulate

2. We need classifiers smarter than the LLMs they intend to regulate3

3. We need non-LLM-based classifiers

Depending on how true any of these points may be, a capital-intensive approach could be needed.

Who Classifies the Classifiers

Of course, classifiers themselves need to be made safe. Ideally, we take advantage of the high variability in AI skill and end up with something that’s really good at classifying and sucks at everything else.

It could also be that you need more classifiers, and it’s AI classifiers all the way down. Hopefully, though, eventually you get to the point where a classifier can be evaluated by a 100% human team, whose abilities and training allow them (for safety’s sake) to function in the place of AI.

That’s right, the ultimate solution: Mentats.

So in the same way, I’ve adopted an essentially Amish approach to technology, except instead of baselining from some point in the 1800’s, I stop at the mid 2000’s (aka the “oughties”)

Why then? In 2005, Reddit launched, and shortly after, Facebook released the News Feed, the progenitor of all the subsequent recommendation algorithms. Delivered to your pocket by smartphones (first appearing in 2007), upvotes and eyeballs tune doom-scrolling into today’s most popular addictive drug.

As I’ve written before, just how coca leaves are turned into cocaine and then crack, innocent raw materials can be continually refined into something totalizing. That’s why I stop at something like 2003 freebase.

Breaking the Cycle

Over the course of several years, I worked up to mostly banishing addictive technology from my life, and have retrospectively realized it looks like avoiding stuff post mid-2000s.

It may sound extreme, but think of it like being a non-smoker in the 60s. Everyone around you is smoking, and generally doesn’t even think about how it might be bad for you. It would seem especially abstemious and oversensitive to avoid not only smoking but also secondhand smoke. Still, it’s worth the funny looks; it’s not that you’re oversensitive, it’s that everyone else doesn’t realize they’re addicted to something bad.

(I have a theory that my resistance to modern social media is from being an early user of then-primitive Facebook. Like milkmaids protected from smallpox because of prior exposure to its weaker cousin, cowpox, I was exposed to a weak version of social media that produced antibodies against its modern forms.)

Smartphones

Some people have gone so far as to revert to flip phones, but the costs seem extremely high, with stories of people getting lost without GPS, and just last month at a dinner, a participant couldn’t split the bill because, without a smartphone, he had no Venmo or similar.

What works for me is turning your smartphone into not a flip phone, but something more like a BlackBerry.1 In practice, this means removing anything fun and addictive:

• Social media apps, like Twitter, Facebook, Instagram, etc.

• Video/media apps like YouTube, TikTok

• Content Feeds: Reddit, and yes, Substack

• Any browser on your phone, because otherwise you will just consume with the browser

The last one seems extreme, but it is completely necessary, and you’d be surprised how little you need it most of the time. For emergencies, I add it back and immediately remove it (via parental controls).

I keep all utilitarian apps and even some entertainment, like music, podcasts (without feeds), Wikipedia, audiobooks, and my lifeline to the culture: an RSS reader.

RSS Quarantine

RSS is an old internet technology that was a popular method of aggregating blogs starting in the 90’s. It effectively lets you make your own feed so you can aggreate content in a less addictive way. Here’s how:

1. Get an RSS reader like Reeder

2. Add feeds URLs of your favorite content. Most platforms have these, for example, just add “/feed” to any substack e.g. jehanazad.com/feed

3. If a newsletter doesn’t have a feed (most do), use kill-the-newsletter.com to turn any email newsletter into an RSS feed.

4. For social media like Twitter posts, use rss.app to make a feed of users you want to follow.

5. For links, add them to the reader app and read them there (since you no longer have a browser)

Browser Blocks

Getting your smartphone under control is 90% of the battle because it’s just so available.

Regarding laptops, try to leave them in your workspace or workplace. You probably don’t need it when you’re not working, and if all you have with you is your sanitized smartphone, your doom-scrolling options will be limited. In your workplace, you have psychological anchoring and possibly nosy colleagues to keep time wasters off your laptop screen.

You’ll still be tempted, so add one of the feed eradicators to your browser. I find these more effective than blockers, because they make the websites boring, not forbidden.

(As many of us have learned from dating, this is much more psychologically effective for giving something up.)

E-Readers

Some of you workaholics will not want to leave the laptops alone (despite also being sucked into unproductive activities). A nice stopgap here are e-readers, in particular the reMarkable, which you can write on and is about the size of a sheet of paper. It’s ideal for PDFs and web articles.

