October 2018 – Militant Futurist

October 31, 2018November 1, 2018

Roundup of interesting articles, October 2018

In spite of what Hollywood would have you believe, submarines almost never fight each other.
https://nationalinterest.org/blog/buzz/showdown-story-only-submarine-vs-submarine-battle-34652

Should we replace our aircraft carriers with cargo ships full of containerized missiles and drones?
https://taskandpurpose.com/maritime-airpower-aircraft-carrier/

The U.S. Navy will buy more of its venerable Arleigh-Burke destroyers.
https://nationalinterest.org/blog/buzz/navy-getting-10-new-flight-iii-destroyers-and-it-could-be-game-changer-32427

The U.S. Army tested an experimental conversion kit that allowed a computer to fly a helicopter designed for a human pilot.
https://www.janes.com/article/84156/us-army-flies-automated-sara-helicopter

The USAF has selected the Boeing “T-X” to be its new training jet. It will replace the T-38, which first flew in 1959. Given the longevity of modern warplanes, the T-X should stay in service until at least the 2060s, by which time fighter and bomber planes might be automated, rendering human pilots obsolete. Thus, the T-X could be the last, or at best the second-to-last, trainer aircraft that the USAF ever makes. AIs won’t need to spend time in a simplified practice plane to learn how to fly. They will just be created in software labs and uploaded directly into frontline combat planes. Someday, the very notion of a “trainer aircraft” will be obsolete.
http://www.thedrive.com/the-war-zone/23889/boeing-wins-big-again-and-claims-the-usafs-huge-t-x-jet-trainer-deal

The T-38 pilot training jet could be converted into a fighter plane, but it would be crappy at its new job, and it would be a poor use of money considering what you’d get in return. Once the U.S. adopts the T-X, we’ll probably transfer our surplus T-38s to poorer allied countries or to aviation museums.
https://www.quora.com/Can-an-aircraft-such-as-a-T-38-be-weaponised-in-a-war-time-situation

Even if a stealth plane is invisible to radar, it will be hotter than the air around it thanks to its jet engine and to air friction against its wings, so you will still be able to see the plane using a thermal camera (e.g. – “Predator” heat vision).
https://nationalinterest.org/blog/buzz/rip-stealth-russian-su-35-reportedly-took-picture-f-22-why-might-be-problem-34267

The U2 spy plane’s camera still uses analog film.
https://petapixel.com/2018/06/08/film-photography-at-70000-feet-in-the-u-2-spy-plane/

The U.S. just withdrew from the Intermediate-Range Nuclear Forces (INF) Treaty because Russia had been cheating for years, and the U.S. needs medium-range missiles to deter China in the Pacific.
https://www.washingtonpost.com/world/2018/10/23/how-china-plays-into-trumps-decision-pull-out-inf-treaty-with-russia/

The U.S. might start mass producing copies of Russian guns for foreign military assistance. Ukraine and Bulgaria already have factories for making these guns, but apparently they can’t keep up with demand.
http://www.thedrive.com/the-war-zone/24283/russia-rages-at-pentagon-plans-to-build-u-s-made-derivatives-of-soviet-era-small-arms

The USSR planned to quickly nuke Austria if WW3 ever broke out even though Austria wasn’t in NATO and has enshrined neutrality in its constitution.
https://www.telegraph.co.uk/news/worldnews/europe/austria/1364037/Vienna-was-top-of-Soviet-nuclear-targets-list.html

A police helicopter was used to break up a rowdy college party by flying low over them. The downdraft blew away their tents and lawn furniture. This is one, overlooked reason why flying cars were never built.
https://youtu.be/j4Au-yCQur0

Science is coming around to buttressing what has been common sense forever: Sunlight exposure raises your mood. This finding is particularly interesting since it establishes a biochemical pathway linking skin cells to brain cells.
https://blogs.sciencemag.org/pipeline/archives/2018/10/12/sunlight-and-the-brain

