A modernized steel helmet is simultaneously lighter than the PASGT and performs better against both fragments and handgun rounds

Thursday, May 25th, 2023

The first modern combat helmet was the French casque Adrian which was designed to address the threats soldiers faced in the Great War:

In WWI, explosive or fragmenting munitions were responsible for roughly 60-70% of all combat casualties. At the battle of Verdun, fragmentation and shrapnel from artillery bombardment caused at least 70% of the approximately 800,000 casualties that both sides suffered. The remainder were, for the most part, inflicted by relatively heavy rifle and machine-gun rounds which even the best helmets of today would not be able to stop.


The first helmet of the war to enter mass production and see widespread use — and the first modern combat helmet — was the French casque Adrian. This was made of mild steel, 0.7 to 0.8mm thick, with a tensile strength of at least 415 MPa and moderate ductility. (18% tensile elongation.) This helmet was capable of resisting a 230-grain, .45 caliber ball round at 400-450 feet per second, which is roughly half the .45 ACP’s muzzle velocity. But notwithstanding this poor performance against bullets, it is estimated to have defeated 75% of all shrapnel impacts from airburst munitions, and it had, therefore, an immediate positive impact on troop casualty rates and morale. In the Adrian’s wake, every other participant in WWI — except for Russia — hastened to develop and issue steel helmets of their own. Like the Adrian, these helmets had very poor resistance to small arms impacts, but were highly effective at protecting their wearers from shrapnel and fragmentation.

These same steel helmets, with minor modifications in some instances, were employed by all American and European forces through WWII. And here they proved even more vital, for whereas fragments and shrapnel accounted for approximately 65% of all WWI casualties, they accounted for 73% of WWII’s wartime wounds. The widespread use of the steel helmet shifted patterns of wounding and was highly effective at preventing fatal head injury. When the war was over, it was calculated that of all hits upon the US military’s M1 helmet 54% were defeated and, in fact, of all incapacitating hits upon the body, the M1 helmet prevented 10% of them.

Needless to say, all of the helmets of the war were totally incapable of stopping 8mm Mauser, 7.62x54mmR, or .30-06 bullets at most engagement distances — and in fact they would, invariably, fail to stop 7.62x25mm Tokarev handgun/submachinegun rounds within 100 yards under normal ballistic test conditions — but that wasn’t their intended function.


Interestingly, the soft, large, and extremely heavy .45 ball ammo that was used as the test projectile for the M1 couldn’t possibly have been more different from the fragment-simulating projectiles (FSP) used to test helmets today. The FSPs are much lighter — ranging from 2 to 64 grains — and they’re made entirely of AISI 4340 steel heat-treated to 30 HRC. With no jacket, no deformable lead core, and much lighter weights and lower diameters, they’re a qualitatively different threat in every respect.


In the mid 1960s, duPont chemists working on materials for automobile tire reinforcement identified a high-modulus polymer fiber which was first named PRD-49-IV was later trademarked and sold as Kevlar® 29. This material was of immediate interest to the US military. For at the time of its production it was 2.5 times as strong as any other textile fiber, and its performance was 60-100% better than ballistic nylon on a weight basis. Little time was wasted in replacing the nylon and fiberglass flak jackets with more protective and lighter Kevlar vests. And, taking a page from the Hayes-Stewart, Kevlar-laminate helmets — stiffened with about 20% by weight of a polymeric (PVB-phenolic) resin — were developed. Both the vests and the helmets were introduced as the PASGT program, and were issued to the troops in 1983. Some U.S. soldiers wore PASGT helmets in Grenada (Operation Urgent Fury) in 1983, Panama (Operation Just Cause) in 1989, and in the Middle East (Desert Shield/Desert Storm) in 1990-1991.


The PASGT, though not officially rated to stop handgun rounds, was also demonstrably capable of stopping 9mm FMJ service ammunition at typical muzzle velocities.

All of this is tempered somewhat by the fact that the PASGT helmet is markedly heavier than the M1. A size XL PASGT weighs 4.2 pounds; a size XL M1 weighs 2.85 pounds. (The M1 was only offered in one size, which corresponds to an XL in dimensions and coverage.) Were the M1 made 47% heavier, thicker, out of a more modern steel alloy, it stands to reason that its protective capabilities could have kept pace, at a much lower cost and with superior performance against small-arms projectiles. Indeed, we know that this is the case, for a modernized steel helmet — the Adept NovaSteel — is simultaneously lighter than the PASGT and performs better against both fragments and handgun rounds. It is frankly surprising that something along such lines was never attempted or, seemingly, considered. As things stand, it could be argued, and very convincingly, that the introduction of the Kevlar helmet was a mistake.

And that’s without taking into consideration the fact that the PASGT was perhaps an order of magnitude more expensive than the M1, which cost the military $3.03/unit in the early 1950s. ($1.05 for the manganese steel shell, $1.98 for the liner.)

While ballistic protection provided by helmets has increased significantly since WWI, blast protection has not.

Containerize the batteries, charge them nearby the port, and load them like regular cargo

Monday, May 22nd, 2023

Batteries and cargo ships don’t seem like a natural fit, because, Austin Bernon argues, ships need cheap batteries instead of high performance ones:

Let’s examine powering a 10,000 TEU container ship from New York to Rotterdam with the cheapest volume battery chemistry we have, lithium iron phosphate (LFP). At the max continuous speed of 22 knots, power usage is ~40 MW. The trip distance is 3800 miles, so we need ~150 hours of power or 6 GWh. These ships carry ~2 million gallons of fuel, which equates to ~7.5 million liters. The latest CATL Qilin LFP packs contain 290 Wh/l. We can fit nearly 2.2 GWh just in our fuel storage space. Two-stroke marine diesel engines are massive, and we can get another ~0.5 GWh from reducing engine room space and ballast. The last 3.6 GWh can fit in less than 450 TEUs. So we can cross oceans losing only 4.5% of cargo space while gaining the benefits of electrification. The mass performance is not as good, with the batteries using 25% of the cargo mass after netting out fuel and engines. Ships carrying lighter containers would see little cargo penalty, but those carrying heavier goods would.