Anything that needs reading can be saved there for after-hours reading, and it’s particularly good for scientific papers.

What about LLMs?

I, like the Amish, am allowed to ask for answers from a higher power.

The idea is that the universe is like a “dark forest” where one must hide and stay silent, or risk being eaten by a more powerful creature. If ever detected, it means near-certain death because others will locate and destroy you preemptively, out of fear that if found, you would do the same to them.

This thus proposes that there are many extraterrestrial civilizations, but they all choose to keep quiet, and the ones that didn’t are no longer around to hear from. The concept has been popularized by the excellent sci-fi novel The Three-Body Problem, and it plays a prominent role in the plot.

However, though working as a narrative device, in real life, Dark Forests are surprisingly hard to find. For example, listen to this short clip from the Amazon:

What you’ll immediately notice is that the jungle is not always silent, as anyone who’s spent time in dense woodland can attest. So it seems even dark forests aren’t really “dark forests.”

That said, perhaps “dark forest” is just a convenient ecological metaphor. Once the entities are civilizations, rather than organisms, does the dynamic emerge?

But Why Namibia?

If you think that Dark Forest is the stable equilibrium that civilizations reach in response to each other, you’ll have to answer the question: Why does Namibia still exist?

The theory claims that rational actors will destroy any new civilization they find before it can become a threat. But if this is true, shouldn’t NATO have annihilated Namibia (or any other small nation) by now?

There are two reasons why NATO doesn’t go on a rampage, and here, both apply:

1. There are values and norms against arbitrarily destroying other countries.

2. Namibia is not, even to the tiniest degree, considered a threat.

NATO doesn’t consider Namibia even remotely a threat. Yet, Namibia is technologically closer to NATO than any interstellar civilization would be to NATO.

Unless you think NATO is secretly harboring a desire to end Namibia, and is only held in check by modern values, this is evidence that a civilization considers preemptive annihilation only for near-peers.

So, when this has been the case, how has it played out?

Near-Peers

There have been many near-peer rivalries among human societies, but most of them have occurred before preemptive total annihilation was straightforward. You do have examples such as the Mongols massacring whole societies, but that was part of conquest, and if the victims had surrendered, they would have been spared.

The closest example is during the short post-war period, 1945-1949, when the US had atomic weapons, but its new rival, the USSR, did not.

During this period, there were serious discussions of preemptive war with the Soviets before they could develop their own nuclear arsenal, and a group of top Air Force generals tried pushing this viewpoint onto civilian leadership.

However, even then, when an adversary with doomsday capability was about to emerge, this never gained traction among US leaders, and they didn’t even think an intentional war was likely.

The Best Example is Submarines

Submarine warfare is probably the closest example of a real-life Dark Forest dynamic, and militaries do go to great lengths to silence their vessels as much as physically possible.

First-mover advantage is enormous with torpedoes, and a direct hit is nearly certain to be fatal. A military vessel in wartime can assume it will be fired upon if detected, so it must remain undetected or destroy any adversary it finds immediately.

But to bring this back to extraterrestrial civilizations, we have to ask if our relationship to them would be more like 1) two submarines at war, or 2) late 40’s US and USSR, or 3) NATO and Namibia today.

Humbly, I’ll suggest that it probably isn’t even like NATO and Namibia, it’s more like homo sapiens and termites. Which means we’re considered no threat, and are safe as long as we don’t mess up the floorboards.

1. Hailing is More Predictable - Other rideshare apps show an expected wait time is really an optimistic estimate, based on what drivers might like, while Waymo can be sure because it’s both the dispatcher and the driver.

2. They’re better drivers - They see 360 degrees, are never distracted, and won’t drive like a maniac to earn slightly more. Waymo claims they’re 5-10x safer.

3. (Usually) No Worry About Star Ratings - You won’t get dinged unless it’s something really egregious, as opposed to say, making a driver wait.

4. People can be Annoying - Drivers of course (music, come-ons, smells), but also me (conversation, luggage, smells). Now, I can be as annoying as I like without consequence.

Aside from the benefits to the users, though, it’s possible everyone will benefit because they could have a positive effect on the commuters’ bane: traffic.

Reducing Total Cars

On the one hand, if self-driving rideshares become incredibly cheap on average, you’d expect people to use them more often. However, rideshares can be a substitute for car ownership, so as uptake increases, individual car ownership should decrease.

