They stopped asking Bill to fill out the form

Friday, April 10th, 2026

I somehow missed Andy Hertzfeld’s –2,000 Lines Of Code back in the day:

In early 1982, the Lisa software team was trying to buckle down for the big push to ship the software within the next six months. Some of the managers decided that it would be a good idea to track the progress of each individual engineer in terms of the amount of code that they wrote from week to week. They devised a form that each engineer was required to submit every Friday, which included a field for the number of lines of code that were written that week.

Bill Atkinson, the author of Quickdraw and the main user interface designer, who was by far the most important Lisa implementer, thought that lines of code was a silly measure of software productivity. He thought his goal was to write as small and fast a program as possible, and that the lines of code metric only encouraged writing sloppy, bloated, broken code.

He recently was working on optimizing Quickdraw’s region calculation machinery, and had completely rewritten the region engine using a simpler, more general algorithm which, after some tweaking, made region operations almost six times faster. As a by-product, the rewrite also saved around 2,000 lines of code.

He was just putting the finishing touches on the optimization when it was time to fill out the management form for the first time. When he got to the lines of code part, he thought about it for a second, and then wrote in the number: -2000.

I’m not sure how the managers reacted to that, but I do know that after a couple more weeks, they stopped asking Bill to fill out the form, and he gladly complied.

(Hat tip to Gaikokumaniakku.)

Because of the way that CRAM lights up the night sky and how much noise it makes, it is hard to keep its operation secret

Wednesday, April 8th, 2026

David Hambling explains why U.S. Gatling guns are not stopping Iran’s Shahed drones:

The Centurion C-RAM (“Counter Rocket, Artillery and Mortar”) was first deployed in Iraq in 2006 and is a land-based variant of the original Phalanx CIWS (“Close In Weapon System”) used by the Navy since 1980. It is the last line of defence when urgent action is needed to prevent casualties. The land version is a self-contained unit weighing around 24 tons and costing something over $4 million.

As the name suggests, C-RAM was introduced to protect bases against rocket, artillery and mortar fire. It has an integrated radar which tracks incoming projectiles as well as the stream of rounds fired by the 20mm M61A1 Gatling gun to put them on target.

The cannon is the same as that carried by F-15 and F-16 fighters. Its distinguishing feature is its phenomenal rate of fire, the six electrically-powered spinning barrels selectively firing 3,000 to 4,500 rounds per minute – that is 50 to 75 per second — producing a sound like a buzzsaw, often rendered as “Brrrrt.”

While the Navy version fires solid tungsten projectiles, CRAM uses the M940 Multi-Purpose Tracer – Self-Destruct round. This weighs 3.5-ounces/99 gram and consists of a tungsten cone to punch through the target skin, and a body which explodes in a dense mass of fragments inside the target. “Tracer” means the round produce a visible glow, and in operation the stream of projectiles appears as a bright ribbon reaching out towards the target. Automated tracking shifts aim until that ribbon overlaps the target.

The “Self-Destruct” part means that the rounds automatically explode at a range of around 2,300 meters if they miss the target, an effect also highly visible on videos of CRAM engagements. CRAM is a point defense system placed to protect high-value assets. If a drone strikes just couple of miles away the operators can only watch.

CRAM has a magazine of 1,500 rounds. This sounds like a lot, supplying 30 one-second bursts of 50 rounds each. But it is only enough for 10 two-second bursts at the higher rate of fire. It reportedly takes some 30 minutes to reload CRAM manually with 15 boxes of ammunition each weighing around 60 pounds.

Each M940 round costs $168, so a 150-round burst costs around $25k, comparable to the price of a Shahed.

Because of the way that CRAM lights up the night sky and how much noise it makes, it is hard to keep its operation secret.

[…]

What we do know is that the reported success rate against rockets and mortar shells in Iraq was reportedly 70-80% with an average of 300 rounds per engagement. These are relatively convenient targets because, although they are moving fast, they come in on a very predictable trajectory and they descend from high in the sky making them easy to pick out on radar.

Tackling drones may be more difficult. Being made of composite material rather than metal, they may have a small radar reflection. And Shaheds can fly at extremely low level, sometimes at under 100 feet with a flight path that takes them between buildings. The level of background clutter will make radar tracking challenging.

[…]

Unlike rockets and artillery, Shaheds do not need to fly on a predictable path. Some of the Russian versions automatically carry out evasive maneuvers when they sense a threat. As the videos show the stream of rounds can be seen and potentially evaded. This dodging would at the least increase the number of rounds needed for a kill. Russian Shaheds are also accompanied by numbers of low-cost Gerbera decoys to distract and deplete defenses. Iran does not yet seem to have either capability.

[…]

The U.S. Army only acquired about 20,000 rounds of M940 this year, which one weapon could burn through in five minutes of firing.

Many people in this country believe secret weapons are proper public news

Tuesday, April 7th, 2026

Fleet Tactics and Naval OperationsFleet Tactics and Naval Operations explains the trends and constants of technology:

DePuy, in unpublished papers, accumulated evidence that in ground combat the impact of a new weapon upon the outcome of a war usually has been local and almost always has been transitory. He believed that a technological surprise by itself never has won a war on land, but that technology accompanied by a tactical revolution has. Napoleon’s tactical use of mobile artillery was revolutionary; the field artillery itself was not new. It is ironic that the Germans exploited tanks so effectively with their Blitzkrieg, for one of their victims, the French, possessed more and better tanks, and another, the British, had invented them. In these instances the new tools, artillery and armor, were no secret at all. In contrast, when tanks were a surprise, first used in substantial numbers by the British at Cambrai in World War I, the British forces achieved local successes but could not exploit their new weapons. Some argue that the British prematurely squandered tank technology before the accompanying tactics had matured.

[…]

Because there are fewer big battles at sea, the potential for decision by technological surprise is greater. At least one weapon is comparable in decisiveness to cryptanalysis, which wrought the great increase in Allied scouting effectiveness: it is the kwi-suns, or turtle boats, of Korean Admiral Yi Sun-Sin, which in 1592 helped win two decisive battles against the Japanese at Pusan and in the Yellow Sea.

Another secret weapon sprung long after its prewar invention was the Japanese Long Lance torpedo. As late as the summer of 1943, the U.S. Navy did not know exactly what the Japanese weapon was or why it had been so effective. The Long Lance had been developed in the early 1930s, and Japanese cruiser and destroyer men had trained extensively with it. American scorn for Japanese technology takes much of the blame for the U.S. Navy’s overconfidence at the start of the Pacific war, which was almost as foolhardy as German and Japanese overconfidence in the immunity of their own ciphers.

Then there is the atomic bomb. Although it was not specifically a naval weapon and not numerous enough to be regarded as tactical, the bomb was the shocking weapon that administered the coup de grace to Japan in 1945. The science and technology took four years to develop, and only two bombs were built.

Is it possible to keep the development of an “ultimate weapon” a secret in peacetime? Evidence suggests that it is not possible, at least not in the United States. Many people in this country believe secret weapons are proper public news.

