I pulled the public orbital elements on all three satellites in the viral "SpaceX froze a NATO spy satellite so Russia could intercept it" brief and ran the numbers myself. The data shows two ordinary things happened. The explosive story welded onto them isn't supported, and in specific, checkable places it's flatly false. WHAT ACTUALLY HAPPENED (public Space-Track elements, SGP4): 1) A Russian inspector satellite, COSMOS 2612, made a close co-orbital approach to the commercial radar satellite ICEYE-X36. It closed to tens of meters at ~0 m/s on 16-18 Oct, then withdrew. Same orbital plane (incl 97.85°), and it actively maneuvered to get there. This is a real, notable Russian proximity operation. It required ZERO SpaceX involvement, because Russia tracks and maneuvers with its own sensors. 2) A Starlink (STARLINK-35856) had a routine conjunction with ICEYE-X36, ~0.45 km on 16 Oct 23:54 UTC. Real, but ordinary. Starlink sits in a totally different plane (incl 70° vs 98°), so this was a ~15 km/s crossing that lasts seconds. Every object in LEO gets these constantly. That's the whole factual content. Everything else is refuted by the data or isn't an orbital measurement at all. WHAT THE DATA REFUTES: ❌ "Intercept / immobilized / kill." All three satellites are on orbit and intact right now (elements epoch 25-26 Oct). ICEYE-X36 is alive and tracked. Cosmos 2612 withdrew to ~490 km and is still drifting away. A close approach that ends in withdrawal is an inspection, not a kinetic intercept. ❌ "X36 dodged a Russian approach in June, so it had to be frozen this time." Neither half happened. Cosmos 2612 never came within 50 km of X36 until 16 Oct. In June they were thousands of km apart. And X36 shows no June maneuver: smooth drag decay, with weekly averages falling from 546.1 to 545.0 km and a largest single step of -433 m (noise). A real 0.5 km avoidance burn would be a clean -500 m step. There is none. ❌ "SpaceX pinned/froze X36 in place." X36 never maneuvers, period. Over 165 days it deviates from its natural decay curve by no more than 278 m and holds inclination to 0.022°. There was nothing to "pin." It's a passively decaying microsat. For contrast, Cosmos 2612 shows a 47.8 km altitude burn in a single day. That is what a real maneuver looks like in the data. ❌ "SpaceX inflated the danger 5x, claiming the real distance was 1.9-2.0 km." Independent public elements give ~0.45 km for the Starlink pass, consistent with the ~0.4 km figure cited, not 1.9-2.0 km. That "5x safer" claim needs SpaceX's private ephemerides that no one outside SpaceX can see, and the public data points the opposite way. CREDIT WHERE DUE: the brief's raw archive numbers are real. The TLE update gaps and the conjunction time match my own pull. But multi-hour public TLE gaps are routine for every LEO object, and SpaceX having good space tracking is public knowledge used for collision avoidance. Neither is evidence of collusion. WHAT ORBITAL DATA CAN'T SETTLE: "coordination" with Russia, collision-alert intent, Starlink throttling, the 23% figure, Iran, finances. Those are assertions, not measurements. The brief's "99.9% probability" is that inference restated as a number, and several of its load-bearing inputs (the June evasion, the 5x inflation, "X36 could maneuver") are false, so it's garbage in. Orbital data can't prove a negative on intent. I can't prove no one coordinated. What I can show is that the data provides no evidence for it and contradicts the specific mechanisms claimed. Bottom line: a real Russian satellite inspection, which was already public and known, plus a routine Starlink flyby. Two unrelated events, one narrative bolted on top. All checks from public Space-Track two-line elements via SGP4. Happy to share the numbers.