Overall, I’d expect a reduction in total cars due to much higher utilization rates. The average car is parked about 95% of the time; if any significant portion of this time was used for ridesharing, you’d need many fewer cars overall.

Congestion happens when road usage outstrips capacity, with most of it occurring during peak times. Reducing the total number of cars reduces the total possible peak traffic.

Less Parking, More Road

On urban roads, between 25 and 45% of space is used for parking, much of which could be converted into travel space. Additionally, in some neighborhoods, 45% of all traffic is of cars just circling for parking.

Fewer cars parked means more lanes and less useless circling.

Universal Price Signal

Rideshares are priced according to supply and demand, and users are relatively sensitive to pricing. This means that during peak hours, we’d expect fewer trips as the price rises, unlike private vehicles, which are mostly fixed-cost.

However, if drivers’ personal vehicles can be easily switched to ride-share mode (as Tesla has proposed), they will be foregoing a lucrative rush-hour opportunity to earn extra money. Thus, even many private vehicles will be affected by rush-hour surge pricing, and private drivers may plan their own trips around this.

Similarly, delivery services would need to make the same trade-off, and so may organize to avoid peak times. Effectively, higher car-utilization rates caused by self-driving add a new variable price to every vehicle on the road, disincentivizing peak-time driving.

Avoided Accidents

The US Department of Transportation estimates that traffic accidents are responsible for 25% of all congestion. As mentioned, if self-driving cars can reduce accidents by 80-90%, then 20%+ of congestion can be avoided.

Most self-driving accidents are still human-caused, just by the other driver. If most cars on the road were self-driving, however, the accident rate may drop even further.

Coordination and Other Efficiencies

More efficient driving from Waymos doesn’t just mean fewer accidents, but also:

• They know the road perfectly, so they won’t miss a turn, slow down when lost, or get in the wrong lane.

• Aren’t distracted, so won’t be slow on a green light.

• Are polite, so will let you in and not cut you off.

These benefits are before even considering futuristic city-wide coordination or caravan trains on the highway. We’ve only just begun to see what's possible with greater adoption and coordination.

So, after all the tech-dystopia we’ve seen emerge, I’m thankful for the tiny bit of tech-utopia that is a Waymo.

The most plausible decision however is to slightly lower your level of ambition. Consider a few of the core scenarios.

If the aliens go rogue on us and end it all, the efforts you might be making now will have been for naught.

If the aliens are here to cap the level of human achievement, for instance to keep us on Earth and prevent us from exploring the galaxy, yet without harm, you also can scale back your ambition a bit. You do not need to invest so much capital in supporting the space program. Most of your more local ambitions however should remain untouched. You might even become more ambitious in keeping the Earth a safe place, since escape hatches are now less likely. Alternatively, you might think the aliens are our “saviors of last resort,” but that too probably makes you less ambitious.

A more general Bayesian update is simply that human efforts, in the broader scheme of things, have lower relative marginal products than you might have thought. The aliens apparently have lots of powers, at least if they managed to get here. That too militates in favor of lowering your ambitions. Conversely, if you start believing we are the only intelligent, agentic beings in the galaxy, arguably, you should increase your ambitions. There will be fewer outside forces to stop, limit, or reverse your efforts.

If UFO reports are true, and they represent a technologically advanced alien intelligence, in my experience, they should actually make you more ambitious.

1. Incredible Physics is Possible

Accepting his premises, Alien UFOs imply travel across light-years and are reported to display incredible performance, like:

• Instant acceleration and turns, yet g-forces don’t crush the craft

• Flight propulsion without reaction mass or aerodynamic control surfaces (i.e., no rockets, turbines, rotors, or wings)

• Medium-independent travel: Moving in and out of water at speed, flying supersonic without skin-heating or a sonic boom

If this level of technological mastery is physically possible, we’ve been underestimating possible performance envelopes by orders of magnitude. For example, if you’ve been working on solar sails or fusion rockets for interstellar travel, this should update you towards more ambitious Alcubierre drives and wormholes.

The same could be said for anyone working on energy, materials, and, of course, fundamental physics. The physical ceiling of what’s possible is taller than expected, so you should be reaching higher.

It also applies to anyone downstream of potential technological developments. Just as AI should make a lawyer more ambitious because it’s a force multiplier for their work, introducing advanced power and materials technology should similarly make an AI scientist more ambitious. However, materials and energy go into everything (indeed, they are the only physical inputs), so it’s hard to think of any builder who wouldn’t need to update.