[…]

Here are some examples of weapons, mostly naval, that brought disappointment in World War II:

  • Magnetic influence mines. Germany introduced them against shipping in the estuaries of the British Isles. They were effective, but they were used prematurely. As a result, they turned out to be vulnerable to countermeasures.
  • Magnetic exploders in American torpedoes. Developed before the war, they worked badly and were a great setback to U.S. operations. In a short war, American torpedoes would have been an unmitigated disaster. The British and Germans also experienced early problems with their sophisticated torpedoes.
  • Proximity fuzes. For much of the war they were restricted to use over water out of fear that the Germans would recover one and adopt the technology against U.S. strategic bombers.
  • Night fighters. These were highly effective, but there were too few of them to be decisive.
  • Submarines. They had a powerful impact, but their role against warships was well recognized before World War I.
  • Sonar. This was a crucial response to the submarine, developed in secrecy. It was not enough to neutralize the threat.
  • “Window,” the strips of aluminum foil used to jam enemy fighter-direction radars. The Germans had window early in World War II, but they delayed its application until the Allies used it in the bombing of Hamburg in July 1943. Both sides appreciated the fact that window was a doubled-edged tool of war—of value to both sides.
  • Jet aircraft, V-1 and V-2 missiles, and snorkeling submarines. All arrived too late in the war to have much effect.

Here are some reasons that new weapons, whether secret or known, do not always deliver what they promise:

  • Production limitations, as with magnetic mines
  • Testing limitations, as with torpedo exploders
  • Great complexity, requiring skilled operators and integration into fleet tactics, as with radar and night fighters
  • Great simplicity, threatening adoption and exploitation by the enemy, as with window
  • The risk of failure after introduction, as with the U.S. magnetic torpedo
  • Exaggerated expectations, as with sonar
  • The penalty for maintaining secrecy during a lengthy period of development, as with Nazi Germany’s secret weapons

[…]

There are many examples in which important improvements in combat capability have been hidden. One is the rifling of gun barrels. Another is the improved fire-control systems in dreadnoughts. New engines barely can be detected from an aircraft’s appearance, but they can vastly change the plane’s performance. Changes in computer reliability or cryptology or in scouting systems in outer space are invisible, at least to an amateur observer.

Karl Lautenschlaeger asserts that the most important characteristic of the Soviet Oscar-class submarine was not its great size, but the likelihood that its missiles were guided by space-based sensors.

Submarines that depend on acoustic stealth are in a continuing competition to operate more quietly than the enemy; the quieter they become, the more “invisible” they are.

[…]

Vannevar Bush once said that the unity of decision under a totalitarian regime was a recipe for making colossal technological mistakes, whereas the prevalent confusion of decision-making in a democracy was more efficient. He could not have anticipated the tortuous system of procrastination that characterizes modern American defense procurement.

[…]

Usually more than one piece of technology is required to create a revolution. Sail and cannon together replaced the oared galley. Steam power alone was not enough to replace the ship of the line; it took the steam engine, the screw propeller, and the metal hull all together, which in turn made possible the big gun and the marriage of rifling, breech-loading, and an effective fire-control system. Big aircraft carriers were nothing without powerful aircraft engines to lift bomb-loads worthy of the name, and big aircraft required powered elevators, catapults, arresting gear, and the science of long-range navigation over water.

[…]

Even the Polaris submarine, the embodiment of a naval revolution as neat and swift as we are apt to see, would not have arrived without the inspired marriage of two technologies, nuclear propulsion and solid-fuel rocketry; and the work of two great technical leaders, ADM Hyman Rickover, USN, and VADM William F. “Red” Raborn, along with Arleigh Burke, a Chief of Naval Operations who understood warfare, politics, and the value of swift action.

[…]

It is impossible to design the perfect weapon for large-scale production and employment without practicing with it; even then, it takes three or four generations of hardware before a weapon realizes its full potential.

People who understand complex systems also understand the importance of minimising that complexity wherever possible

Saturday, April 4th, 2026

Explaining tech debt is, the evidence would suggest, impossible:

Like many I’ve seen archaic systems where something that should take an hour might take a week. The trouble, as anyone who’s ever been in this situation can attest to, is that non-technical managers invariably fail to understand the problem.

Of course the standard claim at this point is that engineers are just bad at communicating. They simply need to re-express their ideas around refactoring etc. in terms of ‘value to the business’ or some such. But I’ve never seen this work. Instead I’ve seen the same thing over and over: people who understand complex systems also understand the importance of minimising that complexity wherever possible; people who have never understood a complex system in their life never ever grasp this and cannot be convinced.

Imagine you’re a dumb non-technical manager. You think coding is basically magic. Sometimes when you ask the engineers to add a feature (which is like casting a spell), they give some weird story about how they could cast the spell quickly, but it’s better to do so slowly to prevent their magic becoming impotent and making future spells take longer. This is obviously unconvincing; it sounds like something slackers would say. So why are we surprised when managers who think like this fail to grasp the importance of controlling tech debt?

To this day, most warships have little staying power

Wednesday, April 1st, 2026

Fleet Tactics and Naval OperationsThe development of Germany’s V-1 and V-2 and the US’s atom bomb led the Navy, Fleet Tactics and Naval Operations explains, to develop missiles — first Regulus and then Polaris — that could deliver warheads at very long ranges with reasonable accuracy.:

The new missiles were called “strategic” weapons because, much like strategic bombers, their purpose was to destroy an enemy’s means of waging war.

[…]

This led to a bitter rivalry between the newly created U.S. Air Force, which claimed the mission as its own, and a recalcitrant Navy, which saw difficulties with intercontinental bombing at the time and balked at the hidden costs of maintaining bombers at fixed bases far forward in host countries.

The Navy proposed delivering nuclear bombs from carrier-based aircraft, arguing that the mobility that ships offered would enable the bombers to fly shorter distances and would be less vulnerable than land-based airfields.

[…]

In the 1950s, with remarkable energy and technological acumen, the Navy developed and deployed Polaris missiles—and long-range submarines to carry and fire them—arguing that the undersea craft constituted a more stable and survivable deterrent than bombers and land bases because they could not be pinpointed for attack.

[…]

Some of the early Soviet missiles were cruise missiles, fitted with nuclear warheads and designed to be fired by Russian warships—submarines, surface ships, and long-range land-based aircraft.

Their targets were to be American surface ships, particularly aircraft carriers. Since detonating even one nuclear weapon in the vicinity of a ship was certain to destroy it, staying power derived from armor, compartmentation, damage-control techniques, and large displacement would have little value.

Using antiaircraft guns in an effort to shoot down an attacker would be useless if a nuclear weapon were designed to detonate when the warhead was hit.

The U.S. Navy developed surface-to-air missiles (SAMs) to destroy a bomber or missile far enough away for the ships to be able to survive. Indeed, Talos, Terrier, and Tartar SAMs—all defensive weapons—were the Navy’s first substantial venture into guided-missile technology.

The tight defensive formations of World War II no longer were appropriate; adjacent ships would be incapacitated by the massive explosion and poisonous radiation.

