Sci-Fi Lab
A story that names a machine has done half an engineer's job: it has written the requirement. What is left is to put units on it and go looking for the nearest thing already running. That search keeps coming back with something. Sound really does lift and carry a sixteen millimetre bead through open air. A wall made of nothing but hot gas really does hold a vacuum against the atmosphere. A quantum state really has been moved fourteen hundred kilometres to a satellite. Eight fictional machines here, each with the built thing that rhymes with it, and a number on the distance still to go.
Fiction goes first more often than the order suggests
The habit of treating a made-up machine as a design brief is older than most of the machines. It works because a story has to make its device do something specific, and specific is what an engineer needs.
Three relays in a high orbit
Arthur C. Clarke published a four-page piece in Wireless World working out that three stations parked 35,800 kilometres up, going round once a day, would cover the whole populated world. Syncom 2 reached that orbit in July 1963, and Intelsat I began carrying paying traffic in April 1965, nineteen and a half years after the article.
The waldo
Robert Heinlein's story of that name gave remote manipulators their nickname before anyone built a good one. Master-slave arms behind thick glass are now standard in nuclear handling cells and in keyhole surgery, and the people who use them still call them waldos.
A phone that flips open
Martin Cooper made the first handheld mobile phone call for Motorola on a New York footpath. He has said many times since that the communicator flipped open on Star Trek was on his mind, and Motorola's StarTAC of 1996 wore the shape openly.
None of that makes fiction a forecast. It makes fiction a place where requirements get written down early, in plain language, before anyone knows whether they can be met. The eight below are handled that way: the story told straight, the numbers it implies pulled out, and the closest working machine set beside it with its own figures. Where the fiction asks for more than anything built, the shortfall gets a factor rather than a shrug.
Eight machines on one ruler
Pick one and the sheet fills in. The bar underneath is how many times short the nearest built thing falls on the single quantity the story leans on hardest, drawn so that each step across the track is ten times the last, because a factor of three hundred and a factor of a thousand million million million million cannot share a picture any other way.
The warp drive
The warp drive
A ship at warp does not travel through space. The space around it travels, contracting ahead and expanding behind, and the ship rides in a flat pocket where nothing local is moving fast at all. That is how the fiction usually describes it, and in May 1994 Miguel Alcubierre, then at Cardiff, wrote it out as a solution to Einstein's field equations and published it in Classical and Quantum Gravity. The warp drive is one of the few pieces of screen technology with a published set of equations to its name.
Space really does stretch and squeeze, and the stretching has been measured
On 14 September 2015 the two LIGO detectors recorded a passing gravitational wave from a pair of black holes merging about 1.3 billion light years away. As it went through, it stretched one four kilometre arm and squeezed the other. The change in length was about four billionths of a billionth of a metre, four times ten to the minus eighteen of a metre, roughly one four-hundredth of the width of a proton. Both detectors saw the same shape seven milliseconds apart, close to the ten milliseconds light needs to cover the three thousand kilometres between them; the exact lag is set by which way the wave came in.
The larger version is going on everywhere at once. Every million parsecs of space between distant galaxies grows by about seventy kilometres each second, which is what the redshift of far-off light reports. Stretching spacetime is not the unbuilt part of the brief. It is measured, twice over, by two different kinds of instrument.
What is unbuilt is doing it on purpose, in one place, on demand. Nothing anyone has made moves spacetime by an amount an instrument could pick up.
- Instrument
- LIGO Hanford and LIGO Livingston, 4 km arms
- Arm length change
- 4 × 10-18 m, about a four-hundredth of a proton's width
- Strain measured
- about one part in 1021
- Published
- Physical Review Letters 116, 061102, February 2016
A bubble roughly 100 metres across, carrying a crew compartment, holding a steady apparent speed well above light. Alcubierre's own worked example takes a bubble of about that size and asks how much energy and pressure the bubble wall would need.
Negative energy density on the bubble wall, in bulk. Michael Pfenning and Larry Ford applied quantum inequalities in 1997 and found the wall would have to be a few hundred Planck lengths thick, with a total negative energy about ten orders of magnitude beyond the mass-energy of the visible universe. Chris Van Den Broeck changed the geometry in 1999, shrinking the outside of the bubble to about 10-15 metres while keeping the inside roomy, and brought the requirement down to a few solar masses. Harold White's sensitivity work at NASA in 2011 reported that thickening the wall and oscillating it brings the figure towards the mass-energy of Voyager 1, about 700 kilograms.
Negative energy would have to be producible in quantity, held steady, and shaped. That is a materials problem before it is a relativity problem. Alexey Bobrick and Gianni Martire argued in 2021 that a general class of slower-than-light warp shells can be built from ordinary positive energy, and a 2024 solution by Jared Fuchs and colleagues gives a constant-velocity shell that satisfies every energy condition. Whether such a shell can be accelerated, and at what cost, is open.
