How does James Webb telescope work? What will it see?

Space is cold. We’ve all seen people freeze solid when blown out of the air lock (at least in movies and TV). And we’ve all heard ludicrously low temperatures quoted by science communicators: -270 degrees Celsius (-450 Fahrenheit), or something close to that depending on whether you’re next to a relatively “warm” star. But space isn’t cold. Space is . . . space. It’s nothing, or near enough to nothing to not really matter. That makes the concept of temperature in space, in some sense, complete nonsense.  Because temperature is a measurement of the amount of energy in some kind of matter. Air on earth is a delicious mixture of nitrogen, hydrogen, and oxygen. It’s a bunch of atoms that, when they absorb energy, vibrate faster and faster, and thus get hotter. If they lose energy, they slow down, and get cooler. That is temperature. In space, no one can hear you scream, because there is no air. And no temperature, really. That means that space is not, in fact, cold.  Earlier this year, SpaceX founder Elon Musk announced he’s planning to put AI data centers in orbit to take advantage of the “free” cooling and solar power. Amazon founder and tech oligarch Jeff Bezos quickly followed, probably hoping his Blue Origin rockets could be used to launch Amazon AI servers into space. But there are some major issues with putting AI data centers in space, starting with the “free” cooling. Cooling things in space, it turns out, is very difficult. Because there isn’t any stuff in space, heat energy doesn’t really have anywhere to go. On earth, a hot object can dump its energy into the surrounding medium, whether that’s air or water. Imagine a vacuum-insulated coffee mug: The inner container is separated from the outside air by empty space. Heat can only escape through the lid, which is why your coffee remains blisteringly hot for so long. In space, everything is in a vacuum-insulated thermos. In space, objects can only get rid of their heat by radiating it via a form of light: Infrared radiation. If you’ve seen the Predator, you’ll be familiar with infrared. It’s the wavelength that the Predator senses, allowing him to “see” the temperature of his prey. This makes visual camouflage useless, but thermal camouflage very useful, as Arnold discovered by applying a generous coating of cool mud to his skin. This also illustrates how easily infrared radiation can be blocked. In fact, it can be blocked by any solid object. Arnold would only become visible to the Predator when his body heat warmed the mud coating, which would then release its own infrared radiation for the Predator to home in on.  Infrared is not very good at transferring heat. In fact, it’s the least efficient method of heat transfer, behind conduction and convection. Conduction is the most efficient, the transfer of heat energy directly from one object to another via contact. Convection, heat transfer through the movement of fluids, is next. Radiation is last, and is far slower and less efficient than the other two. To get an idea of how bad radiative cooling is, we can deconstruct one of the most pervasive tropes in science fiction: Instantly freezing when hurled out of an airlock. While you would die quite quickly due to lack of oxygen and external pressure, you would not freeze solid in a matter of seconds. In fact, according to astrophysicist Paul M. Sutter in an article for Forbes, it would take anywhere between 12 hours and never to freeze solid in space. “Never” applying to anyone close to a warm-ish star or simply in strong enough starlight.  To get a better idea of how difficult it is to cool stuff in space, we can look at the International Space Station. The station was launched in pieces starting in 1998 and was assembled in orbit over the next several years. Now the station has between 16 and 20 pressurized modules (depending on how you count them) that have an internal pressurized volume of 1,005.0 m3 (35,491 cu ft). That’s about the same internal volume as a Boeing 747 airplane. Altogether, living and working spaces occupy about 520 m2 (5,600 sq ft), about the same size as a six-bedroom house. The whole station is about the length of a football (soccer) field and has a mass of 460 tons.  Such a massive orbital platform needs a serious cooling system to keep astronauts comfortable and to cool their scientific equipment. Designed by Boeing, the Active Thermal Control System (ACTS) circulates water throughout the station in closed loops to keep things cool. That water exchanges its heat with liquid ammonia, which flows through two gigantic radiators. Liquid ammonia is used for its high thermal capacity—it’s ability to absorb heat—and it’s extremely low freezing temperature. Each radiator consists of eight panels and is 23.3 m (76.4 ft) long, 3.4 m (11.2 ft) wide, and weights 1,122.64 kg (2,475 lbs). This is the ISS main cooling system, but there are several other systems distributed around the station to keep various components cool. In short, keeping the ISS cool requires tons of complex mechanical equipment and huge radiators.   