Designers initially intended that SAMs would cover several ships at the same time, employing the World War II tactic of defending your neighbor while defending yourself. SAMs were expensive, however, and any one ship could only carry so many. They had to be delivered accurately because commanders could not fill the sky with them by the hundreds the way 40-mm and 20-mm shells were expended in World War II. If anything, SAM distribution against incoming aircraft or missiles had to be coordinated so that commanders could rely on an efficient system of assigning targets to individual ships.

In time the formations were loosened even more and spread out in dispersed configurations. One such was a “haystack” disposition, developed so that enemy bombers could not easily locate the vital ship—the carrier—especially where commercial shipping resulted in the generation of many radar contacts. The fleet’s prime targets were supposed to disappear like needles in a haystack.

[…]

The modern U.S. Navy is a victim of outmoded nuclear war thinking. To this day, most warships have little staying power. One or two hits with modern missiles such as an Exocet or Harpoon will put most warships out of action.

To survive an attack and continue to perform a task, a modern American warship depends heavily on reduced susceptibility—avoiding detection and carrying the kind of technology that will enable it to prevent incoming missiles from hitting at all.

[…]

The Vietnam War contributed to loosening up American formations because warships were able to stand off at sea to deliver ordnance while they themselves were relatively safe from attacks.

[…]

History’s most profuse application of cruise missiles has been against tankers and other commercial ships in the Persian Gulf. The attacks started in May of 1981 and continued for seven years, until mid-1988, ending a year after U.S. intervention that provided protective escorts for ship traffic.

[…]

French arms sales equipped Iraq well to carry out air-launched Exocet missile attacks.

Seemingly, missiles had been used between 257 and 261 times, or in about 80 percent of all Iraqi attacks on commercial ships.

[…]

Only a quarter of the ships hit were destroyed; large tankers proved to be the sturdiest and most resilient.

The so-called Tanker War constitutes by far the biggest campaign against shipping since World War II.

[…]

Estimates show that by 1986 the tonnage damaged beyond economic repair already had reached some 20 percent of all Allied merchant ships sunk during World War II.

Navias and Hooton estimate that less than 1 percent of the 800 to 1,000 ships that entered the Gulf each month were hit—about the same overall total as the fraction of sailings lost in the Battle of the Atlantic, although not as bad as the worst of that period, when up to 20 percent of merchant traffic was lost. 4 Also reminiscent of the Battle of the Atlantic, there was a remorseless buildup of shipping losses in the Gulf until the United States responded to pressure from the neutral states there and started to convoy reflagged Kuwaiti tankers.

[…]

Like torpedoes, tactical missiles were conceived and developed to attack warships.

[…]

Broadly, the carrier battle groups of the U.S. fighting fleet could not offer direct protection for tankers sailing up the Persian Gulf; only individual convoy escorts could fend off attacks by the Iranian threat, which in this instance comprised land-based aircraft and a flotilla of assorted small coastal combatants. But the security of the escorts depended upon air cover, present or prospective, from the American carriers standing outside the Strait of Hormuz.

Safe transit through the Gulf waters also depended on mine-clearance operations, carried out largely by European countries, which had joined the effort by the mid-1980s.

[…]

The axiom that “a ship’s a fool to fight a fort” is tempered by the caveat that in order to influence events on land, navies must either circumvent or destroy the enemy’s ability to send land-based aircraft and missiles over the coastal seas.

[…]

In the first cruise-missile attack on a ship, during the Arab-Israeli War of 1967, an Egyptian salvo of four Soviet-made Styx missiles sank the Israeli picket-destroyer Eilat. In 1970 the Egyptians conducted what was in effect a live-target test of the ability of the Styx to home on targets smaller than a destroyer; they fired four missiles and sank an Israeli fishing boat, the Orit. In the Indo-Pakistan War of 1971, India successfully employed nine Styx missiles against Pakistani warships and merchant vessels, some of which were in port.

Next came the best wartime laboratory for study of missile combat—the Arab-Israeli War of 1973. The two sides exchanged 101 Styx and Gabriel missiles in five separate battles with devastating effects on the Syrian and Egyptian flotillas and no harm whatsoever to the Israelis.

After that came the South Atlantic War of 1982, in which Argentina achieved well-publicized results with air-launched Exocets and, for the first time in combat, with land-launched missiles as well. In the same war, but less well-known, Royal Navy helicopters launched Sea Skua air-to-surface missiles at two Argentine patrol boats, sinking one and severely damaging the other.

In February 1991, during the Persian Gulf War, two Silkworm antiship cruise missiles (ASCMs) were launched from a land site in Kuwait, aimed at the USS Missouri (BB 63), which was bombarding Iraqi positions with 16-inch shells. Although the Silkworms malfunctioned and did not inflict any damage, the incident is noteworthy as the first and only time in a war that a ship-fired surface-to-air missile has shot down an ASCM, the honor going to a Sea Dart fired by HMS Gloucester.

[…]

Whether in terms of incidents, damage achieved, weapons fired at a target, or cost of ordnance expended, missiles and missile warfare dominate modern combat at sea.

Briefly, large, defenseless commercial ships showed very high hit-probabilities, but the damage by no means has been uniformly fatal. Hit-probabilities against warships that defended themselves were far lower, yet substantial and usually with devastating effect. Perhaps the most interesting and alarming statistic is the number of successful attacks on defendable ships, such as HMS Sheffield, that failed to protect themselves.

Disconcerting in its tactical implications is the case of the Atlantic Conveyer, hit and destroyed in the South Atlantic War. Two Exocets, launched by a pair of Argentine Super Étendard jet fighters, homed on HMS Ambuscade, one of the screen ships in the Royal Navy formation stationed east of the Falklands. The Ambuscade launched chaff, which distracted the ASCMs and saved her from harm. But once the Exocets had flown through the chaff cloud they searched for another target and found the SS Atlantic Conveyor, destroying the ship and the important cargo on board. By saving herself, Ambuscade failed in her mission to protect the other ships in the formation.

A further irony is that the Argentine pilots actually had hoped to hit the aircraft carrier HMS Hermes, which was also in the formation and had a flight-deck full of Harrier jets.

[…]

If there is a new lesson from the South Atlantic War, it is not that warships are vulnerable to missiles, but that aircraft armed with bombs cannot compete against warships that are equipped with modern defenses.

[…]

Large, protected ships such as battleships are valuable partly because they can take hits and continue fighting.

[…]

Ships must have warning in order to deal successfully with missile attacks. In modern sea warfare the outcome between two forces armed with missiles will often be decided by scouting and screening effectiveness before any missiles actually are launched.

[…]

In World War II it took a lot more punishment to sink a warship than to incapacitate it. Comparing tables 7-1 and 7-2, the average was five times as many 1,000-pound bombs and two or three times as many torpedoes.

[…]

Beall’s conclusion is that vulnerability is proportional to the cube root of displacement. Since displacement is roughly proportional to the three dimensions of length, beam, and draft, the cube root reduces the measurement to one dimension. The Brookings study concluded that a hit by one large warhead would incapacitate a modern warship up to 300 feet long, and another similar warhead is required for every additional 100 feet. By that measure, the Proceedings article concluded that to kill (not sink) an aircraft carrier would require seven missile hits, three missile hits would kill an Aegis cruiser, one or two were required for a frigate, and one would be enough for a patrol craft.