Gravitational wave detectors are already the right instrument. Katy Clough, Tim Dietrich and Sebastian Khan simulated what a warp shell losing containment would radiate and published the waveform in July 2024: a short burst, at frequencies above the band LIGO listens in. Building detectors for that higher band turns a thought experiment into a search with a null result or a signal at the end of it.
The gap: the most optimistic published requirement, 700 kilograms of mass-energy, is 6.3 × 1019 joules, about five weeks of everything the world uses. But it has to be negative. The largest negative energy density anyone has produced sits between Casimir plates, and a stamp-sized pair at a hundred nanometre gap holds about 43 picojoules of it. The ratio is 1.5 × 1030: thirty steps of ten, which is an atom's width set against sixteen thousand light years, a sixth of the way across the Milky Way. The plates that make that negative energy are on the Zero Point page.
The tractor beam
A beam goes out, takes hold of something at a distance, and pulls it in. No cable, no contact. On screen it is used to catch a drifting shuttle and to hold a rock still while somebody looks at it. Stripped of the drama, the requirement is a projected force that grips a specific object and can be steered.
Two working tractor beams, one made of light and one made of sound
Arthur Ashkin at Bell Laboratories pushed tiny clear glass beads about with laser light in 1970, and in 1986 built the single-beam gradient trap that holds a particle in three dimensions at the focus of one beam. In 1987 he found live bacteria caught in the trap, unharmed, still dividing. Optical tweezers now hold single cells, stretch single strands of DNA and measure the step of a motor protein. The forces are one to a hundred piconewtons, which is the weight of about ten nanograms, a thousandth of a grain of table salt. Ashkin received half the 2018 Nobel Prize in Physics for it, at ninety-six.
Sound does the heavy lifting. Asier Marzo, Mihai Caleap and Bruce Drinkwater reported in Physical Review Letters in January 2018 that an array of 192 small ultrasound loudspeakers running at 40 kilohertz, in ordinary air, trapped and held a polystyrene sphere 16 millimetres across, nearly twice the sound's own wavelength of 8.6 millimetres. Two sound beams meeting head on cancel at still points spaced half a wavelength apart, about four millimetres here, and that gap had been the ordinary ceiling on what would fit between them; the bead is close to four times it. The trick was to alternate two oppositely winding acoustic vortices about a thousand times a second, so the bead feels only the average and stops spinning itself out of the trap. It can be held, moved and rotated on command, from one side, with nothing touching it.
- Sound array
- 192 transducers, 40 kHz, air
- Largest held
- 16 mm polystyrene sphere, about 54 milligrams
- Light trap force
- 1 to 100 piconewtons, the weight of about 10 nanograms
- What light holds
- single cells, bacteria, viruses, nanospheres
Hold a shuttle, tens of tonnes, at a range of hundreds of metres, and tow it. Call it 10,000 kilograms held against a gentle acceleration, with the beam crossing vacuum.
Light carries momentum, so a perfectly reflecting surface feels a push of twice the power divided by the speed of light. To hold one adult of 70 kilograms up against gravity by reflected light takes about 103 gigawatts, roughly three and a third times what Australia's whole electricity system delivers on an average day. Sound gives far more force per watt, but only through a fluid: acoustic levitation needs air or water to push against, so it stops at the edge of the atmosphere.
Either a coupling to ordinary neutral matter that does not come with the momentum bill, and nothing known behaves that way, or the beam is used where it already works: small things, short range, in a fluid, with the array scaled up in transducer count rather than in raw power.
Trap stiffness against object size. Push the transducer count up and record the largest mass held and the force at the edge of the trap, which is the number that decides what else could ever be held. In vacuum, the equivalent measurement is on levitated nanospheres: a Vienna group cooled one to its quantum ground state in an optical trap in 2020, so the control is already finer than the force is strong.
The gap: the built beam holds 54 milligrams. A person is 70 kilograms. That is a factor of 1.3 million, which is the step from a poppy seed to the person holding it. The distance is not the hard part; the mass is. Why only certain waves fit a given space, which is what makes the trap, is on the Resonance page.
The replicator
Ask a wall for a hot meal and it appears. The show set its own limit carefully: the replicator works at molecular resolution rather than the finer quantum kind, which is the in-story reason it can make a cup of tea but not a person. Read as a brief, that is a machine holding a library of patterns and a way of laying matter down to match one, fast enough to be worth waiting for.
A design travelled as data and the object appeared where it was needed
On 17 December 2014 the astronaut Barry Wilmore printed a ratchet spanner aboard the International Space Station. The design was drawn by Noah Paul-Gin at Made In Space, checked by safety reviewers, and the file was sent up from Marshall Space Flight Center. From idea on the ground to tool in orbit took under a week, and the print itself took about four hours. Nothing physical was launched. That is the replicator's actual shape: the pattern moves, the matter is already there.