And the ISS isn’t a machine whose primary job is turning energy into computation and heat. In general, a computer will emit as much heat as it consumes in electricity. For example, a 900W server will emit about 900W of heat energy, according to the data center specialists at Server Room Environments in the U.K. When people talk about putting data centers in space, they’re really talking about AI data centers that use energy hungry graphics processing units (GPUs). According the server power equipment provider Schneider Electric, a single refrigerator-sized rack of these GPUs in an AI data center can consume as much as 400 kW, and today’s AI models need multiple racks to tackle complex tasks. The giant one-tonne ISS cooling system has a total heat rejection capability of just 70 kW. Of course there have been many technological advances since then, and space radiators can be made lighter. But they still face the real problem of inefficient heat transfer via infrared radiation. SpaceX, currently the biggest cheerleader for putting data centers in space, have created a concept for an AI data center satellite based on the current Starlink satellite. The concept satellite is 30 m (98 ft) tall with a 75 m (246 ft) wingspan and is projected to have a 175 kW computing capability. The company says the data center satellite will use custom-designed AI chips that are more efficient and rated to withstand radiation and the harsh conditions of low-earth orbit. The concept satellite would use a 160 m2 foldable radiator and a set of solar panels good for 210 kW of power. SpaceX envisions swarms of these AI satellites communicating with each other with lasers to create space-based data centers. The concept is straight out of a William Gibson novel and is undeniably cool in a terrifying dystopian sort of way, but may not make financial sense. According to the “space economy” trackers at Orbital Radar, the cost to launch a kilogram into orbit is currently about $3,000 via the SpaceX Falcon 9. This is a drastic improvement over the Space Shuttle, which cost about $54,000 per kilogram, however that’s probably not a fair comparison as the shuttle was a whole reusable spaceship with a crew aboard. SpaceX doesn’t list the weight for their AI satellite, but according to FCC filings, the latest Starlink satellites weigh between 2,000 to 2,500 kg (4,400 to 5,500 lbs). SpaceX AI satellites will likely weight more. Based on these numbers, it will currently cost approximately $6.7 million to launch one of these new Starlink Satellites into low-earth orbit. It’s true that SpaceX plans to drop the cost to orbit to less-than $1,000 with its giant Starship reusable rocket, but that rocket is still in development and has exploded on the launchpad several times.  Launch costs are just part of the overall price tag for orbital data centers. SpaceX does not say how much each AI satellite will cost to build, but a 2024 Motley Fool article estimated that a single Starlink satellite costs approximately $500,000—and newer satellites may cost up to $1 million each. The new AI satellites, with custom chips, more cooling, and larger solar panels, will likely cost more. In the data center space, costs are measured in dollars per kilowatt of computing power. This gives data center builders and operators a rough idea to project expenses, revenues, and profits. Doing some very rough, cocktail-napkin math based on these estimates, I calculated that the SpaceX AI satellites would conservatively cost somewhere around $46,000 per kilowatt of computing power to put into orbit. According to “leading global provider of integrated, technology-enabled commercial real estate services and investment management” Jones Lang Lasalle Incorporated’s 2026 Data Center Outlook report, data center constructions costs between 2020 and 2025 on earth were about $10,000 per kW of computing power. The same report estimates that AI data center equipment could cost up to an additional $25,000 per kilowatt. So right now earthbound AI datacenters cost around $35,000 per kilowatt to build. There’s a lot more math to be done, but it seems that SpaceX’s proposed AI satellites are conveniently close to being price competitive with land-based data centers, plus they would theoretically enjoy free cooling and power. But it’s still unclear how often these satellites would need to be replaced or repaired, which would incur incredible costs. Launching a human into space to fix a satellite is far more expensive than sending an electrician down to the data center. Likewise, replacing a single rack here on earth costs far less than launching a new AI satellite.  