The voyage

Pushed by light

A lunar laser array accelerates the seed without onboard rocket fuel.

Why can't a starship just carry its fuel?

A rocket moves by throwing something backward. The faster you want to go, the more you have to throw, which increases the launch mass. Fuel needed later in the flight must itself be accelerated, requiring additional fuel.

A chemical rocket throws its exhaust out at about 3.7 kilometres a second. The seed has to reach a tenth of the speed of light, which is 29,979 kilometres a second. That is 8,102 times the exhaust speed, and every extra exhaust speed multiplies the rocket's mass by 2.7, so the multiplication happens 8,102 times in a row. Counted in kilograms, the fuel for one kilogram of ship is, on this first count, a number with 3,519 digits. The whole observable universe holds roughly 10⁸⁰ atoms.

The required mass rules out chemical propulsion for this mission. A fusion rocket with a much higher exhaust speed reduces the mass ratio substantially. Even so, a ship that must also stop at the far end has to leave with 804 kilograms for every kilogram that arrives, with an exhaust about as fast as the one Project Daedalus, the British Interplanetary Society's 1978 fusion starship, was designed for. Even a one-tonne payload would require a large industrial effort.

An external source of propulsion avoids this fuel requirement. Light has momentum. When it bounces off a mirror, the mirror is shoved, the way a ball shoves a wall. The push of sunlight is tiny, 9.1 × 10⁻⁶ newtons on each square metre of mirror at Earth's distance from the Sun, but it is real and it never runs out: in 2010 a Japanese probe called IKAROS became the first spacecraft to sail on sunlight between planets.

Guess first A laser beam hits two flat sheets of the same size: a mirror and a sheet that soaks up all the light. Which is pushed harder?

The mirror, twice as hard. The black sheet only stops the light. The mirror stops it and then sends it back out the other way, so it is pushed once for each.

The launch design uses a laser to provide much higher intensity than sunlight. A laser array at the Moon's south pole focuses a beam on a sail 9 kilometres across, a film so thin that a whole square metre of it has a mass of only 5 × 10⁻⁵ kilograms. The beam follows the sail outward. For 76 days the seed is pushed at 0.47 times the pull of Earth's gravity, and when the push ends, far beyond Neptune, the sail is cut loose and the seed coasts on at a tenth of light speed. The array stays on the Moon. It can fire again for the next seed.

The launch consumes electricity on the Moon. At the destination there is no array to push back, and neither the gas between the stars nor the light of the star Ember can stop a ship this fast, so the seed carries a fusion stage and its propellant, and the laser launches those too (Stopping from 0.1c). That makes the launch 22.7 tonnes, of which the seed is 4 percent. Under the story's software-mind premise, later residents can travel as data over a light link (Bits travel, atoms do not).

How hard does light push, and what does it cost?

The launch-stack mass includes a stopping radiator at 1.0 kg/m². That target extrapolates Juhasz's hot-radiator projection and is not demonstrated at the stage's temperature challenge. Stopping from 0.1c gives the source and the sensitivity of the stage mass.

For comparison, consider the rocket fuel requirement. The rocket equation says the mass ratio, the ship's mass at the start divided by its mass at the end, is

mass ratio = e^(speed change ÷ exhaust speed)

For a speed change to a tenth of light speed, the same change the sail makes, the engines compare like this (the table uses the equation above, which ignores relativity; the exact form in the first fold raises each exponent by a few parts in a thousand):

Engine	Exhaust speed	Mass at launch per kilogram that arrives
Methane and oxygen	3.7 km/s	a number with 3,531 digits (launch only)
Fusion rocket, on paper	0.03 c	28 to launch, 804 to launch and stop
The ideal fusion rocket	0.088 c	3.1 to launch, 9.7 to launch and stop
Laser-pushed sail	none: the engine stays home	1.14 to launch (only the sail is left behind); 8.51 to launch and stop, with a 0.05 c fusion stage for the stop, against 55 for that engine used both ways

Now the sail. Light carries momentum equal to its energy divided by c. A beam of power P brings P ÷ c of momentum every second, and a mirror that sends it straight back gets twice that:

force on a mirror = 2 × power ÷ c

Dividing by c makes the push tiny: one watt of light pushes with 6.7 × 10⁻⁹ newtons, and one newton takes 1.5 × 10⁸ watts. Accelerating the seed requires a high-power beam and a low-mass sail. In the recommended design, the beam that lands on the sail starts at 1.6 × 10¹³ watts, which gives 104 kilonewtons. The launch stack, the seed with its fusion stage, propellant and sail, has a mass of 22,715 kilograms, so its acceleration is 4.56 metres per second squared, or 0.47 g. The light on the film is about 200 times as bright as sunlight at Earth's distance from the Sun.