The scale runs both ways. Two-photon printing lays polymer down in voxels about 100 nanometres across, small enough to build a working lens on the end of an optical fibre. At the other end, Relativity Space's Terran 1, thirty-four metres tall and 85 per cent 3D printed by mass, flew from Cape Canaveral on 22 March 2023 and held together through the worst aerodynamic loading of the climb before its second stage failed to reach orbit.
And the front half of the machine, the pattern library, is being filled right now. Google DeepMind's GNoME release put 2.2 million predicted crystal structures into the open, about 380,000 of them judged likely to hold together. Berkeley's A-Lab, a set of robot arms that weigh out powders, load furnaces, bake, cool and read their own results, made 41 of 58 attempted targets in 17 days. Knowing what to make, and having a machine make it without being told how, is running.
- Finest print voxel
- 100 nm in every direction
- Printed in orbit
- ratchet spanner, 17 December 2014, about 4 hours
- Largest printed flier
- Terran 1, 85 per cent by mass, March 2023
- Pattern library
- 2.2 million predicted crystals, 380,000 judged stable
The prediction side of the replicator has its own page: GNoME →
Four hundred grams of hot, structured, edible matter delivered in about thirty seconds from a stored pattern, repeatable, with no feedstock visible to the user.
Placement at the scale of molecules. Placing atoms one at a time is possible and has been since 1989, when Don Eigler and Erhard Schweizer at IBM Almaden nudged 35 xenon atoms into three letters with a scanning tunnelling microscope, roughly 22 hours of work. At that rate, one gram of carbon, about a pencil lead's worth and 5 × 1022 atoms, would take 3.6 × 1018 years, some 260 million times the age of the universe.
Assembly would have to be massively parallel, the way chemistry already is. A saucepan beats a needle because every one of those 1022 molecules reacts at once without being addressed individually. A replicator is therefore a chemistry problem wearing a printer's clothes: the question is not how to place an atom, it is how to make 1022 of them place themselves correctly.
Two numbers set everything else: placements per second per print head, and errors per placement. Published together they say exactly how big an object a given technique can finish before the defect count ruins it. On the pattern side, the number to watch is how many of GNoME's 380,000 survive contact with a furnace, which A-Lab is finding out one batch at a time.
The gap: the finest printer's smallest brick, a cube 100 nanometres on a side, still contains about 176 million carbon atoms at diamond density: six and a half times Australia's population, in one voxel. Molecular resolution means placing every one of those. That is a factor of 1.8 × 108 between the smallest thing a printer can put down and the smallest thing the fiction says it puts down.
The force field
A hangar with its doors wide open to vacuum, and the air stays in. A doorway you can see through and cannot walk through. The screen version is one field doing two jobs at once: holding gas on one side, and stopping solid objects. Those two jobs are worth separating, because one of them is done.
A wall made of nothing but hot gas really does hold a vacuum against the atmosphere
Ady Hershcovitch at Brookhaven National Laboratory built the plasma window: a wall-stabilised electric arc struck across a small opening, running in argon or helium, at close to 11,700 degrees Celsius at the edges and 14,700 in the centre, roughly twice as hot as the surface of the Sun. Counted up from absolute zero instead of from the freezing point of water, which is what the kelvin scale does, those are 12,000 and 15,000 kelvin. At that temperature the gas is about forty times thinner than room-temperature gas at the same pressure, and much more viscous, and those two facts together make it a plug. A 3 millimetre opening holds a pressure difference of more than 2.5 atmospheres, about what a car tyre carries above the air around it. A 2.36 millimetre arc, four centimetres long, separated a vacuum chamber from open air, and an electron beam at 175 thousand electronvolts passed straight through and out into the room. It ran steadily for 2,000 hours without maintenance.
There is no solid material in the beam path at any point. That is the hangar door, at three millimetres.
The other half of the fiction is also real, in its own way. At the Joint European Torus in Oxfordshire a plasma far hotter than the Sun's core sat clear of the vessel wall on magnetic field alone, and in October 2023 it produced 69 megajoules of fusion energy over 5.2 seconds from 0.2 milligrams of fuel. EAST in Hefei held a plasma near 70 million degrees for 1,066 seconds in January 2025, close to eighteen minutes; at that heat the 273 degree offset between Celsius and kelvin is far too small to show, so one figure does for both scales. Wendelstein 7-X in Greifswald set a record for the fusion triple product over 43 seconds in May 2025. Holding something violently hot away from a wall, with nothing but a field, is a solved and improving art.