Another astounding discovery: cardiovascular exercise decreases your risk of death! Of note is that fact that there was no observed ceiling to the benefit.
https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2707428

Mushrooms and plants that contain hallucinogenic compounds evolved them as protection against insects. The fact that the chemicals also have effects on human brains is entirely coincidental.
https://blogs.scientificamerican.com/artful-amoeba/magic-mushroom-drug-evolved-to-mess-with-insect-brains/

Another medical study has found the drug Ecstasy can treat PTSD.
https://reason.com/blog/2018/10/29/another-ptsd-study-finds-dramatic-improv

Taller people might be at higher cancer risk because they have more cells in their bodies.
https://www.theguardian.com/science/2018/oct/24/tall-people-at-greater-risk-of-cancer-because-they-have-more-cells

Gene editing in utero might soon be tried in humans that have congenital diseases caused by single-point mutations.
https://blogs.sciencemag.org/pipeline/archives/2018/10/09/gene-editing-freely-you-have-received-freely-give

‘In technical papers my research group anticipated years ago that even very complex traits would be predictable once a [human genome] data threshold was crossed. The phenomenon is related to what physicists refer to as a phase transition in algorithm performance. The rapid appearance now of practically useful risk predictors for disease is one anticipated consequence of this phase transition. Medicine in well-functioning health care systems will be transformed over the next 5 years or so.’
http://infoproc.blogspot.com/2018/10/population-wide-genomic-prediction-of.html

Rapid progress is being made identifying the genes linked to many human traits, including intelligence.
http://infoproc.blogspot.com/2018/10/advances-in-genomic-prediction.html

‘[The] fact that intelligence or personality are caused by many thousands of genes, each of minuscule effect, means that it will be impossibly difficult to create a super-intelligent designer baby.’
http://www.rationaloptimist.com/blog/the-genes-of-human-behaviour/

‘Universal Family Tree — Eventually we will sequence the full genomes of everyone living, and as many of the recent dead as we have access to. Together with genealogical records, this huge trove of data will give us our first universal family tree. Everyone living will have a place on it in relation to everyone else. ‘ –Kevin Kelly, 2012
http://www.latimes.com/science/sciencenow/la-sci-sn-dna-genealogy-privacy-20181012-story.html

Over 100,000 varieties of rice are being held at a global rice “gene bank” in the Philippines.
https://www.bbc.com/news/science-environment-45828150

The “dbSNP” is an open database of all recorded human alleles.
https://www.ncbi.nlm.nih.gov/projects/SNP/

Will cryonically preserved humans ever be revived? The consensus among the disinterested interviewees (e.g. – the people who aren’t running human cryonics companies and who haven’t written anti-cryonics books) is “No”, unless we get Star Trek-level technology. The freezing process damages the brain at the cellular level, and reversing it would require nanomachines.
https://gizmodo.com/will-cryogenically-frozen-people-ever-be-revived-1829905516

Increased use of keyboards and smartphone screens and the decline of handwriting are eroding fine motor control across the population, with particularly harmful impact on prospective surgeons.
https://www.bbc.com/news/education-46019429

A new device called the “Everlast” notebook saves writings and drawings as data files. What kills it is the fact that you have to take photos of what you’ve written on the pages to save them digitally. The pages themselves should be able to detect what the user has written on them, and to upload it to their remote storage drive.

Prediction: Within 20 years, books and computer tablets will merge into a single type of device that could be thought of as a “digital book.” It will be a book with several hundred pages made of thin, flexible digital displays (perhaps using ultra-energy efficient e-ink) instead of paper. At the tap of a button, the text on all of the pages will instantly change to display whichever book the user wanted to read at that moment. They could also be used as notebooks in which the user could hand write or draw things with a stylus, which would then be saved as image or text files. The devices will fuse the tactile appeal of old-fashioned books with the content flexibility of tablet computers.
https://www.kickstarter.com/projects/642311833/everlast

