In the six months ending May 31, SpaceX satellites performed 207,152 propulsive collision-avoidance maneuvers. That is more than 1,100 a day. The figure is not evidence of recklessness. It reflects a conservative safety threshold, a fleet operating at unprecedented scale, and an unusually transparent public accounting.
It is also a policy alarm bell. At this volume, a rule inside software is no longer merely advice to an operator. It becomes physical behavior in orbit. Before America’s satellite fleets grow more autonomous, Washington must decide what turns a collision warning into permission to move a spacecraft.
The tempting answer is a number. Software calculates a probability of collision. Above a set threshold, the satellite maneuvers; below it, the satellite holds course. That sounds efficient. It also confuses two different things: an estimate of risk and an authorization to act.
NASA’s conjunction-assessment process shows why the distinction matters. It separates screening, risk assessment, and mitigation. Screening identifies close approaches that deserve attention. Assessment weighs uncertainty and consequence. Only then does mitigation begin. A collision probability is not a sensor reading carved into nature. It is a model-dependent estimate built from predicted trajectories, covariance data, assumptions about object size, tracking quality, and encounter geometry.
Two encounters can carry the same reported probability and still justify different choices. One may involve fresh, reliable tracking and a cooperative operator. The other may involve stale data, an unrealistic covariance estimate, a maneuverable secondary object whose planned actions are unknown, or a satellite whose mission cannot tolerate an abrupt course change. The maneuver itself consumes propellant, alters future conjunctions, and may prompt another operator to move as well.
President Donald Trump’s Space Policy Directive-3 anticipated this scale problem. The 2018 policy directed the government to provide basic space-safety data and services, placed the public-facing function in a civil agency, encouraged automation, and preserved room for American commercial leadership. Its division of labor was sound: government supplies a common safety picture; owners and operators assess the warning and retain responsibility for what their spacecraft does.
The Commerce Department’s Traffic Coordination System for Space, or TraCSS, is now turning that policy into infrastructure. As of August, it had 70 pilot users covering more than 11,345 satellites, along with national government accounts from 10 countries. That is a strong American platform for safer and more transparent space commerce.
But automation can erase the Trump framework’s boundary without changing a single agency box on an organization chart. If a government or commercial warning flows directly into an autonomous maneuver rule, a risk estimate can become a flight command without anyone marking the boundary. The operator remains responsible, but the decisive policy choice may be buried in a threshold selected months earlier by an engineer, vendor, or default setting.
Washington should not answer that risk by prescribing a single universal collision-probability threshold. SpaceX reports maneuvering above a probability of 3 in 10 million, more than two orders of magnitude below the 1-in-10,000 threshold it cites as the industry standard, while also applying miss-distance criteria. That may be sensible for its spacecraft, data, propulsion, and fleet design. The same number could be too cautious or not cautious enough for another mission. A single federal trigger would create false precision and flatten safety judgment into central planning.
The better approach is to regulate the boundary, not the number. Three narrow rules would preserve innovation while keeping authority accountable.
First, “risk detected” and “maneuver authorized” should be separate machine states. A collision estimate may trigger attention, additional tracking, coordination, or the preparation of maneuver options. A maneuver should execute only inside a preapproved envelope that defines permitted displacement, timing, fuel cost, mission effect, and expected residual risk.
Second, degraded evidence should change the authority envelope. Stale tracking, implausible ephemerides, conflicting models, failed communications, or uncertainty about another operator’s plans should not silently expand a machine’s freedom merely because the clock is running down. A system may need a preauthorized fail-safe maneuver, but poorer inputs should narrow its unsupervised options and trigger stronger review when time permits. Confidence displayed on a screen is not confidence earned from the data.
Third, every autonomous collision-avoidance maneuver should create a durable “maneuver receipt.” The record should identify the warning, data age, covariance and model versions, alternatives considered, coordination attempted, authority state, expected residual risk, and the owner or operator responsible for the policy. Regulators do not need a live joystick. They do need a record that makes it possible to reconstruct why a machine was permitted to move.
These rules can preserve machine speed. Small, routine avoidance maneuvers within a certified envelope could be preauthorized and executed automatically. Higher-consequence, fuel-intensive, or strategically ambiguous maneuvers would require stronger evidence, coordination, or human approval. The point is to make autonomy proportional to consequence.
The distinction matters beyond commercial safety. Civil, military, allied, and foreign spacecraft share the same orbital environment, even when they cannot share every piece of data. A poorly governed maneuver can create risks across those boundaries. An auditable authority architecture allows the United States to demand responsible behavior without requiring operators to expose proprietary algorithms or national-security secrets.
America also has an opportunity to export this model. The European Space Agency is already working toward automated collision avoidance, while TraCSS is attracting foreign government participation. A U.S.-led standard that combines common safety information with decentralized, accountable maneuver authority would be more credible than either a global traffic controller or an orbital free-for-all.
SpaceX’s 207,152 maneuvers should not trigger panic or a campaign against the company that disclosed them. They should focus policy on the transition already underway. As automated fleets move from calculating options to selecting and executing them, the U.S. must preserve the difference between evidence and permission.
CODING COSTS NOTHING NOW. AMERICA IS ABOUT TO STRIKE GOLD
Trump’s policy supplied the right foundation: public safety data, commercial innovation, and operator responsibility. The next step is to carry that logic into the software. A warning may inform the decision. The authority to maneuver must remain separately defined, limited, and attributable.
The goal is not to put a bureaucrat in every loop. It is to keep a probability from becoming a permission slip.
Burak Oktenli is an independent researcher based in Washington, D.C., working on human-machine authority architecture and the governance of autonomous decision systems. He holds an MBA and is pursuing a Master of Professional Studies in Applied Intelligence at Georgetown University. His analysis has appeared in RUSI, RealClearDefense, the Washington Examiner, Eurasia Review, Geopolitical Monitor, Britain’s World, and The Space Review.