The startup Fleet Zero has a solution for the charging problem — containerize the batteries, charge them nearby the port, and load them like regular cargo. Ships carry so much fuel today because few fuel sales points offer the required scale, quality, and cost. Filling up at the best places and using large fuel tanks is the way to minimize cost. Recharging at each port will be the lowest cost strategy if batteries and electricity are cheaper than fuel, even if there is variation in prices between ports.

The issue with using LFP batteries is cost. 6 GWh of LFP batteries would cost over $600 million. The cost per cycle would be $110/MWh plus charging costs that could be anywhere from $30/MWh to $100/MWh. Heavy fuel oil at $80/barrel in a 45% efficient engine equates to ~$110/MWh. Business models like Fleet Zero assume multiple stops on routes (which cargo ships commonly do, anyway) and slower speeds. Ships carry fewer batteries on each leg, and the system requires fewer batteries. Cost per cycle drops to reasonable levels while cargo penalties decrease. But it would make adoption easier if cheaper batteries allowed ships to carry more batteries or increased their savings.

These numbers contrast with calculations from analysts like Vaclav Smil because he assumes a 30+ day non-stop trip from Asia to Europe. But shipping companies are in the business of making money,

These missiles are likely completely stealth to the Patriot radar for the majority of their ballistic arc

Friday, May 19th, 2023

Simplicius breaks down Russia’s recent Patriot attack:

Russia was said to have conducted a layered, multi-vectored attack which came from various sides including north, east, and south, which included both Geran drones as screening cover, Kalibr missiles, Kh-101s, and finally the Kinzhals. The attack also likely included other cheaper types of drones as decoys to saturate the air defense, and in fact Kiev does attest to that, as in their official ‘shoot down’ graphic they include several drones they comically ID’d as Orlan ‘Supercum’ which was later changed to ‘Supercam’.

First, let’s break down how such an attack happens. Most logically, the cheaper decoy drones are sent in first to see if they can bait out any of the air defense into opening up on them. Kiev would try to use only its less important SHORAD (Short Range AD) systems against them, such as German Gepards and any Tunguskas/Shilkas and such that they might have.

Next would come the cruise missiles in order to bait out the true high value AD that may have held back with the first wave, and which Ukraine’s SHORAD systems may be useless against.


It should be stated that there are certain positions Russia already knows are likely, and are prefigured into their search matrices. For instance, Mim-104 Patriot system is an extremely complex and large system, you can’t just set it up anywhere, like in the middle of an apartment building courtyard or something like that. These systems not only require a lot of room but also, since they are much less mobile than drivable units like Gepards and such, they are preferably situated somewhere that doesn’t have a lot of civilian ‘eyes’ in the area, so that no one films or rats them out, whether accidentally or not.

This leaves only a few real, solid choices where you can put such a system. And they are almost always put in airports, as an example. It comes as no surprise then that during the attacks on 5/16, word now has it that two of the Patriots were located at Zhuliany airport in Kiev and one at or near the Zoo…

The launch angle of Patriot rockets is fixed at 38° above horizontal. Many other missile systems fire straight up.

This brings up the next issue: a lot of the Patriot missiles appeared to fail. These fallen pieces are not ‘discarded rocket stages’ or anything like that, but the actual missile heads themselves. In fact, we have photo proof that several of them “failed” mid-flight and did the famous ‘Patriot maneuver’ caught long ago in Saudi Arabia:

The Russians’ infamous hypersonic missile is the Kinzhal:

If we take its alleged Mach 10 value, a Mig-31K / Tu-22M3, flying approximately 100-150km north of Kiev over the Russian border, could fire the Kinzhal and it would take a mere 90 seconds or so to arrive in Kiev.

This means that, using the above methods of monitoring, tracking, and observation, once the Russian MOD homes in on a Patriot battery / radar location, it can transfer the coordinates to the Mig-31Ks already in the air, and the Patriots would only have 90 seconds, which is no where near enough time for them to move or do anything to really save themselves.


The other important thing to note is that no one actually knows how fast the Kinzhal or any hypersonic weapons system goes at the point of terminal impact, however it is almost certainly not hypersonic at that point. Yes, you heard that right: no hypersonic weapon on earth actually impacts the target at hypersonic speed.

No where is it actually stated it hits the target at hypersonic speed; this is merely a misleading assumption that people make. In fact, the official description for most hypersonic vehicles like the Kinzhal is that it hits hypersonic velocity at burnout speed. Burnout speed typically means when its engines finish firing during the peak of its ‘ballistic arc’.

People wrongly assume that the point of a hypersonic missile is “to hit the target at a hypersonic speed”. That’s actually not the main advantage. The real point of a hypersonic vehicle is to get to the target as fast as possible, and faster than any other conventional munition, which gives your enemy very little chance to react, such as trying to scramble or hide underground, etc.

The fact is, no manmade object can travel at hypersonic speeds at ground atmospheric levels. The atmosphere is way too thick and any object going such a speed would quickly heat up to astronomical levels and then vaporize. How do space rockets hit hypersonic speeds then, you ask? They accelerate very slow and don’t actually cross the hypersonic threshold until they’re basically already in space.