In 1911, Thomas Edison erroneously predicted that most household furnishings and appliances would someday be made of steel.
https://www.inverse.com/article/19331-thomas-edison-alternate-futures-steel-home-design

Winston Churchill predicted that humans would someday eat lab-grown meats.
https://qz.com/1383643/seven-weird-food-predictions-from-the-past-including-churchills-lab-grown-chicken-wings/

Ben Goertzel 2008: ‘My own (Ben Goertzel’s) personal intuition is that a human-toddler-level AGI could be created based on OpenCogPrime within as little as 3-5 years, and almost certainly within 7-10 years.’
https://opencog.wordpress.com/2008/07/

Ben Goertzel 2017: ‘I’ll be pretty surprised if we don’t have toddler-level AGI in the range 2023-25, actually.’
https://www.nextbigfuture.com/2017/12/ai-researcher-ben-goertzel-launches-singularitynet-marketplace-and-agi-coin-cryptocurrency.html

Ilya Sutskever, co-founder and research director at OpenAI, predicts that the possibility of AGI be created in as little as five years “can no longer be discounted.” Skip to the 27:00 mark in his speech:
https://youtu.be/w3ues-NayAs

Stephen Hsu, physicist and all-around smart dude: ‘So a timescale > 30-50 years for AGI, even in highly optimistic scenarios, seems quite possible to me.’
http://infoproc.blogspot.com/2018/09/intuition-and-two-brains-revisited.html

In 2013, some guy built a perfect Tetris-playing AI. It clears lines from the screen just as fast as new pieces fall from the sky. As far as anyone can tell, it could play forever.
https://codemyroad.wordpress.com/2013/04/14/tetris-ai-the-near-perfect-player/

The hype about self-driving cars is dying down.
https://www.washingtonpost.com/technology/2018/10/18/shaken-by-hype-self-driving-leaders-adopt-new-strategy-shutting-up/

The Tesla Model 3 is, by some measures, the safest car ever tested by the NHTSA.
https://www.tesla.com/blog/model-3-lowest-probability-injury-any-vehicle-ever-tested-nhtsa

A teardown of a Tesla Model 3 reveals it’s an extraordinary piece of technology, but its process of manufacture could be sharply simplified.
https://youtu.be/Lj1a8rdX6DU

ReThink Robotics, the company that made the “Baxter” general-purpose robot, went bankrupt due to low sales and sold itself to a German company.
https://www.therobotreport.com/hahn-group-acquires-rethink-robotics-ip/

We don’t know if there’s enough CO2 sequestered in Mars’ soil to create an atmosphere via terraforming machines. I think that, by the time we have the ability to send large numbers of people to Mars, intelligent machines will probably dominate Earth and cancel any senseless plans to send more than token numbers of resource-hogging meatbags like us there. Multi trillion dollar plans to terraform Mars will also be considered too wasteful to proceed.
http://nautil.us/issue/65/in-plain-sight/so-can-we-terraform-mars-or-not

A simple explanation of how the Asteroid Belt, Kuiper Belt and Oort Cloud are different.
http://ryanmarciniak.com/archives/390

‘The total mass of the asteroid belt is estimated to be between 2.8×10^21 and 3.2×10^21 kilograms, which is just 4% of the mass of the Moon.’

That sounds small until you think about this: A Ford-class aircraft carrier is 9.1 million (9.1×10^6) kilograms. A space warship ten times that size–which is in the same ballpark as a Star Destroyer–would thus be 9.1×10^7 kg. If we had space factories and converted just ONE PERCENT of the asteroid belt’s mass (I used the lower of the two estimates) into space warships, we could build 30.7 QUADRILLION ships.
https://en.wikipedia.org/wiki/Asteroid_belt

In theory, a particle accelerator wider than our Solar System could unlock all the mysteries of physics and cosmology.
https://gizmodo.com/we-could-solve-the-mysteries-of-time-and-space-if-we-ha-1829207595