Three arrays could do the job, and they trade size against speed of push. A bigger array keeps the beam narrow for longer, so the push is gentler and covers more distance:

Design	Array (km)	Sail (km)	Sail mass (kg)	Push (AU)	Days	g	Electric power (W)
Long push	1,300	10	3,927	32,300	3,731	0.0095	6.5 × 10¹¹
Shackleton array (recommended)	30	8.5	2,853	658	76	0.47	3.0 × 10¹³
Polar plateau array	100	5.6	1,216	1,433	166	0.21	1.3 × 10¹³

The long push needs far less power, but its array would be 1,300 kilometres across, 36 % of the Moon's width. The recommended one fits around a crater. design

What does the energy bill look like?

At low sail speeds, reflected light retains most of its energy, so the beam transfers only a small fraction to the sail. Averaged over the push to a tenth of light speed, the sail keeps 0.095 of the beam's energy, a little under a tenth. At the very end it keeps 0.18.

The launch stack ends with 1.0 × 10¹⁹ joules of motion, the mass-energy of 114 kilograms (see A gram is a fortune); the seed's own share is 4.5 × 10¹⁷ joules. To put that into the sail the array sends 1.1 × 10²⁰ joules onto it, and drawing that from the wall takes 2.1 × 10²⁰ joules of electricity at the lasers' efficiency of 0.60. That is 1.90 times the electricity the world produced in 2024, and 23 times what the seed alone would need, because the fuel for the stop rides the same beam. This supplies the launch energy. Coasting requires no further propulsion, and the same array can launch another seed. design

The other 8.4 × 10¹⁹ joules leave the lasers as heat, and the heat has to go somewhere. In the story it goes into ice under the pole.

With JavaScript on, this is a calculator: pick an engine and a speed and it shows the fuel a rocket would need against the electricity a sail would. The tables above give the same numbers for the three engines.

Open everything

Where do the rocket equation and the factor of two come from?Conservation of momentum, a little calculus, and one Doppler shift

The rocket equation

A rocket of mass $m$ throws out a small piece of mass $dm$ backward at speed $v_e$ relative to itself. The rocket loses that mass and gains speed $dv$ . Momentum has to balance:

$m\,dv = -v_e\,dm$

Rearranged, $dv = -v_e\,dm/m$ , which integrates directly:

$\Delta v = v_e \ln\frac{m_0}{m_1} \qquad\Longleftrightarrow\qquad \frac{m_0}{m_1} = e^{\Delta v / v_e}$

The mass ratio grows exponentially with the required speed change. Each added $v_e$ of speed multiplies the mass by $e$ . For methane and oxygen at 3,700 m/s and a speed change of 29,979 km/s, the exponent is $8{,}102$ , and $e^{8{,}102}$ is the mass ratio quoted above.

At a tenth of light speed the speed is no longer small next to $c$ , and the exact answer swaps $\Delta v$ for the sum of many small boosts, each added in rapidity $\phi = \operatorname{artanh}\beta$ :

$\frac{m_0}{m_1} = \exp\!\left(\frac{c\,\phi}{v_e}\right) = \left(\frac{1+\beta}{1-\beta}\right)^{c/2v_e}$

Here $v_e$ is the effective exhaust speed, thrust divided by mass flow. For the methane engine this gives a number with 3,531 digits instead of 3,519. The relativistic correction increases the required mass ratio.

Light's push

A photon of energy $E$ carries momentum $p = E/c$ . A beam of power $P$ arriving head-on at a mirror at rest delivers momentum at the rate $P/c$ . Reflected straight back, the light leaves with $-P/c$ . The mirror's momentum changes by the difference:

$F = \frac{P}{c} - \left(-\frac{P}{c}\right) = \frac{2P}{c}$

An absorbing sheet keeps only the incoming $P/c$ . That is the answer to the question in the story.

A sail that runs away from the beam

The seed moves at $\beta c$ away from the array, so the light undergoes Doppler shifts on arrival and reflection. Seen from the sail, the light arrives redshifted by the Doppler factor

$D = \sqrt{\frac{1+\beta}{1-\beta}}$

( $D = 1.1055$ at $\beta = 0.10$ ). A mirror reflects each photon unchanged in its own frame, so in the array's frame the photon comes back redshifted by $D$ a second time, with its energy cut by $D^2$ . The sail gains the difference. Per unit of beam energy $E$ that lands on it, the sail's kinetic energy rises by

$\eta = 1 - \frac{1}{D^2} = \frac{2\beta}{1+\beta}$

At $\beta = 0.1$ this is 0.182, and at low speed it falls toward $2\beta$ , which is why a slow sail wastes nearly everything. The momentum push falls by a similar factor, with $P$ the power the array sends:

$F = \frac{2P}{c}\,\frac{1-\beta}{1+\beta}$

Holding the acceleration constant therefore means raising the beam power over the push: by $(1+\beta)/(1-\beta)$ for the Doppler loss, and by another $\gamma^3$ because momentum grows faster than speed as $\beta$ rises. At the end of the push the beam must be 1.24 times as strong as at the start.