Most missile types like ballistic missiles and even air to air missiles fired by jets actually shoot up to a very high altitude for most of their cruise, and then come down only as they’re nearing the target. The point is to fly where the atmosphere and air resistance is much thinner to get maximum fuel mileage and acceleration/speed. Cruise missiles are an exception as the exigencies of needing to be ‘below the radar’ require most of them to fly very low.


The second most important thing is that hypersonic vehicles, as noted above, generate a plasma shield around them. This has been by far the biggest reason behind the ‘difficulty’ of creating hypersonic weapons. To accelerate something hypersonically, especially with a basic rocket motor, is easy enough. The problem is then communicating with the object. The plasma shield completely negates all electromagnetic waves, making the object completely impermeable to waves which means you can’t send it any signals to ‘guide’ it to a target.


No one knows which method Russia settled on and uses for the Kinzhal, it’s all classified. However, the likely fact is that the Kinzhal, as well as the Iskander, simply are no longer hypersonic by the time they reach the target, which allows radio signals to give them mid-course correction to the target. The reason is, once they accelerate to their hypersonic ‘burnout speed’ at the top of the ballistic arc, everything after that begins to bleed speed. No one actually knows for certain, but it is likely that by the time of target impact they may be going somewhere in the range of Mach 3-5.

This is still very fast, but keeps them from the ‘plasma field’ problem. How do we know this? Well, there are some videos of Iskander impacts, and while Iskander is said to top out at Mach 6-7 at burnout speed, its impacts do not look hypersonic, though they do look much faster than any other conventional missile types.


Because clearly, if it’s no longer being propelled by thrust, and is merely a ‘glider’ after the zenith of its ballistic arc, then the hypersonic speed it reaches from that point on will be slowly bled little by little. This is likely naturally timed such that the missile is no longer creating a plasma shield or disintegrating itself, such that it’s still going faster than anything else, but can receive course-correction data. This is why my best guess is these missiles actually impact at something like Mach 2-5 at the most.

Also, note that during the May 16 attack, on the night camera footage there was no “glowing objects” descending in the sky. If a Kinzhal was actually traveling Mach 5-7+ when it hit those Patriots, it would have streaked down like a meteor, glowing and throwing plasma.


But there’s one other important aspect not yet mentioned. A plasma bubble absorbs all electromagnetic signals, making the vehicle impervious to them. Guess what that means? That’s right—a hypersonic vehicle is essentially ‘stealth’ and cannot be detected by radar. The radar waves are simply absorbed and ionized by the plasma bubble, and in fact there have been many long years of stealth research in this field.

So the point is that, apropos the argument of whether the Patriot can intercept the Kinzhal or even the Iskander, the fact is, these missiles are likely completely stealth to the Patriot radar for the majority of their ballistic arc. Once they hit the arc and go into ‘glide mode’ and begin slowing down, they slowly come out of stealth, but the problem is, at that point they are already likely over the target and only 15-30 seconds at most from impact, maybe less, and still going a very fast Mach 4-5 at the beginning of the slow down.

A digital leviathan that wields power through opaque algorithms and the manipulation of digital swarms

Friday, May 19th, 2023

Something monstrous is taking shape in America, Jacob Siegel says:

Formally, it exhibits the synergy of state and corporate power in service of a tribal zeal that is the hallmark of fascism. Yet anyone who spends time in America and is not a brainwashed zealot can tell that it is not a fascist country. What is coming into being is a new form of government and social organization that is as different from mid-twentieth century liberal democracy as the early American republic was from the British monarchism that it grew out of and eventually supplanted. A state organized on the principle that it exists to protect the sovereign rights of individuals, is being replaced by a digital leviathan that wields power through opaque algorithms and the manipulation of digital swarms. It resembles the Chinese system of social credit and one-party state control, and yet that, too, misses the distinctively American and providential character of the control system. In the time we lose trying to name it, the thing itself may disappear back into the bureaucratic shadows, covering up any trace of it with automated deletions from the top-secret data centers of Amazon Web Services, “the trusted cloud for government.”

It’s not quite the Dwarven lightning axe of the same name

Thursday, May 18th, 2023

The Air Force plans to spend $320 million buying 1,500 units of Raytheon’s 204-pound GBU-53/B StormBreaker precision glide bomb:

These relatively small (7” diameter) but sophisticated weapons will be built at a facility in Tucson, Arizona through June of 2027. European missile manufacturer MBDA will contribute the pop-out wings that swing out from the bomb upon launch. The latest order is comparable to past unit costs, equating to $213,000 per bomb.


While it’s not quite the Dwarven lightning axe of the same name used by Thor in the Marvel Cinematic Universe, it still has a whiff of the supernatural thanks to its three-eyed “tri-spectral” seeker, offering the option of laser-guidance, an uncooled infrared seeker and a millimeter-wave radar—all mounted on the same moveable gimbal in the nose.

Those sensors can be used in concert to improve accuracy, or used individually if one sensor type is degraded by counter-measures or the explosive device encounters smoke, fog, or rain (which is why ‘StormBreaker’ is all-weather capable). On average, the bomb lands within a meter of its designated target.

While gliding to its target, the bomb’s sensors also allow it to function as a reconnaissance system, feeding back sensor data to be used in locating additional targets or updating mission plans. It can even be instructed to search for specific enemies, using its infrared system to classify possible targets and send back targeting suggestions for approval or refusal by a human operator. This allows use in a fire-and-forget manner, improving survivability of the launching aircraft.

For a good measure, StormBreaker also uses jam-resistant GPS and inertial guidance, and can receive course-corrections from other aircraft or ground forces via its two-way Link 16 datalink. That could potentially allow re-directing of strikes to avoid collateral damage to civilians, or to hit higher priority targets as they’re detected.