[…]

The results are disconcerting to the tactician because all of them show the flatness of the kill curve. In fact, the BuShips data indicate that only a few more hits were required to sink a battleship or carrier than to sink a heavy cruiser. Can modern designs be effective against cruise or theater ballistic missiles to keep a modern combatant in action? The classified 1990 study by NSWC Carderock asserted that a great deal can be done; moreover, the toughening will come at only a modest increase in cost. Whether this is so, the Navy’s current inventory is mainly in large warships that are potent offensively but depend almost entirely for survival on reducing susceptibility by a layered defense of combat air patrols, SAMs, and hard-kill and soft-kill point defenses. Even more important, American warships depend for survival on out-scouting the enemy and attacking him not only effectively, but decisively first. These are tactics suitable for a fleet in the open ocean. The tactics will lose their efficacy in littoral waters.

[…]

Since a large ship enjoys economies of scale, it will carry more fuel, ordnance, aircraft, or Marines than several smaller ships of the same total cost. The analytical conclusion is, therefore, “bigger is better.” The important disadvantage of a large, supposedly efficient ship is the hazard of putting many eggs in one basket. Indeed, the Beall, Humphrey, Schulte, and BuShips studies all reflect a diseconomy of scale. If a 60,000-ton ship carries twenty times the payload of a three-thousand-ton ship but can only take three or four times as many missile or torpedo hits as a small one before it is out of action, then that is a substantial disadvantage offsetting its greater payload.

[…]

Coastal navies use land installations to scout and attack from as safer, cheaper, and more resilient than large warships. Their fighting ships are small and heavily armed. They depend for success on stealthy attack and surprise by out-scouting the enemy. Their ships are short-legged with austere habitability, because they can sortie to perform brief, stressful tasks.

[…]

Borrensen puts the operational aim of a competent coastal defense in full strategic context: a coastal state will not attempt to defeat the navy of a maritime state, but instead will endeavor to inflict sufficient pain on that navy in an extended campaign so that the enemy will not think the game worth the candle.

[…]

Joergensen offers a pointed warning that the U.S. Navy is not sufficiently configured or practiced to defeat a coastal power without severe losses. The implication of both articles is that it will not take a high-technology coastal defense to inflict pain and suffering on a high-technology, blue-water navy.

[…]

The U.S. Navy’s principal responsibility is to safeguard the oceans almost anywhere, though not everywhere at once. The other side of the coin is to deny movement of enemy shipping and the means of war—an easier mission that usually comes with the territory when the first mission is achieved.

Weakness that comes from disregarding these two missions invites another country to build up a blue-water fleet to move into the power vacuum.

The A-10 wasn’t designed for drones

Saturday, March 21st, 2026

The A-10 Warthog is the ultimate drone hunter for the modern battlefield:

In an era where cheap, slow-moving drones like Iran’s Shahed-136 (and its Russian Geran-2 cousin) are flooding the skies flying at just 115 mph while costing as little as $20,000–$50,000 apiece traditional air defenses are bleeding money dry.

[…]

Here’s why the A-10 is built for this mission like no other platform.

1. Speed & Loiter Time: The Perfect Match for Slow Drones

The Shahed-136 cruises at a leisurely ~185 km/h. The A-10’s top speed is ~420 mph, but its real strength is its cruise and loiter speed around 300–340 mph at low altitude. It was designed to loiter for hours over the battlefield, giving pilots plenty of time to spot, track, and engage slow-moving targets that fast jets would blast right past. (Helicopters like the AH-64 Apache can do similar work, but the A-10 is faster, has far greater range, and can cover more ground without needing to land and refuel as often. In saturation attacks, one Warthog can patrol a wide area and knock down drone after drone on a single sortie).

2. Firepower: Cheap, Precise, and Devastating

The A-10’s legendary GAU-8 Avenger 30mm cannon is overkill for tiny drones one burst would shred a Shahed into confetti. But the real gamechanger is the APKWS II (Advanced Precision Kill Weapon System). These 70mm laser-guided rockets cost roughly $20,000 each a fraction of an AIM-9 Sidewinder or AMRAAM. An A-10 can carry dozens of them, turning the jet into a flying rocket truck with massive magazine depth. The FALCO software upgrade (cleared on the A-10) gives the rockets a proximity fuse and laser guidance perfect for subsonic, low-maneuverability drones. Pilots use targeting pods to paint the target the rocket does the rest. And if the drones get too close, the cannon is always there as backup.

3. Built Like a Tank

The Warthog’s famous titanium “bathtub” armor protects the pilot from ground fire up to 23mm. In drone-hunting missions, it can operate low and slow in contested airspace where fragile fighters or expensive stealth jets would be too vulnerable or too fast to be useful. Self-sealing fuel tanks and redundant systems mean it can take hits and keep flying exactly what you need when hunting cheap drones that might be escorted by basic air defenses.

4. Cost-Effectiveness That Actually Makes Sense

This is the killer argument in the drone age. Shooting down a $20k Shahed with a million-dollar missile is economic suicide. The A-10 flips the script: cheap rockets, reusable platform, and the ability to stay on station for extended periods. Analysts have called it a “sweet spot” platform faster than helicopters, slower and more persistent than F-16s or F-35s for this specific threat.

[…]

The A-10 wasn’t designed for drones, but the drone wars have found the perfect aircraft for the job. Its combination of loiter endurance, low-speed agility, massive cheap firepower, and legendary toughness makes it the ultimate drone hunter. While fifth-generation fighters chase high-end threats, the Warthog can stay low, stay long, and swat Shaheeds (and their kin) out of the sky for pennies on the dollar.

The autonomy software wasn’t supposed to be enabled until the boats were suitably far out to sea

Tuesday, March 17th, 2026

Project Maven by Katrina MansonOne day in June, 2025, a group of self-driving military boats lined up for a test event at Channel Islands Harbor Marina, a mile north of Port Hueneme Naval Base:

The boats were part of the Replicator program, which was then two years old and less than two months away from the official deadline to deliver thousands of maritime and air drones. The pressure was building.

Things began with support vessels towing autonomous boats out to sea; the drones’ engines were set to neutral and their autonomy mode turned off. The test focused not so much on the vehicles themselves—known as global autonomous reconnaissance crafts, or GARCs—as on the software that allowed them to function on their own. Two separate companies, the defense contractor L3Harris Technologies and Anduril, had made autonomous operating systems for the boats. That day, Replicator was testing GARCs that ran on each company’s product.

As a safety precaution, the autonomy software wasn’t supposed to be enabled until the boats were suitably far out to sea. But one drone running L3Harris’ system suddenly lurched forward. Its autonomy mode, which had somehow turned on, required it to keep a distance of 80 meters (262 feet) from all other objects. The robo-boat sped away, still tethered to the towboat. It alternately accelerated and decelerated, then started crisscrossing in front from port side to starboard side in a semicircling action.

The captain of the towboat had no way of taking over control of the automated vehicle, whose erratic movements caused his own vessel to capsize, throwing him into the water. Still tethered to the towboat, the drone turned back toward it and—for reasons that remain unclear—started advancing at rapid speed.