A Russian Soyuz space rocket exploded after takeoff, but miraculously, the two astronauts it was carrying safely ejected.
https://www.npr.org/2018/10/11/656473889/rocket-launch-failure-forces-astronaut-and-cosmonaut-to-make-ballistic-landing

NASA’s “Space Launch System” (SLS) heavy rocket is in even worse shape than thought.
https://www.washingtonpost.com/technology/2018/10/10/program-build-nasas-moon-rocket-could-double-price-billion-ig-says/

Yes, we still have the blueprints to build Saturn V rockets, but no, it wouldn’t make sense to make more of them for many reasons.
https://space.stackexchange.com/questions/6281/why-not-build-saturn-vs-again

In the 1990s, the “DC-X” experimental rocket did test flights that proved rockets could vertically land and be 100% reusable. However, the technology wasn’t commercialized for over 20 years.
https://youtu.be/39cjZTCay24

The “To the Stars Academy,” which was co-founded by the former lead singer of Blink-182 and which publicized last year’s big UFO report, says it has very few assets and huge debts.
https://www.sec.gov/Archives/edgar/data/1710274/000114420418050766/tv503167_1sa.htm

Scientists have created cheap, artificial wood that is better than real wood.
https://www.scientificamerican.com/article/scientists-create-artificial-wood-that-is-water-and-fire-resistant/

There have been recent technological breakthroughs in our ability to view the three-dimensional structures of small molecules.
https://blogs.sciencemag.org/pipeline/archives/2018/10/18/small-molecule-structures-a-new-world

Lowered plane travel costs and the growth of the global middle class have caused the number of tourists to explode. The trends will only continue, and I fear someday all the best places in the world will be overrun.
https://www.cnn.com/travel/article/overtourism-solutions/index.html

I predict that “Choose-your-own-endings” like this will become a common form of entertainment in the future. To appease different factions of fans, the same TV series will exist as “parallel universes” where the plots diverged at critical junctures. A mix of viewer focus groups and instant surveys will guide each divergence, and fanfiction crowdsourcing and AI will pick up the slack writing the multiple scripts. The logical endpoint of this is entertainment custom-tailored to individual people.
https://www.bloomberg.com/news/articles/2018-10-01/netflix-is-said-to-plan-choose-your-own-adventure-black-mirror

October 31, 2018January 31, 2019

What would a robot tank look like?

Bigger tanks may be scarier, but they are also bigger targets. Less is more.

Lately, I’ve read about some interesting experiments to jury-rig helicopters and tanks to be autonomously controlled:

I’m unsure if any of these projects will get past the prototype stage, but they’re fascinating nonetheless since, if adopted, they could extend the useful lives of many old pieces of military hardware by enhancing them with machine intelligence and, maybe in some cases, with robot crews. As with all types of jobs, those in the military will inevitably be taken over by machines of some sort, and drop-in-kits installed into the cockpits of old helicopters could be the “bridge” in that transition.

However, in the longer run, planes, tanks, ships, and other pieces of military hardware will be redesigned around the needs of machines instead of humans. Returning to the helicopter example, a clean-sheet design meant to be flown by a computer wouldn’t have a cockpit at all: its shoebox-sized computer “pilot” would just need a small, armored compartment in the fuselage, which could be accessed through a little door. Deleting the chairs, controls, displays, and windows that a human pilot/co-pilot duo would have needed would make the autonomous helicopter lighter, sleeker, faster, and cheaper to make. In a fight between the old helicopter jury-rigged with a drop-in computer and the new helicopter designed specifically around a computer pilot, the latter would win.

This makes me wonder what a “robot tank” that was as good as the best modern tanks would look like. A tank’s quality is determined in aggregate by its 1) firepower, 2) speed, and 3) armor, so the theoretical robot tank will need to equal or surpass the U.S. M1 Abrams or Russian T-14 Armata. For this exercise, I think the best place to start is with the first criterion, “firepower.”