- Plasma window aperture
- 2 to 11 mm arcs tested, 3 mm holding >2.5 atmospheres
- Arc temperature
- about 11,700 °C at the edge, 14,700 °C in the centre (12,000 K and 15,000 K)
- Power scaling
- about 10 kW per centimetre of arc diameter, 7.5 with a venturi
- Magnetic version
- JET, 69 MJ in 5.2 s, October 2023
A doorway two metres by nine hundred millimetres, holding one atmosphere of pressure difference indefinitely, clear enough to see straight through, and stopping a thrown object as firmly as it stops the air.
For gas, nothing new: the arc already does it, and the published scaling is about ten kilowatts for every centimetre of arc diameter. A 900 millimetre doorway on that scaling comes to 900 kilowatts, roughly 375 kettles boiling without a break, and it would be a wall at close to 11,700 degrees Celsius (12,000 kelvin). That is an extrapolation eighty times beyond the largest arc tested, and three hundred times the 3 millimetre one that holds 2.5 atmospheres, an estimate for comparison rather than a design anyone has drawn.
The second job is the one nothing on the bench does yet, and naming what is missing beats calling it hard. Fields push on charge directly, and a thrown rock carries no net charge, so a field that is the same strength everywhere pushes on it nowhere. A field that gets stronger across the object is a different matter: it pulls the charge inside the material slightly apart, giving it a weak plus end and a weak minus end, then tugs harder on whichever end sits in the stronger part of the field. That is an induced dipole in a gradient, and it lifts things. A 16 tesla magnet at Nijmegen held a live frog up on exactly that force, at a field times rate of change of about 1,360 tesla squared per metre. How big the force gets follows how easily the material's charge pulls apart, multiplied by that rate of change, which is why holding a frog takes a whole magnet bore, and why bismuth, the element that pushes back hardest against a magnetic field, about twenty times harder than water, is what a bench reaches for when a magnet has to hang in mid air with nothing touching it.
Of the four interactions anyone has measured, gravity is the one that acts on neutral mass without needing a gradient, and at the mass of a rock it is far too weak to notice. Everything else reaches neutral matter only through what it polarises, and that reach falls away fast with distance. So a doorway-sized field is a long way from turning a rock, and the quantity that would say how far is newtons per tesla squared per metre for the material being stopped: nobody has published one for a doorway. A missing coupling, not a missing megawatt.
Build the widest plasma window anyone has attempted and record whether the ten kilowatts per centimetre holds past a centimetre, and where cooling rather than power becomes the wall. That single curve, power against aperture, decides whether the hangar door is a scaling job or a different machine.
The gap: the air-holding job is short by a factor of 300 in aperture, which is a scaling problem with a published law attached to it. The object-stopping job is short by an unknown mechanism, which is a different kind of shortfall and worth naming as one. Real materials doing real structural jobs are on the Engineering page.
The tricorder
A box the size of a paperback. Point it at a rock and it names the minerals; point it at a person and it names what is wrong with them. It is the least glamorous device on the list and the one that has come closest, because both halves of it turned out to be spectroscopy, and spectroscopy got small.
Point a handheld reader at a piece of metal and it names the elements in under two seconds
An X-ray fluorescence reader fires a small X-ray source at whatever it touches, and every element in the sample answers with light at its own particular energies. Current handhelds read from magnesium to uranium, from single parts per million up to pure, in under two seconds for a standard alloy and under ten for a rock. Scrap yards use them to sort stainless grades by eye-speed. Archaeologists use them to read pigments without taking a sample. They weigh about as much as a full water bottle.
Raman readers do the molecular half. Shine a 785 nanometre laser at a substance, and a tiny fraction of the light comes back shifted by the vibrations of its bonds, which is a fingerprint. A handheld matches that fingerprint against an onboard library in seconds. Spatially offset Raman, published by Pavel Matousek and colleagues at the Rutherford Appleton Laboratory in 2005, shines the laser at one spot and collects from a spot beside it, which reaches deeper into the sample: enough to read the contents of a sealed container without opening it, and enough to read bone chemistry through skin.
The version working furthest from home is on Mars. SuperCam on the Perseverance rover points a laser at a rock two to seven metres away, reads the flash it makes, and reports the composition, with Raman and infrared spectrometers and a microphone alongside. Its five hundred thousandth shot was fired on 18 February 2025.
The medical half has a name and a date too. The Qualcomm Tricorder XPRIZE was judged in April 2017, and the top prize went to Final Frontier Medical Devices, a family-led team out of Pennsylvania, for DxtER: a set of non-invasive sensors that diagnosed 13 conditions and tracked five vital signs, in a person's own home, without a clinician.
- Elements read
- magnesium to uranium, parts per million to 100 per cent
- Time to an answer
- under 2 seconds for an alloy, under 10 for a rock
- Working range on Mars
- 2 to 7 metres, 500,000 laser shots by February 2025
- Medical version
- 13 conditions, 5 vital signs, judged April 2017
Handheld. Names any material by pointing. Names any condition in a person by pointing. Works at a distance, through containers, and on screen it is also used to survey a whole planet from orbit.