A captain towing a separate GARC saw what was happening and raced toward the scene, positioning his vessel between his floating comrade and the advancing drone. A third towboat pulled the captain out of the water, and he escaped without serious injury. It had been just three minutes since the drone had gone rogue.

A safety investigation soon diagnosed the problem: An operator on the dock had inadvertently sent a message to the drone remotely disabling the safety lock meant to prevent it from switching into autonomy mode—a classic “fat-finger mistake.” A spokesperson for L3Harris said in a statement that the operator who caused the issue didn’t work at the company and that its software had “demonstrated its ability to control a mix of uncrewed platforms, payloads, and commercial technologies even if they were produced by different manufacturers.” A physical button was added to drone boats to block such accidental commands, and the boats were tweaked to prominently display the mode under which they were operating. Rival companies would start sharing safety lessons.

But the incident illustrated problems that still existed with the Pentagon’s drone strategy and couldn’t be resolved with the addition of another button or two. Replicator had still not progressed to the point that its creators were comfortable putting live ammunition on an unmanned vessel, let alone sending one into a scenario where it would be expected to coordinate with other vehicles or carry out a specific attack plan. The program did manage to deliver hundreds of drones by the August deadline, but it fell far short of its initial goal.

The evident value of such a submarine tanker for refueling oil-burning surface ships in wartime has kept this concept alive

Sunday, March 15th, 2026

For decades now it has been possible to wield sea power without a navy, and the current situation in the Strait of Hormuz is demonstrating just how vulnerable ordinary shipping is to modern missiles and drones — which got me wondering about the practicality of a submersible oil tanker:

In the early ’70’s there was great interest in economically transporting oil from the large oil finds in the Arctic to the markets in the U.S. and Europe. Either pipelines or marine systems seemed feasible. But, bringing the oil out by submarine tanker — on a year-round basis — appeared to be the most cost-effective approach. Consequently a design study of an Arctic submarine tanker was conducted by General Dynamics’ Electric Boat Division to demonstrate the practicality of this approach.

Though this project never materialized, the evident value of such a submarine tanker for refueling oil-burning surface ships in wartime has kept this concept alive. A battle group of nonnuclear powered carriers and escorts, capable of being refueled from a submerged tanker — on any course and at relatively high speed — would greatly increase transit speeds while ensuring a vital underway replenishment capability, particularly in a conventional war environment of enemy ocean surveillance satellites and enemy long range cruise missiles.

The submarine tanker designed by Electric Boat was most economically sized to carry 250,000 deadweight tons of oil. With a length of 1,000 feet, an 80 foot draft, a submerged displacement of 360,000 tons, an operating depth of 1,000 feet and a sustained speed or 18 knots, this giant submarine could transit efficiently under the Arctic ice, through the restrictions in the Northwest Passage and readily avoid icebergs in Davis Strait.

Since this tanker could and probably would load its oil from a bottom loading pad, its total cycle of operations could be secure from enemy observation. Although designed for peacetime commercial use, it could be considered an asset to be activated as a naval auxiliary in wartime. Thus, an enemy campaign against such a vital element in U.S. logistics should have little chance of being successful. With the U.S. advocating a “forward offensive maritime strategy,” the security or its critical refueling elements “under the gun” of enemy homeland defenses even moreso emphasizes the submarine tanker solution.

When the attractiveness of this submerged commercial tanker for wartime naval operations became evident, a further design study for the underwater refueling system was conducted. A probe and drogue system similar to that used for aircraft refueling from tanker aircraft was shown to be feasible — the submarine positioning itself under the surface ship and pumping oil up through ·its telescopic probe into a bottom drogue on the surface ship. The safety factor in this method of refueling was particularly good because of the stability of the submarine under all sea conditions and the little movement of a surface ship drogue, positioned at its center of flotation.

The vessel is essentially a large, rectangular tanker-like ship hull with the long internal cylindrical pressure-resisting hull, usually associated with a submarine, centered within the outer rectangular hull. The central hull contains the living and control spaces, pumps and auxiliaries, and the propulsion machinery. Except for the free flooding ends of the ship, the remainder is filled with oil cargo in the loaded condition and sea water in the ballasted condition. The variable cargo tanks on either side are provided to compensate for the difference between density of sea water and the oil.

The propulsion is by twin screws driven by steam turbines. Steam is supplied by a pressurized water reactor, similar in design to those presently in use for commercial electric power generation. The nuclear steam supply system produces steam for the two propulsion trains, each plant developing 37.500 SHP at the propeller for a total of 75,000 SHP. The sustained sea speed would be 18 knots.

By the end of World War II, the Germans were using “milk cow” submarines in this role — but submarines have come a long way since then. In particular, modern submarines travel more efficiently while submerged, not less, because they’re designed primarily for undersea travel, where they encounter no wave-making resistance.

They also encounter no air, which is why “true” submarines only became practical with the advent of nuclear power. But there are non-nuclear forms of air-independent propulsion (AIP), like fuel cells:

Fuel cells are not new. They have undergone significant technological improvements from when they were first considered for submarine propulsion by Germany in the 1950s. The principle of producing power is straightforward; hydrogen and oxygen gas react to produce water and an electrical current. It is the reverse process of electrolysis, where a current is sent through liquid water to split the bonds between the oxygen and hydrogen atoms. Through engineering optimization, enough electrical power can be harnessed from this reaction to power a variety of loads. Current uses include cars, buses, remote cell phone towers, and forklifts. The German Navy already has a hydrogen fuel cell–powered submarine class, the Type 212, first launched in 2005, and variants it sells abroad to countries such as Italy and Singapore.

The Gotland-class submarine, a Swedish boat, is the most prominent example of the extreme stealth of non-nuclear AIP submarines. During a joint wargaming exercise in 2005, it tactically sank the USS Ronald Reagan (CVN-76) several times. It was virtually undetectable by all available antisubmarine efforts.

While powered by a Stirling engine, the concept and application of the Gotland-class AIP system are the same as for others. Stirling engines and other forms of non-nuclear AIP, while quieter than nuclear, are louder and less efficient than fuel cells. There are no mechanical parts in the main fuel cell system such as in combustion driven engines. Fuel cells offer the lowest noise levels because almost no sound is produced by an electro-chemical reaction. The only components in the engine room that could contribute to the sound signature are the compressors and pumps for fuel, water, and cooling.

Yet, cooling requirements for fuel cells are much lower than combustion and nuclear because of the low operating temperature of 100°C for proton-exchange membrane fuel cells. Conversely, nuclear-powered submarines need extensive cooling and vibrational dampening because of high operating temperatures, requiring several large coolant pumps and bulky, complicated mechanical systems such as steam turbines and reduction gears.

In addition to the fuel cells, there are advanced lithium-ion batteries on board AIP vessels that can power the electric motor at higher speeds with no loss of acoustic fidelity. Without the nuclear reactor, there also is a smaller infrared heat signature and no radiological trace. There is a significant stealth advantage to fuel cells that lowers the detectable range of the vessel.