When launched from maximum altitude, the glide bomb can engage moving targets up to 45 miles away, or static ones at 69 miles—allowing use from outside the range of short-range air defenses, and even lower-end medium-range systems. Against closer targets, though, the bomb employs an energy-burning ‘spiral mode’ trajectory to avoid overshooting its target.

The weapon’s 105-pound multi-purpose shaped-charge warhead is said to be effective against targets ranging from main battle tanks to infantry, unfortified buildings, and patrol boats. The bomb’s ability to hit moving targets is meant to make it capable of enforcing a ‘no-drive’ zone (the ground-based equivalent to a No-Fly Zone), forbidding traversal of an area by a warring party’s ground vehicles. It also seems useful for battling navies that rely on numerous smaller boats, like those of North Korea or the Iranian Revolutionary Guard Corps.

Between the warhead’s precision and relatively small size—as compared to unguided or GPS-outfitted bombs often clocking in at 500-, 1,000 and 2,000 pounds—Raytheon has argued that this bomb is ideal for minimizing collateral damage in densely populated areas.

Having all of these options built into one weapon streamlines logistics by removing the need to load multiple weapon types on a warplane in the interest of accounting for various contingencies.


Intriguingly, Raytheon has also suggested adding propulsion—likely a rocket booster—to further extend the weapon’s reach. If that can be done at limited additional cost, the glide bomb might transform into a comparatively cheap missile for picking off air defenses and high-value mobile targets from a moderate standoff distance.

The wing is a fuel tank, and the fuel indicator showed 0.000

Wednesday, May 17th, 2023

A lifting body is the opposite of a flying wing; it’s an aircraft or configuration in which the body itself produces lift. Some aircraft with wings also employ bodies that generate lift, like the F-15 Eagle, which produces substantial lift from the wide fuselage between the wings:

A simulated dogfight training took place between two F-15D’s and four A-4N Skyhawks over the skies of the Negev, Israel. The F-15D #957, (nicknamed ‘Markia Shchakim’, 5 killmarks) was used for the conversion of a new pilot in the squadron. Here is the description of the event as described in “Pressure suit”:

“At some point I collided with one of the Skyhawks, at first I didn’t realize it. I felt a big strike, and I thought we passed through the jet stream of one of the other aircraft. Before I could react, I saw the big fire ball created by the explosion of the Skyhawk.

The radio started to deliver calls saying that the Skyhawk pilot has ejected, and I understood that the fireball was the Skyhawk, that exploded, and the pilot was ejected automatically.

There was a tremendous fuel stream going out of my wing, and I understood it was badly damaged. The aircraft flew without control in a strange spiral. I reconnected the electric control to the control surfaces, and slowly gained control of the aircraft until I was straight and level again. It was clear to me that I had to eject. When I gained control I said : “Hey, wait, don’t eject yet!” No warning light was on and the navigation computer worked as usual; (I just needed a warning light in my panel to indicate that I missed a wing…).” My instructor pilot ordered me to eject.

The wing is a fuel tank, and the fuel indicator showed 0.000 so I assumed that the jet stream sucked all the fuel out of the other tanks. However, I remembered that the valves operate only in one direction, so that I might have enough fuel to get to the nearest airfield and land. I worked like a machine, wasn’t scared and didn’t worry. All I knew was as long as the sucker flies, I’m gonna stay inside. I started to decrease the airspeed, but at that point one wing was not enough. So I went into a spin down and to the right. A second before I decided to eject, I pushed the throttle and lit the afterburner. I gained speed and thus got control of the aircraft again.

Next thing I did was lower the arresting hook. A few seconds later I touched the runway at 260 knots, about twice the recommended speed, and called the tower to erect the emergency recovery net. The hook was torn away from the fuselage because of the high speed, but I managed to stop 10 meters before the net. I turned back to shake the hand of my instructor, who had urged me to eject, and then I saw it for the first time – no wing !!!


Ukraine uses artillery fire as a long-range sniper weapon

Tuesday, May 16th, 2023

In Ukraine, as in previous major conflicts, artillery is the biggest killer, accounting for 80% of casualties, but Ukraine appears to be doing more damage with fewer rounds:

In November NBC quoted US officials estimating Russian expenditure of 20,000 rounds per day against 4,000-7,000 for Ukraine. NATO Secretary General Jens Stoltenberg stated in February that Russia was firing around four times as many shells as Ukraine. In March, Spanish newspaper El Pais quoted EU insider sources as saying that Russia was firing 40-50,000 rounds per day, compared to 5,000-6,000 for Ukraine, while Estonia (which has supplied shells to Ukraine) estimated that Russia was firing 20,000-60,000 per day compared to 2,000-7,000 from Ukraine.

So Russia is likely firing something between four and nine times as many shells as Ukraine.

And yet, Russia has suffered much higher casualties. To take just one figure, recently leaked Pentagon documents suggest 189,500-223,000 Russians killed or injured compared to 124,500-131,000 Ukrainians or 1.4 to 1.8 to 1.


While Russia has stuck mainly to Soviet doctrine of massed area fires, Ukraine uses artillery fire as a long-range sniper weapon to pick off individual targets. This has been made possible with the widespread use of two innovations: small drones for artillery spotting, coupled with cheap tablet computers running software like Nettle system to direct fire.

Back in 2014, Ukrainian volunteer organization Army-SOS set out to use its technical skill to help the military. They initially helped soldiers fly and support drones, but soon found the biggest problem was using the data gathered by drone operators efficiently. So they developed Kropyva (“Nettle”) proprietary intelligence mapping software, which can run on any Android tablet.