We must also consider the environmental costs of rocket launches. It is theoretically possible to build and run AI data centers here on earth using nothing but solar, wind, and geothermal power sources, but it is currently impossible to launch anything into orbit without releasing vast amounts of CO2. According to an article published by BBC’s Science Focus, a Falcon 9 burns 112 tonnes of refined kerosene, emitting about 336 tonnes of CO2. That’s as much emitted by the average car driving almost 70 times around the world.   But there’s an even bigger problem with the concept of data centers in space. Large Language Models (LLMs) appear to be reaching the upper limits of their abilities. A recent study published by MIT defined the hard mathematical and financial limits of scaling LLMs. These models absorb vast amounts of text and create mathematical maps of how words relate to each other. Because human language is so complex, these models eventually run out of space to store all the concepts. Instead of just throwing away less-used concepts and connections to make space, LLMs sort of cram them together into what MIT calls superpositions. When the model reaches for a specific concept in one of these superpositions, it can get something unexpected.  Researchers think this is the source of AI hallucinations. To reduce the number of superpositions and hallucinations, AI companies simply give their models more space. For the past six years, this has worked well. Each successive model has been more accurate than the last. But there are diminishing returns to scaling models. According to the MIT paper, every time a model doubles in size its error rate is cut in half. The cost of doubling a model, however, increases by six times. According to the paper, a frontier model today costs about $300 million to train. Doubling that model to halve its error rate would cost $1.8 billion. Double it again for a price tag of $11 billion. Double again, $65 billion. One final time after that puts the cost of training a model at a staggering $390 billion.  Even if a company is willing to pay that kind of money to build an LLM, it would run into yet another limit: Zipf’s Law. This fun-sounding law essentially says that in human language a small number of words are used a lot while a massive number are rarely used. The words “is” and “the” are used all the time while “photosynthesis” and “kaleidoscope” are rarely used. As models get bigger, they spend a disproportionate amount of time learning high-frequency word patterns compared to low-frequency word patterns. Those rare words and concepts are under sampled and remain fuzzy and unclear. The time it takes a model to fully map those fuzzy concepts gets longer and longer. Progress virtually slows to a stop. All of this means that LLMs have a hard limit. That doesn’t mean they aren’t useful, but it does mean we may be fast approaching the point at which it doesn’t make sense to spend more money to make them bigger and more accurate. Instead, AI researchers are exploring different ways of making AI smarter, more accurate, and more efficient. These new methods may very well require more computing power, but they may not. In China, computer scientists were recently able to create much more efficient models that perform nearly as well as top-of-the-line models in the U.S. So there are financial and mathematical limits to scaling LLMs, but there are also financial limits to the entire AI industry itself. According to Goldman Sachs, world investment in AI will exceed $1 trillion this year (2026). That’s more money than we invested in things like the U.S. railroad system and the internet. And as with any private investment, investors expect to make their money back, plus profits. So far it seems that the big AI companies aren’t earning enough money to even cover what they’ve spent on data centers and training, let alone enough to turn a profit fast enough for investors. In June of 2026, independent journalist Ed Zitron obtained leaked financial documents from AI giant OpenAI. While the documents showed OpenAI earning $13 billion of revenue in 2025, its losses were tremendous. Expenses from model training, research and development, and sales and marketing grew from $8.78 billion in 2024 to $20.92 billion in 2025. That means OpenAI burned about $8 billion in 2025. Recently leaked Anthropic financial documents were even worse. The AI company lost $42 billion last year while only earning $4.6 billion, lighting an incredible $34 billion on fire. Elon Musk’s XAI, which recently went public, posted an operating loss of $2.47 billion on $818 million in revenue in Q1 2026. The big AI companies are not making enough money to justify the already massive investment in the technology, let alone enough to launch AI satellites. Then there’s