Signal has to reach the sensor, and that is where the two halves part company. Light of the right colour gets in and out of a rock easily. Tissue scatters visible light within a few millimetres, so most of what a diagnosis needs, the chemistry deep inside a body, never sends anything to the surface. Range is the other limit: light returned falls with the square of the distance, so reading from a 400 kilometre orbit rather than seven metres means 3.3 thousand million times less light coming back, nine and a half steps of ten.
For the medical half, either a wavelength that penetrates tissue and still carries chemical detail, which is what deep-tissue Raman and photoacoustic work are chasing, or an accepted set of surface signals that stands in for the inside. For the range half, a much larger collecting aperture or a much brighter probe, which is simply a bigger telescope with a laser on it.
How many conditions a non-invasive sensor set can separate, and at what error rate, published as a full confusion table rather than a headline count. And for the pointing instruments, the greatest stand-off distance at which a usable spectrum still comes back, measured rather than specified.
The gap: the materials half is close to done, and the readers are on tool belts. The medical half handles 13 conditions against roughly 70,000 diagnosis codes in the current international code book, a factor of about 5,400. That is a hypothesis about what sensors can reach, not a ceiling anyone has proved.
The transporter
A person stands on a pad, dissolves into light, and reassembles somewhere else about five seconds later. The device was reportedly invented for the show because landing a ship every week was too expensive to film, which is a fine origin for the piece of screen technology that ended up sharing a name with a real and working protocol.
A quantum state has been moved from a Tibetan mountain to a satellite fourteen hundred kilometres up
Charles Bennett and five colleagues wrote the protocol down in Physical Review Letters in 1993. Anton Zeilinger's group in Innsbruck and Francesco De Martini's in Rome ran it on single photons in 1997. Then it went outdoors, and then it went up.
Ji-Gang Ren, Jian-Wei Pan and colleagues teleported single-photon states from a ground observatory at Ngari in Tibet, 5,100 metres up, to the Micius satellite passing overhead, across distances from 500 to 1,400 kilometres. The ground source produced about 8,000 attempts each second; enough survived the climb through the atmosphere to give an average fidelity of 0.80 plus or minus 0.01 across six input states, well above what any classical copying could reach. Published in Nature, August 2017. In December 2024 a group at Northwestern reported teleportation through a fibre that was carrying ordinary internet traffic at the same time.
The name misleads, so here is what actually moves. The state travels; the particle does not. The original state is destroyed in the act of sending, so nothing is copied. An ordinary classical message has to travel alongside it at light speed for the receiver to finish the job, so nothing outruns light. And the receiving end must already hold a particle of the right kind, waiting. Quantum teleportation moves what a thing is doing onto a thing that is already there. That is a real and useful trick, and it is not the same trick as moving the thing.
- Greatest distance
- 1,400 km, ground to satellite, August 2017
- Fidelity
- 0.80 ± 0.01, averaged over six input states
- Source rate
- about 8,000 attempts per second
- Largest composite state
- a two-particle system, reported 2006
One adult, about 7 × 1027 atoms, moved 40,000 kilometres in roughly five seconds, arriving alive and continuous.
One entangled pair for every quantum state moved, distributed in advance, plus a classical channel wide enough to carry the measurement results. Write a person down as data and the scale appears: recording each atom's position to a tenth of a nanometre plus which element it is takes about 110 bits an atom, so roughly 7.7 × 1029 bits, near enough to 96 million zettabytes. All the traffic crossing the internet in a year is around five zettabytes, so one person is about 19 million years of everything the world sends.
Entanglement would have to be produced and stored at something like 1027 pairs per event, and the destination would need the matter already assembled and waiting in the right arrangement, since teleportation writes a state onto matter rather than delivering matter. Whether continuity of a person survives that is a question about persons rather than about physics, and the fiction has been arguing with itself about it for sixty years.
Entangled pairs per second across a working link, and how that rate falls with distance. Quantum memories and repeaters are the bottleneck and the whole field knows it: every improvement in memory lifetime shows up directly in that one number. Watch it, and you are watching the quantity the protocol's whole cost scales with: one entangled pair per state moved, so pairs per second is what any version of this has to buy.
The gap: the largest composite system teleported is two particles. A person is 7 × 1027 atoms: one for every grain of sand on a billion Earths' worth of beaches. The factor is 3.5 × 1027. The protocol is not the obstacle; the protocol works. The count is. Where the paths could lead from here is on the Frontiers page.