Nettle is supplied as a tactical system compatible with NATO-standard secure communications and is used from divisional command down to individual vehicles. It maps battle lines and targets and calculates artillery fire missions. It is specifically designed to work with drones, receiving data and using it to calculate the adjustment needed. The gunner changes angle and azimuth accordingly, and shells land on target.

Several other Ukraine-developed software packages — GIS Arta, ComBat Vision, and the major Delta battlefield management system – are also used to share data, locating targets and directing fire.

Ukrainian forces use a wide variety of small drones, including several locally-made military-grade types such as the Leleka-100 and Spectator-M for artillery spotting, as well as thousands of DJ consumer quadcopters. The latter has a range of just a few kilometers and a flight endurance of perhaps half an hour, but their low cost means they are expendable and universally available.

In March 2022, Oleksiy Arestovych, adviser to the office of President Zelensky, told the media that a standard platoon defensive position took normally took 60-90 artillery rounds to destroy, but with drone-guided fire this was reduced to just 9 rounds, and that drones had been supplied to all artillery units. This suggests an improvement of a factor of 7-10, which is roughly what we see in the ratios of artillery shells: casualties above.

Previously, a vehicle, especially in a dug-in, camouflaged position or behind buildings or trees might not be detected until enemy forces were close by. Drone observation changes this, with small drones buzzing overhead spotting everything below in real time – not just vehicles but even individual soldiers. Hiding behind a ridge or hill no longer helps. Given suitable software and communications which Nettle supplies, every potential target can be geolocated precisely, the co-ordinates passed to artillery, and rounds walked on to it.


Increasingly drone-enabled Ukrainian tanks are acting in an indirect fire role, engaging Russian armor beyond normal combat ranges and beyond line of sight. In August 2022, a video posted on social media showed a Ukrainian T-64BV destroying a Russian tank at a claimed range of 6.5 miles, which would make it the longest ever tank vs. tank kill. This required some twenty 125mm projectiles, but the Russian could not fire back to the ‘duel’ was entirely one-sided.


Older, supposedly obsolete weapons are being transformed into effective indirect-fire platforms. Videos show 100mm T-12 Rapira anti-tank guns dating from 1961 in this role, and even a T-12 mounted on an MT-LB tracked vehicles. The 73mm SPG-9 recoilless rifle (from 1962), again originally a direct-fire anti-tank weapon, is being also used for precision indirect fire, as is the AGS-17 Plamya 30mm automatic grenade launcher. In this latter case, there does not seem to be any software, just the drone operator standing next to the gunner directing them, or in some cases the gunner observing the drone feed directly to adjust fire.

Diffused lighting camouflage

Wednesday, May 10th, 2023

Diffused lighting camouflage  was a form of active camouflage using counter-illumination to enable a ship or plane to match its background, the night sky. You could light up a ship or plane to make it harder to see:

An equivalent strategy, known to zoologists as counter-illumination, is used by many marine organisms, notably cephalopods including the midwater squid, Abralia veranyi. The underside is covered with small photophores, organs that produce light. The squid varies the intensity of the light according to the brightness of the sea surface far above, providing effective camouflage by lighting out the animal’s shadow.


In 1940, Edmund Godfrey Burr, a Canadian professor at McGill University, serendipitously stumbled on the principle of counter-illumination, or as he called it “diffused-lighting camouflage”. Burr had been tasked by Canada’s National Research Council (NRC) to evaluate night observation instruments. With these, he found that aircraft flying without navigation lights remained readily visible as silhouettes against the night sky, which was never completely black.

Burr wondered if he could camouflage planes by somehow reducing this difference in brightness. One night in December 1940, Burr saw a plane coming in to land over snow suddenly vanish: light reflected from the snow had illuminated the underside of the plane just enough to cancel out the difference in brightness, camouflaging the plane perfectly.

Burr informed the NRC, who told the RCN. They realized that the technique could help to hide ships from German submarines in the Battle of the Atlantic. Before the introduction of centimetre radar, submarines with their small profile could see convoy ships before they were themselves seen. Diffused lighting camouflage might, the RCN believed, redress the balance.


In January 1941, sea trials began on the new corvette HMCS Cobalt. She was fitted with ordinary light projectors—neither designed for robustness, nor waterproofed—on temporary supports on one side of the hull; brightness was controlled manually. The trial was sufficiently promising for a better prototype to be developed.

The second version, with blue-green filters over the projectors, was trialled on board the corvette HMCS Chambly in May 1941. This gave better results as the filters removed the reddish bias to the lamps when at low intensity (lower colour temperature). The supports too were retractable, so the delicate projectors could be stowed away for protection when not in use. This second version reduced Chambly‘s visibility by 50% in most conditions, and sometimes by as much as 75%. This was enough to justify development of a more robust version.

The third version featured a photocell to measure the brightness of the night sky and the ship’s side; the projectors’ brightness was automatically controlled to balance out the difference. It was tested in September 1941 on the corvette HMCS Kamloops.


The British General Electric Company developed a manually operated diffused lighting system, which was trialled on the ocean boarding vessel HMS Largs and the light cruiser HMS Penelope. The Largs surface observation trials were conducted between 25 January and 6 February 1942; air observation trials, using Hudson bombers, took place on the nights of 4/5 February and 25/26 March 1942. They found an average reduction in the range at which the ship could be seen at night from another ship of around 25% using binoculars, 33% using the naked eye. The results from the air were less conclusive.