the Kessler Syndrome. In 1978, NASA scientists Donald J. Kessler and Burton G. Cour-Palais described a scenario in which the amount of orbital space debris becomes so great that it destroys all satellites in orbit. When a rocket or space shuttle blasts into orbit, it inevitably leaves bits of itself behind. And satellites and space stations also shed bits and pieces over time. All this space junk orbits the planet at around 7 to 8 m/s, or about 15,000 to 18,000 mph. For context, a typical .223/5.56 mm bullet fired from an AR-15 style rifle travels at about 1,840 mph. Space junk, then, moves about 10 times as fast as a rifle bullet, and has 10 times the energy. When it strikes a satellite, space station, or other piece of space junk, it explodes, creating more space junk. Over time, this process repeats and the earth becomes surrounded with a deadly cloud of space junk that destroys any and every satellite. No GPS, no satellite TV, no satellite communications systems, no spy satellites. Nothing in orbit survives for long without being shredded by high-speed space junk.  Currently, several government and private agencies track all the space junk floating around up there in orbit, including NASA, US Space Surveillance Network (SSN), the European Space Agency (ESA), China’s Space Situational Awareness agency, and more. All that data is given to satellite operators, who use it to find safer orbits, and to maneuver satellites out of harm’s way. For now, the system has been very effective and satellite operators are able to avoid deadly space debris. Scientists don’t agree when the Kessler Syndrome may kick in, but in 2023 two researchers from the University of Málaga in Spain did a lot of math and came up with a maximum number of satellites we can deploy without triggering orbital armageddon: 72,000. Beyond that, they estimate, Kessler Syndrome is inevitable. According to Orbital Radar, there are currently about 31,000 tracked objects in orbit. This includes operational satellites, defunct satellites, debris fragments, and discarded rocket bodies. There are probably more objects that aren’t being actively tracked (yet), so it’s tough to estimate how close we are to orbital doom.  All of this makes orbital data centers a questionable idea at best. But the AI boom is not logical, it’s more of a cult. In the minds of many tech leaders, AI is an all-knowing God, a technological savior. OpenAI’s Sam Altman and Google’s Eric Schmidt have both argued that the carbon cost of building AI is irrelevant because it will “solve” climate change by inventing miraculous technology. They and others literally believe that AI will eventually gain consciousness and start an exponential self-improvement loop to eventually become a god-like superintelligence. There is no evidence, however, that any of the current bodiless AI systems are close to consciousness, as we do not have any scientific definition of consciousness. We all have a feeling of what it is to be, a kind of experiencing of the world and the sense that we are in control of our own actions. However there are many compelling arguments against free will, see Robert Sapolsky. Plus, neuroscientists still have very little understanding of how the human brain works, and computer scientists have even less of an understanding of this goopy analog thing between our ears. The idea of AI becoming a god is based on faith and that computers are really good at things like math and statistics. The fact that LLMs can whip up a good approximation of expertise also reinforces this faith.  I know computer scientists and AI experts will say, “but AI is much more than LLMs and statistical inference, much more!” I agree, it is much more, but the human brain is many times more again. There are about 86 billion neurons in a human brain, each with multiple connections, all intertwined with billions of other cells throughout the body through the nervous system. The biochemical systems included in and layered on top of the nervous system use hundreds of chemical compounds for communication, regulation, and more. The sheer complexity of a single human mind-body system dwarfs that of any existing or theoretical supercomputer. Any simulation would thus be an approximation, something entirely new and likely nothing like what we experience as intelligence.  That doesn’t mean AI can’t or won’t solve problems, but it does mean that we should think deeply about how it will affect our world. We have built technology with wild abandon for the past 100 years, forever changing the world that we evolved to live and thrive in. We are just beginning to see the effects of that mad, unthinking technological expansion, and they are not good. Believing that further unhinged technological expansion and growth will solve these problems is insane. Just like the idea of data centers in space.