The light sabre
A hilt about thirty centimetres long throws a blade a metre long, and the blade stops there. A crystal inside sets the colour. It is one of the most copied objects in film, and underneath the sound design it is a serious four-part brief: make a plasma, hold it in a shape, stop it at a fixed distance in open air, and carry the power for all three in one hand.
A handheld plasma blade that cuts steel is a tool you can hire from a hardware shop
A plasma cutter strikes an electric arc through a jet of compressed gas and turns it into a column of ionised gas above 20,000 degrees Celsius. A common workshop machine, drawing about 10 kilowatts to deliver 45 amps, cuts through 16 millimetres of mild steel, hand-held, with a real plasma column steered by a person. Every part of the light sabre's blade behaviour except its length and its lack of a cable is on that bench already.
The shaping is real too. Push a big enough current down a plasma column and its own magnetic field squeezes it inward: that is a Z-pinch, and it needs no external magnet at all. Dense plasma focus machines make a pinched column a few millimetres long, and it lasts about a hundred nanoseconds before the sausage and kink instabilities pull it apart. A self-holding rod of plasma exists. It is short and it is brief.
Even the light has been persuaded to behave a little like matter. Ofer Firstenberg, Mikhail Lukin and colleagues reported in Nature in September 2013 that photons sent through a cloud of rubidium chilled to a whisker above absolute zero attract one another and travel out as a bound pair, behaving on the way through as though they had mass. Lukin told the Harvard Gazette at the time that the comparison with light sabres was not far off. The photons come out of the cloud as ordinary light again, which is the part the analogy has to leave behind.
- Workshop plasma arc
- over 20,000 °C, cuts 16 mm steel
- Power to run it
- about 10 kW input for 45 amps out
- Self-confined column
- a few millimetres, about 100 nanoseconds
- Bound photon pairs
- reported September 2013, in ultracold rubidium
A blade one metre long and about three centimetres across, hot enough to cut steel, on for minutes at a time, ending at a fixed length in open air, powered entirely from a hilt about 30 centimetres long and 4 centimetres across.
That blade has about 0.096 square metres of surface. If it shed heat like a hot solid at close to 19,730 degrees Celsius (20,000 kelvin, and a radiation sum counts from absolute zero, so kelvin is the scale it has to use) it would need about 870 megawatts, roughly two large power station units' whole output. Real arc plasma is partly see-through to its own light, so it sheds less; scaling a six kilowatt cutter arc up by volume to blade size gives about 60 megawatts, some 25,000 kettles. Somewhere between those two, then: engineering estimates for comparison, not a design anyone has drawn.
Three things. A return path for the current, because a plasma column is part of a circuit and open air is not. A confinement scheme stable for seconds rather than nanoseconds, which is the same problem fusion has been working on since the 1950s. And an energy store far denser than anything on the market: a 30 by 4 centimetre hilt holds 377 millilitres, which filled with the best lithium cells sold today comes to about 940 kilojoules, or 262 watt-hours, roughly three of the largest power tool batteries.
How long a self-pinched plasma column can be held before instability takes it, plotted against current and gas mixture. That number has been climbing slowly for seventy years and it decides the whole brief. It is also, not coincidentally, the central measurement in every magnetic fusion machine on Earth, which means the light sabre and the power station are waiting on the same result.
The gap, twice: running a 60 megawatt blade for three seconds takes 180 megajoules, which is 190 times what the hilt can hold, close to the step from a single AA cell to a car battery. And holding a self-confined column for three seconds is 30 million times longer than the built one lasts, which is close to the step from one second to one year. The second number is the harder one, and it is the one being worked on daily.
The cloaking device
In the episode "Balance of Terror", first broadcast in December 1966, a ship simply is not there to be seen. The requirement underneath is exact and rather beautiful: send the light around the object and put it back on the far side travelling as though nothing had been in the way, with the right phase, so there is no shadow behind and no glint in front.
Cloaking works, in narrow bands, and the theory behind it is now ordinary optical engineering
John Pendry, David Schurig and David Smith published transformation optics in Science in May 2006. The idea is a coordinate change: write down the path you want light to take, and the mathematics hands you the material properties that would make light take it. Six months later, in November 2006, Schurig and colleagues built one. Ten concentric rings of copper split-ring resonators printed on circuit board, wrapped around a five centimetre copper cylinder, measured at 8.5 gigahertz. Both the shadow behind the cylinder and the reflection in front of it dropped.
Then it moved to light. Jason Valentine and colleagues in Xiang Zhang's group at Berkeley reported a carpet cloak, which flattens a bump so a mirror over it reads as bare floor, made entirely of non-conducting material in Nature Materials in 2009: a patterned silicon slab 250 nanometres thick that hid a bump 3.8 micrometres wide and 400 nanometres tall, and did it across the whole band from 1,400 to 1,800 nanometres. That is a quarter of the centre wavelength, a genuinely broad band, and it uses no metal and so loses almost nothing to heat.
