Radio spectrum crowding of the sky

Why does every operator's perfectly legal transmitter add up to a sky astronomers cannot hear?

Radio astronomers have been complaining for seventy years that the sky is getting noisier, and the complaint has a strange shape. They are not describing a violation. Almost every emitter they blame is licensed, compliant, and operating exactly as its regulator intended.

So the interesting question is not "who is breaking the rules". It is: how does a system in which every single participant is obeying the rules produce an outcome nobody wants and nobody chose? That is a specific kind of failure, and naming it precisely tells you which fixes can work.

Start with the receiver, because the asymmetry begins there. A radio telescope is trying to detect signals that have travelled for millions of years and arrive with almost no power left in them. Observatories reach for the same comparison to convey it: their instruments could pick out a mobile phone's handset transmitter somewhere out near the Moon. So the threshold at which an unwanted signal becomes damaging is not a little below the level of ordinary radio traffic — it is many orders of magnitude below it.

Now ask what a licence actually certifies. It certifies one device, measured against an emission mask, in its assigned band. Nothing in that certificate mentions how many other certified devices are nearby. And so we can see the first gap: the unit of regulation is the individual transmitter, while the unit of harm is the sum arriving at one antenna.

That gap alone would not matter if the sum were small. So push on it. Suppose the regulator sets each device's permitted spill so that a single one contributes a thousandth of the damaging level. One emitter is harmless. A thousand of them, in view of the same dish, is exactly at the limit. Ten thousand is ten times over. Nothing changed about any device's behaviour; only the count changed. Does any individual operator have a reason to restrain? Their own marginal contribution really is a thousandth — they are telling the truth when they say removing it would not save the observation. And the benefit of their restraint would fall on astronomers, not on them.

That is the collective-action structure in its purest form: the cost of restraint is private and concentrated, the benefit is diffuse and cannot be withheld from anyone, and each participant's honest self-assessment says "not me". No amount of individual good faith resolves it, which is why appeals to good behaviour have never been the mechanism that works.

But there are protected bands, you might object — surely the answer is simply to fence off the frequencies astronomy uses. Two problems. The first is that astronomy does not have fixed frequencies. The hydrogen line is emitted at 1420 MHz, but expansion stretches it, so light from a distant galaxy arrives somewhere lower down, in a band allocated to somebody else. Protecting the rest frequency protects only the nearby universe. The second is that fences leak. A transmitter compliant inside its own band still radiates a little outside it, and satellites radiate something else besides: in 2023 the LOFAR array detected unintended electromagnetic emission from Starlink satellites between roughly 110 and 188 MHz, leakage from onboard electronics rather than from any communications payload, falling partly inside a band reserved for radio astronomy. Nothing about that was illegal. Terrestrial equipment is governed by electromagnetic-compatibility rules for exactly this stray radiation; spacecraft largely are not.

So how do you solve a collective-action problem? The classical answers are to assign rights over the resource, or to build an institution that coordinates the participants. Radio astronomy has tried both. The rights answer is the radio quiet zone — the United States established one around Green Bank in 1958, and Australia and South Africa have since ringed their major arrays with similar protections. It works by geography: radio power falls off with distance, so keeping emitters far away keeps their contribution small. The coordination answer is the ITU-R study group that sets protection thresholds, plus direct negotiation, such as the arrangement under which SpaceX steers beams away from radio observatories.

Now notice what the satellite case does to the first answer. A quiet zone is a circle drawn on the ground. Every satellite in a constellation of thousands has line of sight into that circle from above. Distance no longer helps, and the exclusion does not exclude. The geographic assignment of rights was a good answer to a local externality, and the externality has stopped being local.

Three refinements worth holding onto.

First, this is not a story about wrongdoing, and framing it as one leads to the wrong fixes. Enforcement catches violators; here there are none to catch. The correction has to change what compliance means — an aggregate limit, or an emission standard for spacecraft electronics — rather than police behaviour against the standard we already have.

Second, "spectrum sharing" is not a euphemism for astronomy losing. Sharing schemes that genuinely work exist: time-based coordination, geographic beam avoidance, and the practice of dumping data taken during a known satellite pass. What they have in common is that they operate on the aggregate, by agreement, rather than on one device at a time.

Third, a caution about certainty. How much science is actually lost is contested and depends on the frequency, the observation and the constellation — the mechanism above is solid, the magnitude is an active argument, and I would not pretend otherwise.

• How orbital debris shares the same per-actor-compliant, aggregate-catastrophic structure.
• Which redshifted spectral lines fall outside the protected bands, and why those bands sit where they do.
• What an aggregate emission limit would have to look like to be enforceable across national regulators.

Every term the collection defines is gathered in the glossary.