When I was about eight years old, I became irrationally upset by the fact that I had all these fantastic stories and ideas swirling around in my head that required real work to become, well, real. Why was I gifted with such a fabulous imagination in a world that completely ignored my thoughts? I even wrote an essay about it, titled something like, “On the Extreme Frustration of Not Being Able to Manipulate Reality with my Mind.” As painful as this realization was, I persevered and eventually either learned to do the work or accepted the fact there are some things I simply cannot do. Today, though, it feels like we’re all at the mercy of that 8-year old’s irrational feelings in the form of GenAI. In the minds of many, GenAI represents a kind of magical power that bypasses the messy and difficult process of craft and simply creates the wildest dreams out of thin air. It’s an undeniably alluring idea, which is one reason the whole world is shoveling money into it like a stoker shoveling coal into a runaway train’s boiler.  But for many of us creative types, the craft is the fun part. We enjoy tediously picking away at prose to make it tighter, more understandable, more relatable. We love pushing pixels to make our digital art perfect on a level no one else will ever notice. The craft is cathartic and human and the reason we all got into this in the first place. It brings us joy and peace, distracts us from our overactive and hypercritical minds. So our hearts fall and our souls whither when they tell us, “just put that paragraph in ChatGPT and ask it to streamline the text.” Because for us, pouring over every word and space in a paragraph is the good part of the job. GenAI is made to optimize outcomes and maximize profits. Both of these are of course necessary in a world driven by profits, but creative professionals try to at least shoehorn some humanity into the work, to create some kind of connection. And making those human connections requires human effort, not just probabilistic guesses from a machine. It requires careful thinking and arduous tinkering. This toiling and tinkering also creates a sort of insulation between us (the artists) and the ultimate purpose of our corporate/commercial work. If we can focus on these tiny turns of phrases or minute adjustments to kerning, we can create distance between us and the cold machinery of capitalism. This is the creative process and it’s why we do what we do. This desire to eliminate creativity in the name of production is why I loathe GenAI. I also hate how it has stolen the work of countless humans, dumped it into a mathematical meat grinder, and excreted unimaginable amounts of inedible creative sausage. I am also keenly aware that this process consumes gigawatts of energy and millions of gallons of water (even if it isn’t strictly drinking water). GenAI is not only anti creativity, it’s horrifically inefficient. According to the International Energy Agency, a typical AI-focused hyperscaler data center uses as much electricity as 100,000 households. In comparison, a single human brain, which is capable of everything from scoring 100 points in a basketball game to coming up with General Relativity, consumes about 20 watts. It can also navigate the real world, remember things, and have experiences. The human brain is a miracle and to offload so much of its creativity to an inefficient machine in the name of production and progress is heartbreaking and criminal.  Don’t get me wrong, I believe there are real uses for machine learning and GenAI. Both technologies can be used to model and improve systems and can be invaluable in doing many things in science, like identifying new galaxies in old telescope data or finding cancer cells hidden in medical scans. It can be a tremendous boon for any task that requires sifting through unimaginable amounts of data, or for helping humans comprehend huge, complex systems. It can help in ways I can’t even begin to fathom because I’m not a scientist or engineer. (But even then, scientists tell me, it needs to be thoroughly checked to make sure it isn’t lying, hallucinating, or is just plain incorrect.) In my realm of writing and storytelling, however, GenAI is wrong. Tedious research, real-world experience, and the nitty gritty of the craft are critical to writing anything anyone would want to read. Ultimately, though, GenAI is not truly the problem and the way it’s being used isn’t a surprise. If you prioritize growth, extraction, and profit, you’ll end up with GenAI that vomits countless iterations of Marvel movies into the hungry maws of a hollowed-out populous. It’s an almost inevitable outcome, just one of the beast’s many fierce weapons. So banning GenAI in creative professions is like trimming one claw on a tiger’s paw. If the cat’s hungry, it’s still gonna tear you open and eat your guts. The real problem is we’ve let striving, ravenous, insatiable urges take over everything. Worse, we’ve built our entire civilization around them. They are our core pillars. They are the primary elements of our societal chemistry. Everything must be done now, we must grab all we can before someone else does, we must rush rush rush and grow grow grow. Greed is good. Grabbing hands grab all they can, etc. etc.  