The best case was on the exceptionally clear moonless night of 29/30 January 1942, when Largs could be seen from a surface ship with the naked eye at 5,250 yards (4,800 m) unlighted, but only 2,250 yards (2,060 m) with her diffused lighting, a 57% reduction. By June 1942, Royal Navy commanders considered that camouflage was largely unnecessary, given that the enemy would be using RDF and submarine hydrophones. In April 1943, the Admiralty decided that diffused lighting was impractical, and development was halted, though discussions continued with the Canadian Navy.


Because submarines at the surface could see the dark shape of an attacking aircraft against the night sky, the principle of diffused lighting camouflage also applied to aircraft. However, British researchers found that the amount of electrical power required to camouflage an aircraft’s underside in daylight was prohibitive, while externally mounted light projectors disturbed the aircraft’s aerodynamics.


An American version, “Yehudi”, using lamps mounted in the aircraft’s nose and the leading edges of the wings, was trialled in B-24 Liberators, Avenger torpedo bombers and a Navy glide bomb from 1943 to 1945. By directing the light forwards towards an observer (rather than towards the aircraft’s skin), the system provided effective counter-illumination camouflage with an affordable use of energy, more like that of marine animals than the Canadian diffused lighting approach. But the system never entered active service, as radar became the principal means of detecting aircraft.

Hypersonic aircraft are about 80 percent fuel and oxidizer

Saturday, May 6th, 2023

Andrew Duggleby is chief technology officer of Venus Aerospace, which he co-founded with his wife, Sassie:

Venus Aerospace has the goal of building a hypersonic aircraft that can carry perhaps a dozen passengers and travel at the astonishingly fast speed of Mach 9, or more than 11,000 kilometers an hour.

“How much does the world change if you can get anywhere in an hour?” Sassie Duggleby asked me.


One problem is that Mach 9 is really, really freaking fast. No airplane has ever gone this fast. The speediest airplane ever built is Lockheed’s SR-71 “Blackbird,” which traveled at Mach 3.2. Anything above Mach 9 and you lose communications with the ground, as plasma starts enveloping the vehicle, as if it were a spacecraft returning to Earth through the upper atmosphere.

In terms of passenger travel comparisons, the Concorde supersonic airliner traveled at Mach 2, or about 2,100 km/hour. Most of the newer generation of supersonic aircraft under development today are in about the same range, such as Boom Supersonic’s cruising speed of Mach 1.7

The Dugglebys are proposing a radically different flight profile. They intend for their aircraft to take off and then perform a 10-minute boost with its rocket engine. This will send the aircraft to an altitude of approximately 50 km, or half the way to space. Oh, and they’re aiming for an airport-like operational cadence of four flights a day.

To that end, the company recently decided on a fuel mix for its engine: room-temperature hydrogen peroxide and Jet-A, the fuel used by a majority of jet aircraft already flying at airports. The company’s engineers also recently achieved liquid peroxide and Jet A detonation, which is important for using a stable fuel composition.

One key to making all of this work is using a new type of engine based on “rotating detonation.” Governments around the world have been researching this technology for more than a decade because it has the potential to increase fuel efficiency in a variety of applications, from US Navy ships to rocket engines.

In a traditional rocket engine, a highly pressurized propellant and an oxidizer are injected into a combustion chamber where they burn and produce a tremendously energetic exhaust plume—Newton’s second law of motion in action. A rotating detonation engine is different in that a wave of detonation travels around a circular channel. This is sustained by the injection of fuel and oxidizer and produces a shockwave that travels outward at supersonic speed.


In lab tests, the engines have provided about a 10 percent increase in fuel efficiency.

That may not sound like a whole lot, but it is a make-or-break number for Venus Aerospace. By mass, hypersonic aircraft are about 80 percent fuel and oxidizer. By increasing that fuel efficiency, there is actually mass left over for important things like landing gear, wings, and even some passengers. “It allows us to truly build a vehicle that is like an airplane,” Andrew Duggleby said.


Venus aims to go supersonic with an 8-foot drone before the end of this year and hit Mach 3 by early 2024 with a rotating detonation engine.

The company has about 80 full-time employees and 20 contractors, the majority of whom work at the company’s hangar at the Houston Spaceport. Venus Aerospace so far has raised $41 million, led by Prime Movers Lab, and Sassie Duggleby said she is working on raising a second round of funding.

Windows filled with see-through wood layer help hold in heat

Sunday, April 30th, 2023

A see-through aerogel made from wood could replace air in double-glazed windows and make them as insulating as walls:

Windows with air sandwiched in the gap between plates of glass can be made better insulators by either increasing the number of glass panels, which can affect visual quality, or expanding the width of the air layer — but anything beyond around 1.5 centimetres becomes detrimental to the insulation effect because convection currents circulate more easily.

To address this, Ivan Smalyukh at the University of Colorado Boulder and his colleagues used nanofibres of cellulose to create an aerogel, a solid gel containing pockets of gas, that could function better than air in double glazing.

The model underlying Dolly only has 6 billion parameters, compared to 175 billion in GPT-3

Tuesday, April 25th, 2023

ChatGPT, a proprietary instruction-following model, was released in November 2022 and took the world by storm:

The model was trained on trillions of words from the web, requiring massive numbers of GPUs to develop. This quickly led to Google and other companies releasing their own proprietary instruction-following models. In February 2023, Meta released the weights for a set of high-quality (but not instruction-following) language models called LLaMA to academic researchers, trained for over 80,000 GPU-hours each. Then, in March, Stanford built the Alpaca model, which was based on LLaMA, but tuned on a small dataset of 50,000 human-like questions and answers that, surprisingly, made it exhibit ChatGPT-like interactivity.

