ESO's position paper argues that above ~100,000 objects brighter than magnitude 7, streak density in wide-field exposures crosses a threshold where survey pipelines like Rubin's LSST cannot algorithmically recover the underlying sky. Their brightness models and streak-density calculations show masking works below the threshold but 'eats the data' above it, threatening the scientific premise of decade-long all-sky surveys.
By submitting the ESO paper to Hacker News with the framing 'No more than 100,000 faint satellites should orbit Earth,' the submitter foregrounds the cap as the central argument worth community attention. The 95-point score signals broad endorsement of the concern among the HN audience.
The editorial argues the interesting story isn't the astronomy but the fact that no body has authority to enforce a global brightness or streak-count cap. The FCC regulates spectrum, the ITU coordinates orbital slots and frequencies, and COPUOS lacks teeth — meaning ESO's number functions as a countdown timer against filed constellation intent already approaching the threshold.
The European Southern Observatory published a position paper this week — cataloged as eso2607 — arguing that the total population of faint artificial satellites in low Earth orbit must be kept under 100,000 if ground-based astronomy is to survive as a functional science. The paper is signed by ESO's Director General and coordinated with observatories including Vera C. Rubin, Gemini, and Keck. It is not a press release dressed up as science; it is a working technical document with brightness models, streak-density calculations, and an appeal to the UN Committee on the Peaceful Uses of Outer Space (COPUOS).
The number is not arbitrary. ESO's modeling shows that above roughly 100,000 objects brighter than magnitude 7, the density of satellite streaks in wide-field exposures crosses a threshold where survey pipelines — the Rubin Observatory's LSST especially — cannot algorithmically recover the underlying sky. Below that, streak-masking works. Above it, the masks eat the data. Rubin is designed to image the entire southern sky every three nights for a decade; its whole scientific premise assumes the sky is mostly sky.
As of mid-2026 there are approximately 11,000 active satellites in LEO, dominated by SpaceX's Starlink at ~8,000. Announced constellations from Amazon (Kuiper, 3,236), China's Guowang (~13,000) and Qianfan/Thousand Sails (~14,000), and a long tail of smaller operators put the near-term filed total north of 60,000. The FCC has approvals pending that would push filed intent past 100,000 before 2030. ESO's paper is, functionally, a countdown timer.
The interesting part isn't the astronomy — it's the governance gap. Nobody actually has the authority to enforce a global cap on satellite brightness. The FCC regulates U.S. spectrum licenses. The ITU coordinates orbital slots and frequency assignments. Neither has a mandate over reflectivity, streak count, or the aggregate darkness of the night sky. COPUOS operates by consensus and has no enforcement teeth. The Outer Space Treaty of 1967 predates the concept of a mega-constellation by half a century and treats space as a commons in the loosest possible sense.
This is a familiar shape to anyone who has worked on shared infrastructure. It's the tragedy of the commons with a twist: the commons is a *view*, and the actors degrading it are individually rational. Starlink's marginal satellite is worth several hundred thousand dollars in revenue over its lifetime. The marginal cost imposed on Rubin's LSST is diffuse, non-monetary, and borne by a global scientific public that doesn't write checks. There is no market mechanism to close the loop. SpaceX has, to its credit, engaged — the DarkSat and VisorSat experiments knocked roughly a magnitude off Starlink's brightness, and the current v2 mini design incorporates dielectric mirror films. But even fully-mitigated Starlinks are visible to Rubin's cameras, and China's constellations have made no such commitments.
The community reaction has been sharper than the paper's diplomatic tone. Tony Tyson, chief scientist for Rubin/LSST, has been on record for two years that streak contamination is now the observatory's single largest operational risk — ahead of weather, ahead of instrument failure. A 2024 Nature paper estimated that at 100,000 satellites, roughly 30% of Rubin's twilight exposures would be unrecoverable, and transient-detection pipelines would miss an estimated 50% of near-Earth asteroids in the critical 6-hour warning window. That last number is the one that should land: the same constellations enabling global broadband are quietly degrading planetary defense.
Radio astronomy has an even worse version of the same problem. Starlink v2 satellites emit unintended electromagnetic radiation across bands that aren't formally protected but that radio observatories have used for decades. The SKA (Square Kilometre Array), when it comes online later this decade, will operate in a South African radio-quiet zone specifically to escape terrestrial interference — but there is no orbital-quiet zone.
If you build in space, the regulatory environment is about to get non-trivially harder. Expect brightness and emission constraints to become licensing conditions within the FCC's next rulemaking cycle, and expect the EU to move first with a hard cap on member-state authorizations. The days of "launch first, mitigate if pressured" are ending — not because operators want them to, but because the science community has finally aligned on a specific, defensible number.
For developers working on Earth-observation pipelines, transient detection, or any astronomy-adjacent data product: budget for streak-removal in your CV stack now. Rubin is releasing its streak-masking code as open source (`rubin_sim` and the `dm-stack` on GitHub), and it's the best-in-class implementation. If your pipeline touches wide-field imagery from Pan-STARRS, ZTF, or the coming LSST releases, you'll want to fork it rather than roll your own — the edge cases in streak-detection are brutal (satellite glints look like real transients, real transients get masked, and calibrating the trade-off is a research problem in itself).
And for anyone building on Starlink or Kuiper backhaul: the political tailwind against mega-constellations is now scientific and cross-partisan. It won't stop launches, but it will slow filings, add compliance overhead, and probably raise per-terminal costs over the medium term as operators bake mitigation into hardware.
The honest read is that 100,000 is a ceiling nobody can enforce and everyone will blow through. The paper's real function is to establish a Schelling point — a number the scientific community will now reference in every filing objection, every UN submission, every FCC comment period for the next decade. That's how soft governance works when hard governance is impossible: you pick a number, you defend it publicly, and you make crossing it politically expensive. Whether it's enough to save Rubin's decade-long survey is the question. Whether it's enough to save the concept of a dark sky at all is a different, sadder one.
We don’t take our cities sound and light pollution seriously.I bet it plays important role in all the anxiety and other popular health issues rising.
I am a science and astronomy fan, but I am sorry, in this case progress is more important. If we regret our decision, the LEOs will fall out of the sky by themselves in a few years and it will be ok.
This is a tradeoff we have to make with infrastructure and development in general. How do you balance human needs with pristine nature?Do we put up long-distance power lines and wind farms even though they ruin the views? Do you tear down a forest to put up farmlands and suburbs? Do you build a dam
I'm not sure how exactly they are making these calculations but I just don't see it. Both Reflect and SpaceX are targeting SSO orbits where they are only reflecting for an hour or two at sunset. That isn't true of Starlink, but that constellation is already up there and if its fine ri
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
This is not -progress-, is speculaton, at astronomical level, as 1M sat in orbit, only a very small part of humans will benefit (shareholders, but i doubt all), all the rest not even. We really need to surf web at Gb/s speed in the mid of nowhere?