Greed corrupts and encourages corner cutting, rushing, rashness, and cut-throat behaviour of all kinds. Yes, this is a tremendously oversimplified, kindergarten/Sunday school observation, but it’s true. And even more frustrating, it’s a truth we’ve known for at least 10,000 years, a story that’s been told since we figured out how to scratch little symbols into stone tablets and probably even before then. Yet here we are, our entire civilization running on greed. So until we in the very least attempt to control greed, we will get things like GenAI that will turn creativity into commodity. We will get the grinding, soulless, thoughtless process of business. We need to move beyond it, to start seriously recognizing what we want as humans, what will best fill the void that we all have in our souls (hint, it’s community, art, music, spirituality). We need to consider how we can create a stable, long-lasting civilization that’s in harmony with our planet. We need to slow down, share, and be kind. When we start doing that, fewer people will want to use GenAI to make art. They will have the time for deeper thought, longer practice, and for experiencing the satisfaction of mastering a craft. Or they will find something else that fills that hole—caring for someone, doing heavy physical labor, playing a sport, learning how to play the accordion, I don’t know. Anything that requires repetition and commitment.  So I’m in a sticky situation. I realize that GenAI is abhorrent. I also know it can make the people who pay me very happy. I can use it to make these people happy, to ultimately feed my family and my pride of cats while my soul rots. Or I can try to find somewhere that refuses to use GenAI for writing, some weird business or organization that’s okay with not making money or getting more shit done. I can switch careers entirely, start at the bottom of the ladder doing something seemingly beyond the reach of GenAI like carpentry or plumbing or electrical work. (Some of those professions even have unions!) But I’m old, and frail, and my eyesight is going, and I am becoming increasingly fat and sore all over.  No, right now the best solution is to smile and diligently do my job, focusing on what makes me special and more valuable than GenAI (A sickening sentence to write as humans should always be more valuable than machines, but so it goes). I will also create, even if it does feel futile. My creative human work can create the connections I need to survive and even defeat GenAI. And who knows, one day it may even help us move away from greed and progress for progress’ sake and toward a calmer, more measured, and more humane society.

In Neal Stephenson’s science fiction masterpiece The Diamond Age: Or, A Young Lady's Illustrated Primer, microscopic nanobots build fantastic structures and extravagant objects out of diamond. They assemble mile-high buildings atom by atom, constructing colossal translucent towers stronger and more durable than anything we can imagine today. Of course those opulent towers are only accessible to a select few elites and the rest of humanity lives in nightmarish toxic squalor, but the theoretical technology is still fascinating. And its based on the very real and very strange properties of diamonds.   Diamonds are made of carbon, the fourth most abundant element in the universe, and the second most abundant element in your body. It is the key element of life, the backbone of organic chemistry. Carbon is special because it so easily bonds with other elements to create a huge variety of compounds and materials, everything from simple sugars to bones. Of course most recently it has gained fame as the core component of carbon fiber, a super-light and super-strong material made from carbon threads and resin. Carbon fiber’s strength is due in large part to its unique arrangements of carbon atoms. Carbon can bond to itself in many ways, creating complex structures with many different properties. In carbon fiber, the element bonds together to create extremely tough chains that resist stretching. Arrange those atoms differently, however, and you get bizarre and astonishing results.    Diamond is pure carbon arranged in a cubic crystal structure. Each atom is bonded to four other atoms to form a tetrahedron, or pyramid, shape. Carbon in this arrangement is incredibly hard, in fact it’s by far the hardest material we’ve discovered. Hardness is, simply put, how much a material resists denting or scratching. This is a well-known fact; we’ve all seen cat burglars using diamond glass cutters to steal priceless artifacts, or used diamond-coated drill bits or saw blades to cut through tough materials like steel. But diamonds have other amazing properties that are often overlooked.   Diamond has the highest thermal conductivity of any material. That means it can transfer heat better than anything else, which is why real diamonds can feel warm to the touch whereas cheap knockoffs feel cool. To understand just how amazing diamond is at transferring heat, we first need to understand thermal conductivity. Thermal conductivity is measured in Watts per meter-Kelvin, W/mK. A material with a conductivity value of 1 W/mK will transfer heat at a rate of 1 watt for every degree of temperature difference between opposite faces.    