Fuel cell AIP submarines do not have the nominally infinite endurance of nuclear-powered submarines; however, they can remain underwater for much longer than alternative AIP options such as closed-cycle diesel generator, Stirling, and MESMA (a French steam turbine). Fuel cells are significantly more efficient than diesel engines, thus requiring less oxygen fuel per kWh of energy produced. Diesel-electric boats have a max underwater time of a couple of days because of battery limits. Fuel cell AIPs can last weeks underwater and have a range of up to 2,000 nautical miles. Further, by forward deploying these vessels in ports that are close to their respective operating areas, the ratio of time on station to transit and refueling time is increased.

They wouldn’t be immune to mines and underwater drones, of course, but one thing at a time.

Most managers optimize for being informed

Tuesday, March 3rd, 2026

Anish Moonka summarizes the key points of Elon Musk’s interview with Dwarkesh Patel and notes that Elon’s methodology is always asking, what is the limiting factor right now, and how do I remove it?

Chip output is growing exponentially. Electricity production outside China is flat. By the end of this year, Elon predicts AI chips will be piling up faster than anyone can turn them on. The companies that win are the ones that can plug their chips in, not the ones that buy the most.

[…]

Solar panels produce 5x more power in orbit because there is no atmosphere, no day/night cycle, no weather, and no clouds. And you need zero batteries. Combined, that is roughly 10x the economics of ground-based solar. Space solar cells are also cheaper to manufacture because they require no glass or heavy framing.

[…]

Within five years, Elon predicts SpaceX will launch hundreds of gigawatts of AI compute into orbit annually, exceeding the cumulative total on Earth. That is 10,000 Starship launches a year. One launch per hour. 20 to 30 reusable ships rotating on 30-hour cycles.

[…]

Only three casting companies in the world make the specialized vanes and blades for gas turbines. They are backlogged through 2030. Everything else in a power plant can be sourced in 12 to 18 months. But without those blades, you have no turbine and no electricity.

[…]

Digital human emulation means an AI that can do everything a human worker can do at a computer: read screens, click buttons, type, think, and decide. NVIDIA’s output is “FTPing files to Taiwan.” Apple sends files to China. Microsoft, Meta, and Google produce nothing physical. If you can perfectly emulate a human at a computer, you can replicate the output of every one of these companies. Customer service alone is a trillion-dollar market with zero integration barriers.

[…]

Elon argues that programming AI to be politically correct, meaning to say things it does not believe, creates contradictory axioms that could make it “go insane.”

When humans represent less than 1% of total intelligence, it would be “foolish to assume there’s any way to maintain control.” The best case is AI with values that find humanity more interesting alive than converted to raw materials. Elon compares the ideal future to Iain Banks’ Culture novels, where superintelligent AI coexists with humans because it finds them interesting.

[…]

Elon runs weekly (sometimes twice-weekly) engineering reviews with skip-level meetings where individual engineers present without advance prep. He mentally plots progress points across weeks to determine if a team is converging on a solution. Time is allocated not to what is going well, but to whatever the current bottleneck is. If something is working great, he stays away.

Most managers optimize for being informed. Elon optimizes for being useful at the point of highest leverage.

Tom Clancy Speaks at the National Security Agency

Thursday, February 12th, 2026

I’ve been slowly working my way through the Tom Clancy novels, and I just stumbled across this old talk he gave at NSA, after writing his first two novels:

Own the night or die

Monday, January 19th, 2026

Own the night or die, John Spencer says:

In three major conflicts involving forces that range from professional to semiprofessional—the 2020 war between Azerbaijan and Armenia, the ongoing Russia-Ukraine war, and Israel’s campaign against Hamas after October 7, 2021—large-scale night operations have been notably rare. Outside of highly specialized units conducting limited raids, most decisive fighting has occurred during daylight. At night, both sides tend to pause, reorganize, and recover. In effect, the night is ceded rather than dominated.

That reality stands in sharp contrast to what the US military demonstrated in Operation Absolute Resolve. US forces executed a complex, high-risk mission deep inside a dense capital city at night. The operation required joint and interagency integration across air, land, sea, and cyber domains and fusing intelligence, special operations forces, and other capabilities. Power was cut. Targets were overwhelmed. The mission concluded with zero American casualties and zero loss of equipment. It was a near-flawless demonstration of a capability that takes decades to build and years to sustain.

That success is even more striking when viewed against earlier US experience. Operation Eagle Claw remains a cautionary case of what happens when night operations exceed institutional readiness. The 1980 hostage rescue attempt in Iran required unprecedented joint coordination and depended on a complex, multiphase plan involving long-range infiltration, helicopters, and clandestine ground movement deep inside hostile territory, much of it planned for execution under conditions of limited illumination and degraded visibility. Mechanical failures, severe dust storms, and navigation challenges reduced the assault force below the minimum required to continue the mission. During the withdrawal from Desert One—a staging area where the mission was aborted—a helicopter operating in degraded visibility collided with a transport aircraft, killing eight US servicemembers. Eagle Claw exposed serious deficiencies in joint planning, rehearsal, and integration. Strategically, it revealed the limits of American power projection in denied environments and directly drove sweeping reforms, including the creation of US Special Operations Command.

A decade later, Operation Just Cause marked significant progress but also underscored how darkness magnifies the challenges of identification, control, and coordination. The 1989 invasion of Panama involved approximately twenty-seven thousand US troops and successfully dismantled the Panamanian Defense Forces within days. The operation deliberately began at night, with major assaults initiated around midnight and continuing through hours of darkness, requiring near-simultaneous airborne and ground attacks against multiple objectives across Panama. During the opening night of the operation, including the seizure of Torrijos-Tocumen International Airport and other key sites, fratricide occurred amid limited visibility, compressed timelines, and the rapid convergence of aircraft and ground forces. The Joint Chiefs of Staff history of the operation highlights the extraordinary command-and-control demands created by this nighttime tempo, illustrating how darkness, density of friendly forces, and speed of execution strained identification and coordination even within an increasingly capable joint force. Just Cause demonstrated growing US proficiency in large-scale night operations, but it also showed that darkness punishes even small lapses in control, communication, and situational awareness.

The difference between those operations and more recent successes was not technology alone. It was mastery earned through relentless training, professionalization, and a force-wide expectation that fighting at night is not exceptional. It is preferred.

Adding radars, LiDARs, and other sensors to cameras does not meaningfully advance us toward full self-driving

Saturday, January 17th, 2026

Adding radars, LiDARs, and other sensors to cameras does not meaningfully advance us toward full self-driving, Genma_Jp argues:

Here are the six main reasons:

Marginal information gain: RADAR and LiDAR primarily provide depth and relative velocity — data that modern neural networks can already derive sufficiently from camera images alone, especially given that precision requirements decrease at longer distances.

LiDAR’s fundamental weaknesses: It performs poorly in rain, fog, and on reflective surfaces (blooming), produces sparse and noisy returns requiring fragile clustering, and lacks the angular resolution for reliable classification at distance.

RADAR’s practical limitations: Despite better weather penetration, it delivers extremely sparse detections, suffers from clustering and classification challenges, and often masks weaker objects behind stronger reflectors — particularly problematic for static infrastructure in low-speed scenarios.

Irreplaceable role of vision: RADAR and LiDAR cannot detect critical semantic information — traffic signs, lights, lane markings, or pedestrian intent cues. Stellar computer vision is mandatory anyway; the other sensors cannot compensate for its absence.