The Abrams tank has a 120mm diameter main cannon and the Armata has a 125mm cannon. Their capabilities are about the same. Our robot tank would thus need one or other. Now, a tank’s cannon sets a minimum diameter for the tank’s “turret ring,” which is the big hole in the top of the tank’s hull that the turret is dropped into in the factory. As a general rule, bigger cannons need wider turret rings. And for obvious reasons, the width of the turret ring in turn sets a minimum width for the tank’s hull.

At this WWII American tank factory, the tanks in the foreground have not had their turrets installed, so their turret rings are visible. Fully assembled tanks are visible farther in the background.

A U.S. Sherman tank that was destroyed by a large land mine in Okinawa. Note how the turret has fallen off of the hull.

In the early 1980s, the U.S. Army built an experimental version of the M1 Abrams tank that had an unmanned, 120mm turret. They dryly named this vehicle the “Tank Test Bed” (TTB). It was never put into mass production, but its weapon reportedly worked very well. I couldn’t find figures on the internet, but eyeballing the photo below, it looks like the diameter of the TTB’s turret ring is about 80% of the tank hull’s diameter. The M1 is known to be 12 feet wide, so I’ll estimate the turret ring is 9.6 feet in diameter. To add the necessary structural support for the turret ring and space for side armor, let’s make our robot tank’s hull 11 feet wide.

Take note that the Russian T-14 Armata has a 125mm, unmanned turret, and the vehicle’s overall width is 11.5 feet (3.5 meters), which suggests my estimate is credible, and that the Russians might have made a robotic tank gun that is as compact as physically allowable. Note that the T-14 Armata has a three-man crew, and they are seated in a row inside the hull, so it’s possible the tank’s width was determined by human considerations rather than mechanical ones. That said, I’ll stick with my assumption that a robot tank could be a slightly thinner 11 feet wide thanks to the deletion of space-consuming humans with their huge shoulder widths.

Diagram of the T-14 Armata tank. The three-man crew sits shoulder-to-shoulder in the hull.

As this very long, very awesome RAND report says, the optimal ratio between a tank’s width and length is 1 : 1.5 (ignoring the length of the cannon). Our hypothetical robot tank’s length should thus be 16.5 feet (198 inches). Looking at modern tanks, however, I see the ratio ends up more like 1 : 2.16 (M1 Abrams), 1 : 2.49 (T-14 Armata), 1 : 2.08 (Leopard 2), and 1 : 1.8 (T-90), which suggests to me that there’s some other design constraint forcing tank engineers to make their vehicles longer than they should ideally be. And you guessed it, I think the extra length owes (mostly or wholly) to the need for interior crew compartment(s) for the humans. Look at this diagram of a T-90, to which I added a Little Green Man to indicate where the driver sits.

Sitting down, the driver adds about 3 feet to the tank’s overall length. Subtract that, and the T-90 ends up being 19 feet 6 inches long, giving it a width : length ratio of 1 : 1.56, which is very close to optimal. Additionally, shortening it by that amount might allow for the elimination of one set of road wheels (down from six to five), cutting weight and cost, though I think there’s also a risk that could make the tank unbalanced and back-heavy.

Since I don’t know what all the different design constraints are, I’ll give a range of possible lengths rather than a hard number: The hypothetical robot tank could have a width : length ratio of anywhere from 1 : 1.5 to 1 : 1.7 (slightly less than the T-90), which means its length would be 16 feet 6 inches up to 18 feet 8 inches.

Estimating the robot tank’s weight is harder still, but I’ll give it a shot. The RAND report has weight figures for three types of tanks: the four-man M1 Abrams (66 short tons), a hypothetical three-man tank (61 short tons), and a hypothetical two-man tank (55 short tons). Graphing those on a simple line chart yields the following:

As you can see, the elimination of each human crewman allows a roughly five ton weight reduction, and extending the trend to zero crewmen, our robot tank should weigh 44 short tons. It’s fair to scoff at this straight-line extrapolation as overly simplistic, but consider this: the T-90 is longer (22 feet 6 inches) and wider (12 feet 5 inches) than our hypothetical robot tank (16 feet 6 inches up to 18 feet 8 inches long and 11 feet wide). The T-90 weighs 51 short tons, and trimming length and width to make something the size of our robot tank could absolutely result in a new weight of 44 short tons. A real-life datapoint supports my back-of-the-envelope line graph extrapolation.