(It is possible aliens could want to cap our technology, but as of yet, there are no reports of this, outside of particular weapon systems. If anything, it’s more frequently claimed that they have actually advanced our own technology, including intentional gifts.)

2. Your Priors About the Impossible Should Drop Globally

I, like most people, used to think that UFOs were ridiculous and in the same category as Bigfoot and the Loch Ness monster. Nutcases believed in them because they wanted it to be true, to feel special, and to enjoy confirmation bias.

Now I feel that I owe the UFO nuts an apology. I’m not claiming they’ve been definitively proven right, but evidence has come out that they don’t belong in the same category as other conspiracy theorists.

This means I was globally miscalibrated. My model of the world was too narrow, so I should broadly update to be more open-minded about crazier things.

And if crazy things are more possible than we realize, then we should inch closer to being crazily ambitious. So maybe reconsider that biological immortality, space datacenter, or brain-uploading project.

3. AI Less Likely to Kill Us

If you’ve been less ambitious because you think there’s no point in building if AI might kill all of us soon, Aliens should revise that possibility downwards for two reasons:

1. Safe superintelligence is possible, demonstrated by aliens who have not killed us. They might not be AI themselves, but given how advanced the rest of their technology is, they probably have AI too.

2. They may be, as Tyler said, “saviors of last resort” and reportedly do disable our nukes, though not our data centers, implying they regard nukes as the danger.1

4. We’re In a Competitive Environment

When people are in competition, they work harder if the game seems winnable, and decrease effort if they think they’ll lose. It’s implied that because humans are so far behind aliens, we are uncompetitive and so should put in less effort.

However, we don’t know how many players there are in this game or how they’re organized into factions. There may be exactly one intelligence in the galaxy, but there probably are not exactly two.2

As an example, during the Taiwan Strait crisis, they were weaker than either the PRC or the US, but by playing off the geopolitics, Taiwan still exists as an entity. If we were to enter a galacto-political game, we should do so from as strong a position as possible, as the matter could even be existential.

So if you ever begin to consider that UFOs could be real, I hope it makes you more ambitious. I speak firsthand, because it’s been the single biggest ambition amplifier I’ve ever had: pushing me from wanting to be a “success” to wanting to be a Copernicus.

Exactly 36 years later, on January 3rd, 2026, the US repeats the excursion and captures Nicolas Maduro. However, this time, the entire mission was completed in only hours and without a single American casualty. Why the difference?

This time, the US had the “Discombobulator”:

When first asked about the special weapon that enabled such an lopisde outcome, the president said:

Trump: Well, I’m not allowed to talk about it.

Though immediately goes on:

Trump: Everything was discombobulated. Practically, a shot wasn’t fired…and when we came, they couldn’t do anything.

Intereviewer: …Nothing worked, including humans. It knocks the humans out.

Trump: Well, let’s put it this way, we lost no equipment…in a very bad environment.

Later, White House press secretary Karoline Leavitt retweeted an account of the event, which corroborated the use of a disabling super-weapon:

Security Guard: …At one point, they launched something—I don't know how to describe it... it was like a very intense sound wave. Suddenly I felt like my head was exploding from the inside. We all started bleeding from the nose. Some were vomiting blood. We fell to the ground, unable to move.

Interviewer: And your comrades? Did they manage to resist?

Security Guard: No, not at all. Those twenty men, without a single casualty, killed hundreds of us. We had no way to compete with their technology, with their weapons. I swear, I've never seen anything like it. We couldn't even stand up after that sonic weapon or whatever it was.

So the US has a sonic or energy beam weapon that makes people vomit blood.

Just the US, though, right?

Havana Syndrome

From late 2016, US diplomats1 in Cuba started experiencing unusual health symptoms after an acute episode of pain, pressure, and a loud, localized sound. In almost all cases, symptoms lasted for months and included dizziness, headaches, and cognitive problems.

Named after where it first appeared, Havana syndrome (also known as anomalous health incidents) is “not recognized as a disease by the medical community.” A 2024 review listed possible causes as head trauma, viruses, or mass psychogenic illness.

Still, those affected have some recourse, as the 2021 HAVANA Act “authorizes the Central Intelligence Agency, the Department of State, and other agencies to provide payments to agency personnel who incur brain injuries from hostilities while on assignment.”

But where should there be a HAVANA act if there may not even be a Havana syndrome?

It’s all part of the gaslight.

The Gaslight

Gaslighting is trying to get someone to question their sense of reality. The term has become popular over the last decade, though honestly, it’s often overapplied to ordinary disagreements.