Cameras are robust enough: Modern imagers match or exceed human-eye performance, and practical mitigations (wipers, airflow) handle issues like raindrops. In truly degraded visibility, the safe response is to slow down — something an AV can do systematically, just as humans do.

Fusion as a crutch: Multi-sensor approaches deliver quick early wins by patching vision weaknesses, but they mask the need for true mastery of computer vision through massive data and compute. Companies end up over-investing in complex fusion logic instead of solving the hard problem.

Directed Infrared Counter Measures use a sophisticated laser to disrupt the incoming missile’s infrared “heat-seeking” sensor

Monday, January 5th, 2026

Early MANPADS (Man Portable Air Defence Systems) would lock onto the exhaust plumes of aircraft and were countered by deploying flares. Modern Directed Infrared Counter Measures (DIRCM) use a sophisticated laser to disrupt the incoming missile’s infrared “heat-seeking” sensor:

With a laser energy source embedded in a highly agile enclosed turret system, a DIRCM can be infinite in duration and provide protection for the whole mission, keeping aircrews safe even in dense threat engagement environments.

[…]

Whereas flares are omnidirectional, a DIRCM focusses a beam of light directly at an incoming missile. However, that beam of light needs to:

  • Be able to have line of sight to the missile — a DIRCM with a twin or triple turret system allows for multiple threats to be countered simultaneously, no matter how the aircraft may be manoeuvring
  • Be able to track and engage in a very short space of time — MANPADS can be supersonic in less than a second after firing
  • Emit significant laser energy power to disrupt the missile seeker for long enough so the missile is unable to acquire, re-acquire or track the aircraft

Not so blinding as New Mexico test because of bright sunlight

Sunday, January 4th, 2026

Now It Can Be Told by Leslie M. Groves At about 4:30 a.m. the Duty Officer delivered General Groves the detailed hoped-for cable from Farrell, as Groves explains (in Now It Can Be Told: The Story of the Manhattan Project), which had been dispatched after the bomber returned to Tinian. It read:

Following additional information furnished by Parsons, crews, and observers on return to Tinian at 060500Z. Report delayed until information could be assembled at interrogation of crews and observers. Present at interrogation were Spaatz, Giles, Twining, and Davies.

Confirmed neither fighter or flak attack and one tenth cloud cover with large open hole directly over target. High speed camera reports excellent record obtained. Other observing aircraft also anticipates good records although films not yet processed. Reconnaissance aircraft taking post-strike photographs have not yet returned.

Sound—None appreciable observed.

Flash—Not so blinding as New Mexico test because of bright sunlight. First there was a ball of fire changing in a few seconds to purple clouds and flames boiling and swirling upward. Flash observed just after airplane rolled out of turn. All agreed light was intensely bright and white cloud rose faster than New Mexico test, reaching thirty thousand feet in minutes it was one-third greater diameter.

It mushroomed at the top, broke away from column and the column mushroomed again. Cloud was most turbulent. It went at least to forty thousand feet. Flattening across its top at this level. It was observed from combat airplane three hundred sixty-three nautical miles away with airplane at twenty-five thousand feet. Observation was then limited by haze and not curvature of the earth.

Blast—There were two distinct shocks felt in combat airplane similar in intensity to close flak bursts. Entire city except outermost ends of dock areas was covered with a dark grey dust layer which joined the cloud column. It was extremely turbulent with flashes of fire visible in the dust. Estimated diameter of this dust layer is at least three miles. One observer stated it looked as though whole town was being torn apart with columns of dust rising out of valleys approaching the town. Due to dust visual observation of structural damage could not be made.

At no time was there any idea of testing the gun-type bomb

Monday, December 29th, 2025

Now It Can Be Told by Leslie M. Groves In late June, as the forces under General MacArthur and Admiral Nimitz approached within bombing range of the Japanese homeland, General Groves realized that they had not been told about the ban on certain cities, as he explains (in Now It Can Be Told: The Story of the Manhattan Project), for at the time it was imposed they had been too far away to make it necessary:

This concern was soon removed, however, for when we brought the matter to the attention of the Joint Chiefs, they hastily reserved our targets from all air attack.

We were fairly sure by now that we would be able to test the Fat Man, the implosion-type bomb, sometime around the middle of July. (At no time was there any idea of testing the gun-type bomb.) Planning for this operation, which carried the code name of Trinity, had begun back in the spring of 1944 when Oppenheimer and I decided that a test might be necessary to make certain that the complex theories behind the implosion bomb were correct, and that it was soundly designed, engineered, manufactured and assembled—in short, that it would work.

We thought then that we might want to explode the first bomb inside a container, so that if a nuclear explosion did not take place or if it was a very small one, we might be able to recover all or much of the precious plutonium. Also, we wanted to prevent its being scattered over a wide area and creating a health hazard that would make it necessary to guard the area against trespassers for many years.

Consequently we ordered from Babcock and Wilcox a heavy steel container, which because of its great size, weight and strength was promptly christened Jumbo. To move it from the manufacturing plant in the East to New Mexico, it had to be loaded onto specially reinforced cars and carefully routed over the railroads. At the nearest railroad stop to the test site it was unloaded onto a specially built trailer with some thirty-six large wheels, and then driven overland about thirty miles to Alamogordo.

But by the time of the test we had decided we would not need to use Jumbo, for we had learned enough to be reasonably certain of a fair-sized nuclear explosion. Even if it were as low as 250 tons, as many of our scientists were predicting, the container would only create additional dangers.

It is interesting to speculate about what would have happened, with the actual explosion of almost twenty thousand tons, if we had used Jumbo. That the heat would have completely evaporated the entire steel casing is doubtful. If it did not, pieces of jagged steel would probably have been hurled for great distances.

The scientist in charge of the test was Dr. K. T. Bainbridge, who had the unusual qualification of being a physicist with undergraduate training in electrical engineering.

[…]

I had ruled out using Los Alamos for the test on grounds of security and also because I doubted if the area could be expanded sufficiently. Later, we decided that we would need a site measuring approximately seventeen by twenty-four miles, that it should be in a generally non-populated area, and that it should be no further from Los Alamos than necessary. I added one special prohibition: that it should have no Indian population at all, for I wanted to avoid the impossible problems that would have been created by Secretary of the Interior Harold L. Ickes, who had jurisdiction over the Bureau of Indian Affairs. His curiosity and insatiable desire to have his own way in every detail would have caused difficulties and we already had too many.

[…]

Air travel has improved considerably since those days. The field we used at Pasadena was very small, and our approach to it was impeded by some high-tension lines at the end of the strip. As he came in, our pilot found himself lined up on the taxiway and quite low. Instead of circling the field, he came in over the wires and then side-slipped, landing with a terrific bounce—both horizontal and vertical. Our landing brought everyone out of the small operations office, including one of my security officers who had missed the plane in San Francisco, and who was waiting to rejoin us in Pasadena. He remarked afterward that, if not the first, at least the second thought that flashed through his mind was: “How am I going to explain the accidental death of Bush, Conant and Groves, without publicity to the project and resulting breaches of security?”