And the trick is not limited to light. Stéphane Brûlé and colleagues drilled a grid of boreholes into soil near Grenoble in 2012, shook the ground at 50 hertz, and measured the surface wave being turned away from the region behind the grid. Reported in Physical Review Letters in 2014. That is a full-size cloak, built in dirt, deflecting a wave you could feel, and it has a civil engineering job waiting for it.
- First built cloak
- copper split rings, 8.5 GHz, November 2006
- Optical carpet cloak
- hid a 3.8 µm bump across 1,400 to 1,800 nm
- Ground version
- boreholes near Grenoble, 50 Hz surface waves, 2012
- Bandwidth bound
- passive cloaks trade bandwidth for depth, Optica, 2016
Geometry doing the work of chemistry has its own page: Metamaterials →
An object the size of a person or a ship, unseen across the whole visible band, from every angle, in daylight, while moving, with the crew able to see out.
Bandwidth and feature size. A resonant split ring works over roughly a couple of per cent of its centre frequency, so about 200 megahertz around 8.5 gigahertz: an estimate from how resonators behave, not a figure the 2006 paper reports. Visible light spans 400 to 790 terahertz, a band 390,000 gigahertz wide and about 66 per cent of its own centre. And the structures have to shrink with the wavelength: rings for 8.5 gigahertz are millimetres, while green light needs features about 64,000 times smaller, tens of nanometres, at sizes where metals turn light into heat.
Francesco Monticone and Andrea Alù proved a bound in Optica in 2016 using an old result from radio matching theory: any passive cloak that reduces scattering in one band increases it elsewhere, added up across the spectrum. So broadband invisibility cannot simply be designed into a passive shell; bandwidth has to be bought. Cloaks that put energy in are not bound the same way, and that is where the room is.
How much of the wave the object still throws back, measured right across a doubling of frequency, published as one curve rather than one point on the dial. Almost every disagreement about cloaking would be settled by that plot, because it shows at a glance where the bandwidth was borrowed from.
The gap: a narrow microwave cloak covers about 200 megahertz, which is roughly a thousand FM radio channels side by side. Visible light needs 390,000 gigahertz, about two million times as much. The carpet cloak closed a good part of that at infrared wavelengths on a bump four micrometres wide, which is why the argument now is about size and bandwidth together, not about whether it works.
Where the eight sit against each other
Lined up, the shortfalls sort themselves into three kinds, and the kind matters more than the size.
| Machine | The quantity that decides it | Nearest built figure | Times short |
|---|---|---|---|
| Light sabre | seconds a self-confined plasma column holds | 100 nanoseconds | 3 × 107 |
| Force field | aperture the arc can bridge | 3 mm at 3 kW | 300 |
| Tricorder | conditions a sensor set can separate | 13 | 5,400 |
| Tractor beam | mass held at a distance | 54 mg | 1.3 × 106 |
| Cloaking device | bandwidth covered | about 200 MHz | 2 × 106 |
| Replicator | atoms in the smallest placeable brick | 176 million | 1.8 × 108 |
| Transporter | particles in the largest state moved | 2 | 3.5 × 1027 |
| Warp drive | negative energy available | 43 picojoules | 1.5 × 1030 |
Scaling problems
The force field's air-holding half, the tricorder and the tractor beam are short on quantity, not on mechanism. Every one of them has a published scaling law or a clear engineering path, and the shortfalls run from a few hundred to a few million: large, and of a kind that industry closes routinely when there is a reason to. The plasma window's power law is written down. The transducer count in an acoustic trap can go up.
Counting problems
The replicator and the transporter work already, once. The obstacle is doing it 1027 times at once, which is a different animal from doing it once better. Chemistry solves exactly this problem every time a pot of soup thickens, by never addressing an individual molecule. Whether an assembler or an entanglement source can be made to work that way is the open question, and it is a good one.
Keep exploring
Zero Point
The Casimir force, the photons pulled out of empty space in 2011, and the only place negative energy density has ever been weighed.
Metamaterials
Geometry doing the work of chemistry: cloaks, flat lenses and materials that get fatter when you stretch them.
GNoME
2.2 million predicted crystals, 380,000 judged stable, and a robot kitchen cooking the first batches. The front half of a replicator.
Frontiers
Where discovery could head next, told in could and would, with the instruments that would settle it.
Sources
- M. Alcubierre, "The warp drive: hyper-fast travel within general relativity", Classical and Quantum Gravity 11, L73, May 1994.
- M. J. Pfenning and L. H. Ford, "The unphysical nature of 'warp drive'", Classical and Quantum Gravity 14, 1743, 1997.