That’s all pretty meaningless without a solid example, though. Pure copper, the famously good electrical and thermal conductor, has a conductivity of around 400 W/mK. Gold measures in at 327 W/mK. Diamond, however, has a thermal conductivity between 1,500-2,200 W/mK.   Why can diamond conduct heat so well? Again, the secret lies in its structure. Heat can quickly and easily propagate through its cubic crystal structure in the form of atomic vibrations called phonons. In condensed-matter physics, a phonon is a unit of vibrational energy that arises from oscillating atoms within a crystal. Heat energy zips right through diamond’s crystal lattice.    Electrons, however, can’t travel through diamonds at all. This makes them electrically non conductive, in other words an insulator. Combine high thermal conductivity with zero electrical conductivity and you get a material that’s perfect for dissipating heat away from sensitive electronics. Electrical engineers use diamond to quickly and efficiently channel heat away from sensitive circuitry. These components, called diamond heat spreaders, are used in satellites, airplanes, and other equipment that must endure extreme conditions. They’re also used in high-performance computing, LED lighting systems, electric vehicles, radiofrequency (RF) transmitters, and high-power lasers.     These diamond heat spreaders aren’t made of naturally occurring diamonds, they’re made in a lab. Powerful microwaves blast methane gas, releasing carbon atoms that crystalize around a seed layer of diamond. A diamond wafer grows slowly over time and can then be cut using lasers or other cutting tools with diamond surfaces. Diamond heat spreaders are usually just a few millimeters thick and are placed placed between heat-generating electronic components and larger passive radiators made from other materials like Silicon carbide, copper, or aluminum. Heat passes quickly through the diamond, then into the larger radiator to dissipate further.   Diamond’s other amazing properties make it perfect for extreme conditions. It’s essentially immune to corrosion and resists a wide variety of acids and other chemicals. It can also withstand incredible temperatures.   All of these properties make diamond a dream material for electrical engineers. When mixed or “doped” with other elements like boron, diamond becomes a super semiconductor. Unlike silicon semiconductors, it can withstand tremendous heat and high-frequency, high-power electricity. This makes for more efficient, smaller high-power electronics in power grids, EVs, and more. But unfortunately diamond semiconductors are extremely expensive and difficult to make. Ookuma Diamond Device in Japan, however, has been working on diamond-based semiconductors for decades and is making devices for measuring radioactivity in hazardous environments. The company’s first production facility, and the world’s first dedicated manufacturing base for diamond semiconductors, is planned for Fukushima Prefecture. The organization is also working with Hokkaido University, the National Institute of Advanced Industrial Science and Technology (AIST), the National Institute of Technology, and Fukushima College to develop diamond microprocessors and memory devices. They expect these devices to be used in nuclear decommissioning and space applications.   Diamond definitely has amazing physical properties, but it remains an exotic material despite recent advances in manufacturing techniques. The world of diamond skyscrapers described in Stephensen’s Diamond Age is still far from reality. Hopefully one day engineers will develop an efficient, effective molecular manufacturing or crystallographic technique that can be used to make diamonds in any shape, but for now we’re stuck with slow, energy intensive, persnickety methods. Still, without those methods, scientists wouldn’t have access to amazing equipment like the James Webb space telescope and more.

Many of us never got to experience it, but the air quality in LA in the ‘40s  ‘50s, ‘60s, and ‘70s was deadly. Smog hung in the air like a thick fog and many residents actually said that it hurt to breathe. Sources: https://www.kcet.org/history-society/how-los-angeles-began-to-put-its-smoggy-days-behind https://earther.gizmodo.com/environmental-regulation-has-helped-cut-deaths-from-air-1829875441 https://www.hemmings.com/magazine/hcc/2007/09/After-Burner/1509192.html   https://www.marketplace.org/2014/07/14/sustainability/we-used-be-china/la-smog-battle-against-air-pollution https://www.geek.com/geek-cetera/spaceballs-becomes-reality-canadian-company-is-selling-bottled-air-to-the-chinese-1642442/ https://www.cnbc.com/2015/12/15/canadians-are-selling-cans-of-fresh-air-to-china.html http://www.bbc.com/capital/story/20170515-the-entrepreneurs-making-money-out-of-thin-air https://www.cnn.com/2016/02/08/world/fresh-air-britain-china-bottles/index.html https://ww2.arb.ca.gov/about/history

The atoms in our bodies were forged deep within the bellies of red giant stars eons ago. And if we peer deep enough into the night sky, we can see stars just like them taking shape at the edge of the observable universe. The light from those stars is more than 13 billion years old and thanks to the expansion of the universe, it has stretched out into super-long infrared wavelengths—basically heat. To “see” that heat, you need a really big, and really complex telescope. In space.