We left the next morning from March Field in Riverside in order to be sure that the predicted Los Angeles fog would not interfere with our taking off.

[…]

The main problem was the weather. We had obtained the very best men that the armed forces had on long-range weather forecasting, and, for a considerable period, they had been making accurate long-range weather predictions for the test site. The only time they were not right was on the one day that counted. The weather that evening was quite blustery and misty, with some rain. Fortunately, the wind seemed to be in the right direction.

We were interested in the weather for a number of reasons: First and foremost, we wanted to avoid as much radioactive fallout2 as possible, particularly over populated areas. This was a matter that had not received any attention until about six months earlier, when one of the Los Alamos scientists, Joseph Hirschfelder, had brought up the possibility that it might be a real problem. For this reason, we felt it would be desirable to explode the bomb when rain was unlikely, since rain would bring down excessive fallout over a small area instead of permitting it to be widely distributed and therefore of little or no consequence. In reaching this decision we could not ignore the old reports that heavy battle cannonading had sometime brought on rain, even though no scientific basis was known for a such phenomenon.

Second, it was extremely important that the wind direction be satisfactory, because we did not want the cloud, if one developed, to pass over any populated areas until its radioactive contents were thoroughly dissipated. It was essential that it not pass over any town too large to be evacuated. The city about which we were most concerned was Amarillo, some three hundred miles away, but there were others large enough to cause us worry. The wind direction had to be correct to within a few degrees.

Third, we wanted suitable flying weather so that we could have observation planes flying over the near-by areas; and finally, we wanted to avoid prior heavy rain or continuous dampness, which might ruin our electrical connections, both for firing the bomb, and for the various instruments.

[…]

I was extremely anxious to have the test carried off on schedule. One reason for this was that I knew the effect that a successful test would have on the issuance and wording of the Potsdam ultimatum. I knew also that every day’s delay in the test might well mean the delay of a day in ending the war; not because we would not be ready with the bombs, for the production of fissionable material would continue at full tilt anyway, but because a delay in issuing the Potsdam ultimatum could result in a delay in the Japanese reaction, with a further delay to the atomic attack on Japan. Obviously, a reasonable time had to be allowed for the Japanese to consider the ultimatum.

From a purely technical point of view, also, it was desirable to avoid a postponement, for the chances of short circuits and a misfire would increase appreciably with every hour that our connections were subjected to excessive moisture.

[…]

It had originally been scheduled for 4 a.m. on July 16. This hour had been fixed with the thought that an explosion at that time would attract the least attention from casual observers in the surrounding area, since almost everyone would be asleep. We expected there would be a tremendous flash of light, but thought it would not be great enough to waken many people who were well removed from the burst. Then, too, we wanted the darkness for our photography.

[…]

As the hour approached, we had to postpone the test—first for an hour and then later for thirty minutes more—so that the explosion was actually three and one half hours behind the original schedule. While the weather did not improve appreciably, neither did it worsen. It was cloudy with light rain and high humidity; very few stars were visible. Every five or ten minutes, Oppenheimer and I would leave the dugout and go outside and discuss the weather. I was devoting myself during this period to shielding Oppenheimer from the excitement swirling about us, so that he could consider the situation as calmly as possible, for the decisions to be taken had to be governed largely by his appraisal of the technical factors involved.

[…]

Everyone was told to lie face down on the ground, with his feet toward the blast, to close his eyes, and to cover his eyes with his hands as the countdown approached zero. As soon as they became aware of the flash they could turn over and sit or stand up, covering their eyes with the smoked glass with which each had been supplied.

[…]

As I lay there, in the final seconds, I thought only of what I would do if, when the countdown got to zero, nothing happened. I was spared this embarrassment, for the blast came promptly with the zero count, at 5: 30 A.M., on July 16, 1945.

My first impression was one of tremendous light, and then as I turned, I saw the now familiar fireball. As Bush, Conant and I sat on the ground looking at this phenomenon, the first reactions of the three of us were expressed in a silent exchange of handclasps. We all arose so that by the time the shock wave arrived we were standing.

I was surprised by its comparative gentleness when it reached us almost fifty seconds later. As I look back on it now, I realize that the shock was very impressive, but the light had been so much greater than any human had previously experienced or even than we had anticipated that we did not shake off the experience quickly.

Unknown to me and I think to everyone, Fermi was prepared to measure the blast by a very simple device. He had a handful of torn paper scraps and, as it came time for the shock wave to approach, I saw him dribbling them from his hand toward the ground. There was no ground wind, so that when the shock wave hit it knocked some of the scraps several feet away. Since he dropped them from a fixed elevation from near his body which he had previously measured, the only measurement he now needed was the horizontal distance that they had traveled. He had already calculated in advance the force of the blast for various distances. So, after measuring the distance on the ground, he promptly announced the strength of the explosion. He was remarkably close to the calculations that were made later from the data accumulated by our complicated instruments.

I had become a bit annoyed with Fermi the evening before, when he suddenly offered to take wagers from his fellow scientists on whether or not the bomb would ignite the atmosphere, and if so, whether it would merely destroy New Mexico or destroy the world. He had also said that after all it wouldn’t make any difference whether the bomb went off or not because it would still have been a well worth-while scientific experiment. For if it did fail to go off, we would have proved that an atomic explosion was not possible. Afterward, I realized that his talk had served to smooth down the frayed nerves and ease the tension of the people at the base camp, and I have always thought that this was his conscious purpose. Certainly, he himself showed no signs of tension that I could see.

[…]

These plans proved utterly impracticable, for no one who had witnessed the test was in a frame of mind to discuss anything. The reaction to success was simply too great. It was not only that we had achieved success with the bomb; but that everyone—scientists, military officers and engineers—realized that we had been personal participants in, and eyewitnesses to, a major milestone in the world’s history and had a sobering appreciation of what the results of our work would be. While the phenomenon that we had just witnessed had been seriously discussed for years, it had always been thought of as a remote possibility—not as an actuality.

[…]

Several days after I got back to Washington, Dr. R. M. Evans, of the du Pont Company, came to see me about some of the operating problems at Hanford. After we had finished and as he was leaving, he turned, his hand on the doorknob, and said, “Oh, by the way, General, everybody in du Pont sends you their congratulations.” I quickly replied, “What are you talking about?” He answered, “It’s the first time we ever heard of the Army’s storing high explosives, pyrotechnics and chemicals in one magazine.” He went on to add that the radio announcement on the Pacific Coast had been teletyped in to Wilmington from Hanford. My only response was: “That was a strange thing for the Army to do, wasn’t it?”

[…]

Mr. Stimson’s diary for Sunday, July 22, 1945, is most enlightening:

Churchill read Groves’ report in full. He told me that he had noticed at the meeting of the Three yesterday that Truman was much fortified by something that had happened, that he had stood up to the Russians in a most emphatic and decisive manner, telling them as to certain demands that they could not have and that the United States was entirely against them. He said, “Now I know what happened to Truman yesterday. I couldn’t understand it. When he got to the meeting after having read this report, he was a changed man. He told the Russians just where they got on and off and generally bossed the whole meeting.” Churchill said he now understood how this pepping up had taken place and he felt the same way.