Integrated beam energy

The sail's kinetic energy is $E_k = (\gamma - 1)mc^2$ , so $dE_k = mc^2\gamma^3\beta\,d\beta$ . The beam energy that must land on the sail is $dE_b = dE_k/\eta$ . Integrate from rest to $\beta$ :

$E_b = \frac{mc^2}{2}\int_0^\beta \gamma^3(1+\beta')\,d\beta' = \frac{mc^2}{2}\bigl(\gamma(1+\beta) - 1\bigr) = \frac{mc^2}{2}\,(D - 1)$

For the launch stack this is $E_b = 1.08 \times 10^{20}\,\text{J}$ , the figure in the Numbers section. Divide by the share of the beam that lands, $0.85$ , and the lasers' efficiency, $0.60$ , to get the electricity bill.

Why the push has to end

A laser array of width $D_a$ and wavelength $\lambda$ cannot send a perfectly parallel beam. Diffraction spreads it, and the bright central spot a distance $L$ away has a diameter of about $2.44\,\lambda L / D_a$ . The sail, of width $D_s$ , catches the whole beam only while the spot is no bigger than the sail:

$L = \frac{D_a D_s}{2.44\,\lambda}$

The array focuses its beam on the sail. Until the end of the push the spot is no wider than the sail; past it, the spot keeps widening and most of the light misses.

With $\lambda = 1.064 \times 10^{-6}$ m, $D_a = 30{,}000$ m and $D_s = 8{,}524$ m, the push length is $L = 658$ AU. After that the beam is wider than the sail and the captured fraction falls with the square of the spot size. At a light-year the spot is 819 km across and the sail catches about $1 \times 10^{-4}$ of the light. So the push must finish inside $L$ .

The push distance fixes the rest. With constant acceleration $a$ from rest to $v$ :

$L = \frac{v^2}{2a}\qquad\Longrightarrow\qquad a = \frac{v^2}{2L},\qquad t = \frac{2L}{v}$

Larger optics lengthen $L$ and reduce the required acceleration. The beam power that must land on the sail follows from $F = 2P/c$ and $F = ma$ : $P = mac/2$ , which is the beam power quoted in the Numbers section. A longer push reduces the required power at the cost of a larger optical aperture.

Has anyone done it, and what could go wrong?The experiments, the proposals and the open problems

Light pressure was measured long ago, and flown recently

James Clerk Maxwell predicted in 1873 that light presses on whatever it falls on. The pressure is so small that it took until 1901 for Pyotr Lebedev to measure it in a laboratory, with light falling on thin vanes in a vacuum. The first spacecraft to sail on sunlight in interplanetary space was IKAROS, which Japan launched in 2010. Its square sail unfolded on the way to Venus and gave it a small, measured extra push. A sail pushed by sunlight is slow but proven. What no one had done by 2026 is push a sail through space with a beam built for the purpose.

Who thought of pushing with a laser?

In 1984 Robert Forward showed how a laser-pushed sail could make an interstellar round trip, using a huge focusing lens in space and a sail that splits into stages. In 2016 Philip Lubin laid out a roadmap for the idea: a vast array of lasers pushing gram-scale craft, each built on a single wafer, to more than 0.25c, so that the nearest star is reached in 20 years. The same year the Breakthrough Starshot initiative set out to prove the concept. Kevin Parkin's system model for Starshot found a cost-optimal point design for 0.2c to Alpha Centauri, with a sail 4.1 metres across accelerated for 9 minutes.

The Eventide launch extrapolates from these in one direction only: a much heavier craft, a sail 9 kilometres across with a mass of 2,853 kilograms, and a push of weeks instead of minutes. It keeps the push to weeks by using an array 30 kilometres across, far larger than anything built by 2026. design

What goes wrong?

Staying on the beam. A sail on the edge of a beam is pushed sideways as well as forward. Zachary Manchester and Abraham Loeb showed in 2017 that a cone-shaped sail on a Gaussian beam is unstable without active control, and proposed a spherical shell that rides a beam with no feedback at all. A real sail also has to survive being nudged, spun and heated unevenly.
Heat. The film absorbs a small share of the beam, and it can lose heat only by radiating it. If the film absorbs 1 × 10⁻⁵ of the 2.7 × 10⁵ watts per square metre that land on it, it takes in 2.7 watts per square metre, and a perfect radiator on both faces would sit near 70 K. A real film radiates less than a perfect radiator, so it runs hotter, and a film that reflects the beam almost perfectly while shedding heat well is the central open problem. challenge Harry Atwater and colleagues set out in 2018 the criteria a Starshot sail film would have to meet, and argued that meeting them needs major advances in photonic design and materials science.
Getting the relativity right. Kulkarni, Lubin and Zhang solved the sail's motion fully relativistically in 2018. Diffraction of the beam limits the top speed, and they traced critics' claims that the earlier work was wrong to an improper treatment of causality.
The array. Every laser in it must be locked in phase to a fraction of a wavelength across the whole array, track a moving sail with a pointing error far smaller than the beam, and run for weeks. challenge