Today we are introducing Dolly, a cheap-to-build LLM that exhibits a surprising degree of the instruction following capabilities exhibited by ChatGPT. Whereas the work from the Alpaca team showed that state-of-the-art models could be coaxed into high quality instruction-following behavior, we find that even years-old open source models with much earlier architectures exhibit striking behaviors when fine tuned on a small corpus of instruction training data. Dolly works by taking an existing open source 6 billion parameter model from EleutherAI and modifying it ever so slightly to elicit instruction following capabilities such as brainstorming and text generation not present in the original model, using data from Alpaca.

The model underlying Dolly only has 6 billion parameters, compared to 175 billion in GPT-3, and is two years old, making it particularly surprising that it works so well. This suggests that much of the qualitative gains in state-of-the-art models like ChatGPT may owe to focused corpuses of instruction-following training data, rather than larger or better-tuned base models.

United Airlines will fly an air taxi service between the downtown Vertiport Chicago and O’Hare

Sunday, April 23rd, 2023

In 2025, United Airlines will fly an air taxi service between the downtown Vertiport Chicago and O’Hare International Airport, using electric vertical takeoff and landing aircraft it is purchasing from Archer Aviation:

The Archer Midnight eVTOL aircraft will complete the route in about 10 minutes; according to local resident and Ars Managing Editor Eric Bangeman, that journey by car can take over an hour due to road construction.


United placed an order for 200 eVTOL aircraft from Archer back in 2021 at a cost of $1 billion.


The Archer Midnight has a range of 100 miles (160 km) and a top speed of 150 mph (241 km/h).


Asked about the cost, an Archer spokesperson told the Chicago Sun-Times that the company hopes to make the service competitive with Uber Black, so it will be roughly $100 for the trip.

Why did we wait so long for wind power?

Tuesday, April 11th, 2023

Why did we wait so long for wind power?

The first mention of using wind to generate electricity is credited to William Thomson (better known as Lord Kelvin). In an 1881 address to the British Association for the Advancement of Science, Thomson noted that a windmill connected to a dynamo and a battery could be used to provide electric power, and speculated that, while wind had largely been superseded by steam generated by burning coal, “When the coal is all burned…it is most probable that windmills or wind-motors in some form will again be in the ascendant”.

It didn’t take long for windmills connected to electric generators to be built. James Blythe, a professor in Glasgow, built a windmill in 1887 that used a generator to charge batteries to power the lights in his vacation home. The same year a similar system was built in La Hève Cape in France to provide power to homes. In the US, inventor Charles Brush built a huge windmill to power the 350 lightbulbs in his Cleveland mansion in 1890. (For context, Edison patented his lightbulb in 1880, and the first central electric power station was built at Pearl Street in 1882). And in 1893, arctic explorer Fridjtjof Nansen commissioned a ship that included windmill powered electric lights for a voyage to the North Pole.


However, these efforts were ultimately derailed by the steady march of cheaper electricity. In 1882, Edison’s Pearl Street Station operated at a thermal efficiency of 2.5%. By 1920, steam turbines were operating at 20% efficiency, and by 1960 they had reached 40% efficiency. Fuel prices similarly fell — between 1949 and 1965 the real price of oil and coal fell by 20% and 33%, respectively. Larger power plants were able to capture greater economies of scale. Between 1920 and 1970 the cost of coal-generated electricity fell by approximately 80%.

We must be willing to consider what might at first seem absurd and unworkable

Sunday, April 9th, 2023

A recent paper proposes the deployment of volcanic geothermal energy from the Yellowstone Caldera Supervolcano, using a completely new copper-based engineering approach:

The proposed ideas, if implemented, would allow the production of green, 100% emission-free energy for the United States of America and possibly beyond, to last the years and centuries to come, while having the great added benefit of forestalling the Yellowstone Supervolcano from potentially ever erupting again.

To consider such an implementation, the reader is asked to be willing to think big and bold. We must be willing to consider what might at first seem absurd and unworkable, only later to realize that it is absolutely feasible and realistic to implement, with even today’s current technologies, and with some imagination.


By utilizing thermally conductive copper pillars on an unprecedented scale, this paper proposes a means to draw up this Supervolcano’s mighty energy reserve from within the Earth, to superheat steam for spinning turbines at sufficient speed on a sufficient scale and number, to power the entire USA, from a single, multi-redundant facility that utilizes the star topology in a grid array pattern. In so doing, over time, bleed-off of sufficient energy from the Supervolcano’s magma chamber will potentially forestall this Supervolcano from ever erupting again. In 2017, NASA conducted a study to determine the feasibility of preventing the Yellowstone Supervolcano from erupting. The results of this study showed that cooling the magma chamber by 35% would be enough to forestall another eruption.


This paper proposes to make a power generation facility on the Yellowstone Caldera, with a satellite view shown in Fig. 1 from Google Maps, capable of generating well over twice the projected electrical energy usage in the year 2050 of 5.5 Quadrillion Watt hours for the entire United States of America. We therefore assume well over 11 Quadrillion Watt hours of electrical energy over the course of one full year used by the USA.


The Arabelle steam turbines run on high pressure, superheated steam. Steam is made from water and heat. Shoshone Lake is next to the Yellowstone Caldera [37]. This fresh water lake covers over 8000 acres and has a maximum depth of over 200 ft. Water from this lake could be used for the steam production. In the proposed design laid out in this paper, all water taken from the lake would be returned to the lake, cleaner than it was originally, at ambient temperature, and free of any contamination, as all major plumbing in this proposed design is made with copper, a standard plumbing material. As with all plumbing, scaling buildup on the inside of the piping and corrosion resistance on the outside of the plumbing in contact with the Earth are important considerations for the long-term life, water flow performance, maintenance of all necessary thermal conduction and heat transfer capabilities of the piping. To prevent scaling buildup on the inside of the plumbing, the water passing through these pipes must first be demineralized. All extracted minerals will be returned to the lake with the returning water, so as not to affect the lake’s chemistry, and thus the aquatic inhabitance of the lake. External corrosion resistance of the copper plumbing in contact with the Earth will be obtained through minimally adequate gold (with a nickel interface) plating, as gold is nearly impervious to corrosion. By taking these measures, there should be no degradation of geothermal energy extraction over time.