That said, gaslighting does happen, and Havana syndrome denial increasingly looks like an example.

The critical differences between a disagreement and a gaslight are that 1) the perpetrator knows you know the truth, and 2) they try to manipulate, rather than argue you out of believing it. In this case, there are 4 parts:

1. Deny: Initially, Havana Syndrome was rejected entirely. The State Department blocked the CDC from accessing medical data for nearly a year. The 2023 intelligence community assessment found that five of seven agencies judged foreign involvement was “highly unlikely,” and officials said there was “no credible evidence” that an adversary had these weapon capabilities. Simultaneously, though, the CIA quietly shut down its Havana station after two more agents were affected.

2. Stigmatize: The FBI’s Behavioral Analysis Unit labeled the incidents “mass psychogenic illness” without interviewing affected personnel. A classified report by the JASON advisory group was selectively leaked as evidence that embassy sounds were crickets, though the report itself noted most recordings didn’t overlap with symptom onset and left directed energy as a possible cause. Senior CIA officials privately dismissed victims as exaggerating for financial gain.

3. Retaliate: Those who went public faced consequences. Mark Lenzi, diagnosed with brain injuries at the University of Pennsylvania along with his family, was ordered to undergo a psychiatric evaluation after he warned colleagues of the danger, despite leadership knowing of similar cases. He was denied promotions, blocked from overseas assignments, and, as late as 2024, given a pre-written cable to end his tour for “personal reasons.” At the CIA, a whistleblower alleged potential obstruction and witness tampering, and dozens of officers filed formal complaints over the agency’s delays.

4. Hush: HAVANA Act payments, up to $204,000 for qualifying brain injuries, without admission. Claimants are compensated for injuries that are not formally acknowledged to exist.

The Science Catches Up

While acute symptoms reported in Venezuela and by Havana Syndrome victims are often described as accompanied by loud sounds, the microwave auditory effect is well established and was first observed among radar workers during World War II. Microwaves can cause popping or clicking sounds in someone’s head, with the proposed mechanism being slight brain heating that generates a pressure wave transmitted to the cochlea.

Still, not everyone believed the microwave mechanism was physiologically plausible, so in 2024, a Norwegian scientist decided to prove the theory wrong by building a microwave pulse generator and directing it at his own head.

As the Washington Post reports, “it didn’t go well.” He ended up with symptoms consistent with Havana syndrome, and these results were sent by the Norwegian government to the CIA, prompting visits by officials from the Pentagon and White House.

Directed Energy Weapons

The US has a whole suite of directed energy weapons, from lasers to particle beams to their current favorite, microwaves. The disclosed microwave weapons don’t cook people’s brains, but they do heat skin and destroy electronics.

The US “discombobulator” shutting down both troops and equipment, along with the known perceived sonic effects, suggests a microwave weapon. Combine this with the Norwegian scientist’s unfortunate self-experiment, and we have evidence that:

1. Havana syndrome is caused by microwaves.

2. The US knows about these weapons because it has its own.

3. Havana syndrome dismissals are more likely concealment than confusion.

So it was nice of the Norwegians to share their research with their ally, but the US is way ahead of them on “discombobulation”—and doesn’t want you to know.

Why the Cover-up?

There are a few possible explanations for why the US would want to suppress the reporting of Havana syndrome.

The first is simply to conceal the existence of this technology. Yes, one of their adversaries has these capabilities, presumably Russians operating in Cuba, but smaller regional powers (e.g., Venezuela) may not be aware, and this preserves the element of surprise (e.g., Venezuela).

Secondly, acknowledging the attacks creates response obligations at both the agency and national levels. Intelligence agencies would need to implement countermeasures across all sites, and there would be public pressure on the US to retaliate, thereby escalating the situation. The CIA is constantly waging small-scale, covert warfare against various actors, and it may be inconvenient to have its dirty laundry aired.

Finally, suppressing reporting on Havana syndrome also denies adversaries feedback on the effectiveness of their own efforts. Whether one’s efforts are working is among the most valuable intel, and historically, great effort has been made to obfuscate this. One of the most famous examples is Operation Double Cross, which misled Nazi rocket targeting and likely saved thousands of English lives.

However, we now all know, because despite not being “allowed to talk about it,” the president has described US capabilities in front of new cameras. Is this from a shift in US policy? I suspect it was more of a spontaneous Trump moment, both entertaining and confirmatory.