Several Satellite Constellations Already Exceed the Size at Which Their Debris Turns Unstable, or Runs Away

Ask a national regulator whether a satellite is safe to launch and you will get an answer about that satellite. Will it come down within the required window? Can it be steered out of somebody else's way? Is it reliable enough to be trusted with the maneuver at all? Every question on the form is about the one object named on the form. Ask whether thirty thousand of them are safe to launch together, and there is no equivalent procedure to hand.

A paper posted to arXiv on July 31 sets out to supply one. In "Critical Sizes of Satellite Constellations," Hugh G. Lewis, its sole author, asks how large a constellation can get before collisions among its own members start producing fragments faster than orbits can clear them. The answer is a number, specific to each constellation, and he calls it the critical size. Several constellations that are planned, partially deployed or fully deployed are already above their own.

The physics underneath is not new. Objects in low Earth orbit are removed slowly, by atmospheric drag hauling them down over years or centuries, and added abruptly, by collisions that turn two intact spacecraft into thousands of pieces. When the adding outruns the removing, the population climbs by itself, and every new fragment raises the odds of the next collision. Donald Kessler set the feedback out in 1978, and it has carried his name since. The preprint's abstract states the result rather than the derivation, so the description of critical size in the paragraph above is a plain-language gloss on that physics, not the paper's own definition.

What happens above the line is the part a summary most easily gets wrong. The abstract's phrase is that these constellations exceed their critical size "producing an unstable or a runaway orbital debris population." Two outcomes, not two words for one, and the paper keeps them apart. The distinction is standard in debris modeling, and Lewis has published thresholds for both: in an assessment he presented with Kessler at the European Conference on Space Debris in 2025, the existing population of intact objects in low Earth orbit already sat above the unstable threshold at every altitude between 400 and 1,000 kilometers, and above the runaway threshold at nearly every altitude between 520 and 1,000. The new paper does not say in its abstract which constellations land in which of the two states.

Three qualifications in that abstract do more damage than the headline finding. This happens, Lewis writes, "both for small and large constellations, despite debris mitigation measures that go beyond internationally recognised good practices, and even when the constellations are deployed into a pristine, debris-free orbital environment." Read the last clause twice. A constellation launched into a perfectly clean shell, with no legacy junk in it at all, can still cross its own threshold, because its members are numerous enough to do it to each other. Meeting the disposal rules is not a defense. Neither is being small.

Those rules are mostly about disposal and reliability. International guidance has long asked that a spacecraft clear low Earth orbit within 25 years of finishing its mission, and in 2022 the U.S. Federal Communications Commission cut that to five years for satellites it licenses. Both are per-satellite standards, and both are applied one filing at a time.

The trouble with a per-satellite standard is easiest to feel through an example the paper does not use. A disposal success rate of 99% is an excellent number for a single spacecraft, and a regulator would sign it off without hesitation. Applied to a constellation of 10,000, the same rate leaves 100 dead objects drifting through a shell the other 9,900 have to fly in, topped up every time the fleet is replaced. Nothing about the individual satellite has changed. Only the number of them has.

That is the argument aimed at regulators, and Lewis makes it directly. National authorities are trying to limit the risk through mitigation requirements, the abstract says, but those requirements "fail to address the large scale and long lifetimes of satellite constellations." Instead, "evaluations are based on the risks from single satellites, and the aggregate risks across constellations are neglected." He attributes the gap partly to the absence of any usable method: there has been no simple way to score a constellation plan as a whole, which is the hole his approach is shaped to fill.

The paper runs to 20 pages and six figures, filed under planetary astrophysics and instrumentation, and it carries no journal reference and no note of submission to one. That makes it a preprint. It has not been through peer review, and its numbers are best read as a criterion offered for argument rather than a forecast anyone has yet checked. It is also a single-author analytical model rather than a full evolutionary simulation of the debris environment, which is the standard tool in this field and the one that will eventually be used to test it.

The abstract does not identify the constellations it places above their critical size. Lewis says in his closing line: the results "demonstrate the need for more stringent approaches to regulation and suggest careful consideration of some recent constellation plans." His method is deliberately simple. If it survives review, a regulator holding an application for a fleet of several thousand satellites could work out in a single calculation whether the constellation on the page is above its own line, before the first one launches.