- C. Van Den Broeck, "A 'warp drive' with more reasonable total energy requirements", Classical and Quantum Gravity 16, 3973, 1999 · arxiv.org/abs/gr-qc/9905084
- J. Fuchs, C. Helmerich, A. Bobrick, L. Sellers, B. Melcher and G. Martire, "Constant velocity physical warp drive solution", Classical and Quantum Gravity 41, 095013, 2024 · arxiv.org/abs/2405.02709
- K. Clough, T. Dietrich and S. Khan, "What no one has seen before: gravitational waveforms from warp drive collapse", The Open Journal of Astrophysics, July 2024 · arxiv.org/abs/2406.02466
- LIGO Scientific Collaboration and Virgo Collaboration, "Observation of gravitational waves from a binary black hole merger", Physical Review Letters 116, 061102, February 2016.
- A. Ashkin, "Acceleration and trapping of particles by radiation pressure", Physical Review Letters 24, 156, 1970; single-beam gradient trap, Optics Letters 11, 288, 1986; Nobel Prize in Physics 2018.
- A. Marzo, M. Caleap and B. W. Drinkwater, "Acoustic virtual vortices with tunable orbital angular momentum for trapping of Mie particles", Physical Review Letters 120, 044301, January 2018 · link.aps.org/doi/10.1103/PhysRevLett.120.044301
- A. I. Hershcovitch, E. D. Johnson and R. C. Lanza, "The plasma window: a windowless high pressure-vacuum interface for various accelerator applications", Proceedings of the 1999 Particle Accelerator Conference, New York, 584 · proceedings.jacow.org/p99/PAPERS/FRAL3.PDF
- M. V. Berry and A. K. Geim, "Of flying frogs and levitrons", European Journal of Physics 18, 307 to 313, 1997; frog levitated at 16 tesla at the High Field Magnet Laboratory, Nijmegen · physics.umd.edu/grt/taj/411c/FlyingFrogs.pdf
- EUROfusion, JET DTE3 result: 69 megajoules over 5.2 seconds from 0.2 milligrams of fuel, October 2023, announced February 2024. EAST 1,066 second high-confinement plasma, January 2025. Wendelstein 7-X triple product record over 43 seconds, May 2025.
- D. M. Eigler and E. K. Schweizer, "Positioning single atoms with a scanning tunnelling microscope", Nature 344, 524, 1990.
- NASA, "Space station 3-D printer builds ratchet wrench to complete first phase of operations", December 2014 · nasa.gov
- R. C. Wiens et al., "The SuperCam instrument suite on the NASA Mars 2020 rover", Space Science Reviews 217, 4, 2021; 500,000th laser shot reported by CNES, February 2025.
- P. Matousek et al., "Subsurface probing in diffusely scattering media using spatially offset Raman spectroscopy", Applied Spectroscopy 59, 393, 2005.
- Qualcomm Tricorder XPRIZE final results, April 2017; DxtER by Final Frontier Medical Devices · xprize.org/competitions/tricorder
- C. H. Bennett, G. Brassard, C. Crépeau, R. Jozsa, A. Peres and W. K. Wootters, "Teleporting an unknown quantum state via dual classical and Einstein-Podolsky-Rosen channels", Physical Review Letters 70, 1895, 1993.
- J.-G. Ren et al., "Ground-to-satellite quantum teleportation", Nature 549, 70, August 2017 · arxiv.org/abs/1707.00934
- Q. Zhang et al., "Experimental quantum teleportation of a two-qubit composite system", Nature Physics 2, 678, 2006.
- O. Firstenberg, T. Peyronel, Q.-Y. Liang, A. V. Gorshkov, M. D. Lukin and V. Vuletić, "Attractive photons in a quantum nonlinear medium", Nature 502, 71, September 2013.
- J. B. Pendry, D. Schurig and D. R. Smith, "Controlling electromagnetic fields", Science 312, 1780, 2006; D. Schurig et al., "Metamaterial electromagnetic cloak at microwave frequencies", Science 314, 977, November 2006.
- J. Valentine, J. Li, T. Zentgraf, G. Bartal and X. Zhang, "An optical cloak made of dielectrics", Nature Materials 8, 568, 2009 · arxiv.org/abs/0904.3602
- F. Monticone and A. Alù, "Invisibility exposed: physical bounds on passive cloaking", Optica 3, 718, 2016 · opg.optica.org/optica/fulltext.cfm?uri=optica-3-7-718
- S. Brûlé, E. H. Javelaud, S. Enoch and S. Guenneau, "Experiments on seismic metamaterials: molding surface waves", Physical Review Letters 112, 133901, 2014.
- A. C. Clarke, "Extra-terrestrial relays: can rocket stations give world-wide radio coverage?", Wireless World, October 1945.
- Relativity Space Terran 1 first flight, 22 March 2023, Cape Canaveral. Nanoscribe two-photon polymerisation feature size, manufacturer specification.