In terms of height, the robot tank couldn’t be much lower than the T-90, which is 7.3 feet tall (not counting the machine gun and any sensors mounted on the top of the turret). Returning to the T-90 diagram, this time with a Little Green Man drawn to represent the commander, we can see that the turret might be a little taller than is mechanically necessary because it must accommodate a human. However, the height of the hull can’t be decreased since it is constrained by the height of the engine (outlined in red). The T-90 also can’t have its suspension lowered without sacrificing ground clearance and damaging its cross-country performance.

T-90 with commander’s position and engine highlighted

Aside from flattening its roof, there doesn’t seem to be any good way to make the T-90’s turret lower. The diameter of the cannon and the height of its breech establish a hard limit on how low the turret can be. Additionally, the turret’s ammunition carousel (shown in the diagram below) stacks the shells and propellant horizontally, which already minimizes the overall height of the carousel. There doesn’t seem to be any way to shrink it further.

The “carousel” of a T-90 tank, which stores ammunition for the cannon.

I estimate that getting rid of the bulge at the top of the turret would reduce the T-90’s height from 7.3 feet to 6.5 feet. The hypothetical robot tank would use the same type of autoloading turret and hence would be the same height.

So there we have it. I estimate that an autonomous tank equivalent to today’s best manned tanks in terms of firepower, speed and armor would have the following specifications:

Length	16 ft 6 in (min) to 18 ft 8 in (max)
Width	11 ft
Height	6 ft 6 in
Weight	44 short tons
Armament	120mm or 125mm cannon with autoloader
Armor and mobility	Comparable to T-90 or M1 Abrams

This is close to the Chinese Type 59G tank (a modified version of which is used by Pakistan):

Length	19 ft 10 in
Width	10 ft 8 in
Height	8 ft 6 in
Weight	40 short tons
Armament	125mm cannon
Armor and mobility	Inferior to T-90 or M1 Abrams

Pakistan’s “Al Zarrar” tank, which is an upgraded version of the Chinese T-59G tank.

The fact that tank with similar dimensions and firepower to my hypothetical robot tank already exists shows that there aren’t any engineering or practicality barriers to building the robot tank per my specifications. The Type 59G proves that a 125mm cannon can be mounted in a relatively small, lightweight hull and fired without tearing the vehicle apart. I don’t have the software or artistic talent to make a computer rendering of the robot tank, but combining the hull of a Type 59G with the lower-profile turret of a T-90 or the narrow, “naked turret” of the M1 Abrams TTB would give a fair approximation of its appearance.

Since the robot tank would be much smaller and lighter than an equivalent tank built around a human crew, it would be cheaper to manufacture, harder to hit since it would be a smaller target, and more easily transportable. A computer would take the place of a human commander, gunner, and driver, allowing for a significant reduction in internal volume and overall tank size. A space-efficient autoloader already found in the T-90 would be incorporated into the robot tank. Additionally, to perform maintenance and repairs in the field, the robot tanks would need to internally carry two smaller, human-sized (but not necessarily humanoid) robots, but they could be stored much more compactly than human crewmen during transit. They might be able to curl up into fetal positions and fit into small lockers in the back or sides of the vehicle, or in a bustle at the back of the turret.

Even outside of combat, tanks require frequent maintenance and repair. A robotic tank would still probably carry smaller, human-sized robots in it that would go exit the vehicle to do the work.

Finally, I think the robot tank would carry a small UAV that it could launch to provide aerial reconnaissance footage, vastly improving the tank’s situational awareness. Something as simple as a 25-pound quadcopter could do.

Links