Based on numerous studies, and also reported by the U.S. Geological Survey the temperature in Yellowstone Supervolcano’s magma chamber is approximately 1475°F, and its size is approximately 40 km long by 80 km across, similar in size to the flat, overlying Yellowstone Caldera. The Yellowstone Caldera is large enough for all of the required Arabelle steam turbines and all other required hardware and plumbing to fulfill the proposal set forth in this paper. In terms of depth, the top of the magma chamber is 8 km below the surface, and the bottom of the magma chamber is 16 km below the surface.

In the most general terms, this paper proposes to use metallic thermal conduction as a means to transport the tremendous thermal energy flux of the Yellowstone Supervolcano magma chamber to the surface for utilization. Looking at the selected material of copper, reveals that its melting point is 1983°F, which is well above the internal temperature of the Yellowstone Supervolcano magma chamber. At this temperature, copper has a thermal conductivity value of 350 W/(m x °K). Copper is an excellent metallic thermal conductor, lacking in corrosion resistance. The use of gold (with a nickel interface) plating on copper that may be in contact with the Earth is recommended. This plating has a melting point at or higher than the copper it is meant to protect. Furthermore, this minimalist plating method, used extensively in the electronics industry, provides superior corrosion protection while imposing relatively minimal impact on the facility cost, the overall thermal conduction, and heat transfer performance of the geothermal energy extraction process.


As the magma chamber is 8 km below the surface, 8 kilometer-long copper cylinders (mostly hollow) would be required for this work. We propose that these cylinders be made in 10 m long segments that interconnect with one another at their ends.


As water is pumped down the center bore hole under pressure, high pressure, superheated steam would be forced to return up through each of the surrounding smaller diameter bore holes of the copper cylinder. This returning steam would exit the copper cylinders at the top, above the surface of the caldera, and be piped through the Arabelle steam turbines generating electrical power. After exiting the steam turbines, this exhausted steam would then be collected and cooled with a water condensing network. The re-condensed water would be brought to ambient temperature before returning the water back to Shoshone Lake. This entire process would be done in parallel for all 100 of the copper cylinders simultaneously. In so doing, all 1000 Arabelle stream turbines would be generating power at the same time. This electrical power would be put onto the nationwide power grid, supplying electrical energy to the entire USA, and possibly beyond.

The ones who could solve the problem didn’t appear any “brighter” in conversation than the ones who couldn’t

Monday, March 27th, 2023

When OpenAI released GPT-2, S.R. Constantin remarked that it was disturbingly good:

The scary thing about GPT-2-generated text is that it flows very naturally if you’re just skimming, reading for writing style and key, evocative words.


If I just skim, without focusing, they all look totally normal. I would not have noticed they were machine-generated. I would not have noticed anything amiss about them at all.

But if I read with focus, I notice that they don’t make a lot of logical sense.


The point is, if you skim text, you miss obvious absurdities. The point is OpenAI HAS achieved the ability to pass the Turing test against humans on autopilot.

The point is, I know of a few people, acquaintances of mine, who, even when asked to try to find flaws, could not detect anything weird or mistaken in the GPT-2-generated samples.

There are probably a lot of people who would be completely taken in by literal “fake news”, as in, computer-generated fake articles and blog posts. This is pretty alarming. Even more alarming: unless I make a conscious effort to read carefully, I would be one of them.

Robin Hanson’s post Better Babblers is very relevant here. He claims, and I don’t think he’s exaggerating, that a lot of human speech is simply generated by “low order correlations”, that is, generating sentences or paragraphs that are statistically likely to come after previous sentences or paragraphs.


I’ve interviewed job applicants, and perceived them all as “bright and impressive”, but found that the vast majority of them could not solve a simple math problem. The ones who could solve the problem didn’t appear any “brighter” in conversation than the ones who couldn’t.

I’ve taught public school teachers, who were incredibly bad at formal mathematical reasoning (I know, because I graded their tests), to the point that I had not realized humans could be that bad at math — but it had no effect on how they came across in friendly conversation after hours. They didn’t seem “dopey” or “slow”, they were witty and engaging and warm.


Whatever ability IQ tests and math tests measure, I believe that lacking that ability doesn’t have any effect on one’s ability to make a good social impression or even to “seem smart” in conversation.

If “human intelligence” is about reasoning ability, the capacity to detect whether arguments make sense, then you simply do not need human intelligence to create a linguistic style or aesthetic that can fool our pattern-recognition apparatus if we don’t concentrate on parsing content.


The mental motion of “I didn’t really parse that paragraph, but sure, whatever, I’ll take the author’s word for it” is, in my introspective experience, absolutely identical to “I didn’t really parse that paragraph because it was bot-generated and didn’t make any sense so I couldn’t possibly have parsed it”, except that in the first case, I assume that the error lies with me rather than the text. This is not a safe assumption in a post-GPT2 world. Instead of “default to humility” (assume that when you don’t understand a passage, the passage is true and you’re just missing something) the ideal mental action in a world full of bots is “default to null” (if you don’t understand a passage, assume you’re in the same epistemic state as if you’d never read it at all.)