In Star Trek the crew of the USS Enterprise zips between the stars in mere days, traveling a distance that takes light itself months or even years to cover. To do it they use a fantastic sci-fi technology called warp drive. The antimatter-powered warp drive bends spacetime for the Enterprise, making the distance between stars much, much shorter for the ship. Without it Kirk and the crew would grow old and die before they reached the outer edges of our galaxy—where a lot of the show takes place. And here on earth hundreds of years would pass. It would be a pretty depressing show.

In this off-the-cuff episode I talk about some solutions to the Fermi Paradox (where is everybody in the galaxy?), the intergalactic object ʻOumuamua, and some science fiction. Enjoy!

My current novel is about everyday people who get superpowers from amazing future technology that I totally made up. The tech simply doesn’t exist, and probably won’t in my lifetime—or ever. But I want superpowers NOW. Fortunately, there are a myriad of for-real superpower technologies out there. Sure, they’re rudimentary and rife with life-threatening side effects, but they exist. So keep reading if you want to be faster, stronger, better NOW. Most of us know what steroids and other performance-enhancing drugs do, so I’ll try to stick to little-known facts. Most of this comes from the Mayo Clinic, which has compiled a detailed list of all the drugs you can take to make you awesome at sports.

A new treatment process could make wood stronger than steel. In 1940 the de Havilland aircraft company introduced the Mosquito—a combat aircraft made almost entirely out of wood. A few years later the famously fastidious Howard Huges built the Spruce Goose out of, well, spruce (and other woods). In the ‘60s the British automotive manufacturer Marcos built its GT car using mostly plywood. Tons of other manufacturers have made aircraft, boats, and cars using wood or wood composites. But why isn’t wood construction more mainstream? Cars are steel (or aluminum). Bikes are steel (or aluminum). Skyscrapers and warehouses have bones of cold, hard steel. The stuff is simply stronger and tougher than wood. Until now.  Materials scientist Liangbing Hu at the University of Maryland has invented a method of treating wood that makes it stronger than steel, and even some titanium alloys. If their methods are proven out, we could see more cars, planes, and even skyscrapers built out of wood. It could reduce our reliance on energy-intensive steel and aluminum, giving us a low-carbon alternative.  Wait, what’s all this about carbon? I’m just going out on a limb here, but it’s becoming more and more clear that we need to actively reduce the amount of carbon dioxide in the atmosphere. Growing trees is a great way to do it. If wood can be a good alternative to steel, there’s even more incentive to grow more trees. And once that carbon is locked away in wood, it’s not going back into the atmosphere—unless you burn it. But I’m getting ahead of myself. Let’s leave this alone for now. So how does this special treatment work? First they boil the wood in a solution of sodium hydroxide and sodium sulfite. This boil removes some of the lignin and hemicellulose polymers in the wood, but leaves the cellulose intact. And cellulose is the strong stuff you want to keep around. After the boil the researchers compress and heat the wood, making a much more dense version of the original thing. The cellulose fibers actually fuse together on a molecular level, creating a material that’s much stronger than the original. The stuff is three times as dense as regular wood, fifty times more resistant to compression and just about 20 times stiffer.  In simulated tests the new compressed wood actually stopped bullets. It’s not as tough as traditional kevlar armor, but it’s only about a fifth of the cost.  So what else could you do with this new compressed wood? It’s strong enough to be used in cars (just like that Marcos), buildings, airplanes, whatever. It would be cheaper and lighter than steel, and it would be totally renewable. Sure, you can melt steel down and use it again, but it’s an energy intensive process. Need more wood? Plant some trees. If you plan it right, you could grow more than enough to meet manufacturers needs. Plus, trees are a carbon sink. Remember that bit about climate change earlier? One great way to reduce carbon concentrations is by planting trees. Sure, it takes a while, but trees capture and hold on to a lot of carbon.  Hu would like to work with engineers to scale up his new process to make it commercially viable. Who knows, maybe in the future your Tesla will be made largely from wood.  I stole this story from author Sid Perkins in Scientific American. Go give it a read. Sid goes into more detail about the chemical process and explores a few other futuristic materials made out of wood—including transparent wood.

Yes, yes you should.