The lethal envelope
Focused rather than spread.
Take the five-kilometer array at the reference beam of 100 gigawatts at 1.06 micrometers and, instead of spreading it over a receiver, focus it as tightly as diffraction allows. It holds structural steel at its melting point out to about 6.3 AU and aluminum out to about 22. Inside 1.5 AU it vaporizes anything. Beyond Saturn it cooks rather than cuts.
Range scales linearly with aperture, so a 20-kilometer array reaches Neptune’s orbit as a cutting weapon. Dwell times inside the envelope are short: a centimeter of steel needs about 80 megajoules per square meter, which is 3.7 seconds at 1 AU and 102 seconds at Jupiter.
Those ranges assume the target reradiates from the illuminated face only, which is the optimistic case for the shooter. Two-sided reradiation cuts the steel range to 4.5 AU and the aluminum range to 16. Real targets are neither, and this site quotes the one-sided convention with a ±40 percent spread carried as an open number. Ablation and plume shielding are unmodeled; so is dust-cloud attenuation as a cheap countermeasure.
The dodge criterion
Three quantities, and one division.
The intuition is that a light-speed weapon is unstoppable. The arithmetic says the opposite.
Three quantities. The spot is diffraction-limited at s = 2.44 λL/D. The correction interval has a floor of one round-trip light time, τ = 2L/c, because the observation and the delivery each cross the range at the speed of light and no amount of computing power moves that. And a target under continuous unpredictable thrust a displaces d = 2aL²/c² over that interval. Divide:
d / s = a L D / 1.22 λc²
The dodge margin scales linearly with both range and aperture. Range appears because displacement grows as the square of distance while the spot grows only as the first power, so waiting is worth more to the target than distance costs it. Aperture appears because better collimation shrinks the spot and does nothing to the light time. Every design choice that makes the weapon more capable makes it easier to walk away from.
Break-even is therefore a product rather than a distance, and a designer can trade range against aperture without escaping it.
| Target thrust | D = 100 m | D = 1 km | D = 5 km (Mercury) | D = 10 km | D = 47.6 km |
|---|---|---|---|---|---|
| 10 mm/s² (~1 milli-g) | 0.78 AU | 0.078 AU | 0.016 AU | 0.008 AU | 0.0016 AU |
| 1 mm/s² | 7.77 AU | 0.78 AU | 0.155 AU | 0.078 AU | 0.016 AU |
| 0.1 mm/s² | 77.7 AU | 7.77 AU | 1.55 AU | 0.78 AU | 0.163 AU |
| 0.01 mm/s² | 777 AU | 77.7 AU | 15.5 AU | 7.77 AU | 1.63 AU |
At one millimeter per second squared of target thrust, the five-kilometer array breaks even at 0.155 AU, against a lethal envelope of 6.3. Every range at which the beam has lethal flux is 40 or more times beyond the range at which it can still hit something that does not want to be hit.
And a millimeter per second squared is not demanding. It is a tenth of a milli-g. The beam-fed tug on the departure corridor page runs at 20 times that, and a solar-electric freighter cruises at 0.1 to 0.5.
Both counters fail
For the same reason, at different granularity.
Counter one: spread the beam to cover the volume the target might occupy. This guarantees a hit geometrically and fails energetically, because flux falls as the square of the diameter ratio — the price of a guaranteed hit is exactly (d/s)². At the steel range that turns 474 kilowatts per square meter into 255 watts, a fifth of terrestrial sunlight.
Counter two: tile the uncertainty circle with many independent beams instead of one broad one. Same factor: 1,859 spots are needed at 6.3 AU, so 100 beams cover about 5 percent of the circle per correction cycle at one percent of the flux, against a target that re-randomizes every cycle. The counter fails twice, on coverage and on flux.
Two honest qualifications keep this from being a general result. Inside about 1 AU the guaranteed-hit flux is still half a megawatt per square meter and maneuvering saves nothing. And against a sluggish target at a tenth of a millimeter per second squared, the guaranteed-hit flux at long range is still 19 solar constants, which wrecks thermal control and optics even if it melts nothing. This is a long-range failure of the weapon, not a general one.
What it can hit
Fixed, unpowered, or predictable.
Three categories. The first two are physical, and everything valuable in the architecture sits in them: stations, depots, habitats, surface installations, mass drivers, radiator farms, collector arrays, receivers, and anything under construction are indefensible. Warships, tugs under power and small craft are effectively immune.
A Mercury array is a counter-infrastructure weapon and not a counter-force weapon — the exact inverse of the boost-phase intercept problem, and the inverse of almost every intuition imported from terrestrial directed-energy debate.
The third category is not physical and it is the uncomfortable one. A vehicle is predictable because it has declared its trajectory. A cargo tug on a filed continuous burn is the most predictable object in the solar system, which is what makes it a docile beam customer and also what makes it a target requiring no dodging model at all. The declaration regime that makes commerce verifiable is the same regime that makes commerce targetable, and a fleet that stops declaring becomes untargetable and unverifiable in the same instant. There is no version of a verification architecture that gets one without the other, and the safeguard is still worth more than the exposure.
What dispersal costs
Cheapest where siting is free.
The defense against a counter-infrastructure weapon is dispersal, which is cheapest where siting is free and most expensive where the site is the asset.
At Shackleton the illuminated ridge, the shadowed floor and the corridor between them are a few kilometers apart because that is physically where they are. The polar architecture cannot disperse without abandoning what made it valuable, and the same is true of Mercury’s own power ring. The weapon is specified at Mercury and the terrain is what it is on the Moon and in the belt, and neither side of that trade has ever been priced against the other.
Why this is stable
The array cannot defend itself.
There is no engineering difference between the peaceful and the hostile configuration — same aperture, same phase control, same power, and the only difference is a pointing command, which is software. Any verification regime that tried to separate them would be verifying intent, which no arms control regime has ever managed.
Three facts make that less catastrophic than it sounds.
The array cannot defend itself. Between 56 and 137 square kilometers of thin film in a known heliocentric orbit is the most fragile large object anyone will ever build, and a gravel canister on an intercept trajectory ends it. Mutual vulnerability is total and obvious, which is the classic precondition for deterrence rather than use.
Attribution is trivial. Exactly one facility can do this. There is no proxy and no deniability.
The engagement-authority problem is bounded. Against fixed infrastructure the beam has hours, so a human decision loop at 6 to 12 minutes of light time fits comfortably inside it. Only against maneuvering targets would authority have to be delegated to a machine — which is precisely the case the arithmetic says the beam cannot win. The one weapon in the architecture that would most obviously require an autonomous system to fire is the one that gains nothing from firing fast.
For completeness, because a reader who has done this arithmetic will do it elsewhere: the interstellar sail array is 4.8 to 47.6 kilometers across at the same reference beam, and at the large end it holds steel at its melting point to 60 AU and aluminum to 212 — past Pluto. That array is an order of magnitude more capable than this one and has until now been called a launcher. It also owns the corollary that the sail itself, gram-class with no thrust authority, a deliberately enormous cross-section and a published trajectory, is the ideal target for exactly this weapon.