Un morceau de quatre tonnes d’une fusée SpaceX a passé dix-huit mois sur une trajectoire de collision avec la Lune, commençant à peu près au moment où tout le monde a cessé d’y penser. Le 5 août, il est arrivé.
La partie intéressante n’est pas la collision. C’est que la collision a été programmée dix-huit mois à l’avance par une décision que personne ne se souvient avoir prise.
Oui. Un étage supérieur du SpaceX Falcon 9 épuisé s'est écrasé sur la Lune le 5 août 2026, vers 06h34 UTC – 02h34 HAE – près du cratère Einstein sur le flanc ouest. L'orbiteur sud-coréen Danuri a confirmé l'impact lunaire, en photographiant le site au cours de huit passages qui ont commencé environ trente minutes avant le contact. L'Institut coréen de recherche aérospatiale a rapporté que "des changements dans le terrain autour du site d'impact et des traces de propagation d'éjectas ont été confirmés". À l'observatoire Lowell en Arizona, l'équipe de Carl Schmidt a mesuré un panache de gaz de sodium et de lithium de plusieurs dizaines de kilomètres de diamètre qui a duré cinq à dix minutes.
Ce qui suit n’est pas une histoire spatiale. C’est ce que dit le crash de chaque système qui survit à l’attention qui lui est portée.
L'objet n'a pas de nom, seulement une désignation catalogue : 2025-010D. Il s'agissait du quatrième matériel suivi du dixième lancement orbital de 2025, le 15 janvier, qui transportait l'atterrisseur Blue Ghost de Firefly Aerospace et le Resilience d'ispace vers la Lune. Blue Ghost a atterri avec succès en mars. Resilience a perdu le contact environ quatre-vingt-dix secondes avant l'atterrissage et s'est écrasé en juin.
Les étages supérieurs du Falcon 9 sont normalement désorbités peu de temps après le lancement et brûlés dans l'atmosphère. Cela ne s'est pas produit ici. Pousser deux atterrisseurs vers leur point de déploiement a consommé presque tout le propulseur de l'étage, de sorte que le brûlage d'élimination n'était plus disponible. Selon les termes d'une étude menée par Benjamin Fernando du Laboratoire national de Los Alamos, la scène "a été abandonnée alors qu'elle traversait la Lune sur une orbite terrestre élevée".
Il a ensuite passé plus de dix-huit mois sur une orbite déséquilibrée de 26 jours, oscillant entre environ 220 000 et 510 000 km de la Terre, croisant à plusieurs reprises la trajectoire de la Lune, la ratant à chaque fois, jusqu'à ce qu'elle ne le fasse plus. L'astronome Bill Gray du projet Pluton a estimé la vitesse d'impact à 2,43 km/s et l'énergie libérée à environ 14,5 milliards de joules, soit environ trois tonnes de TNT.
Personne n’a choisi d’écraser une fusée sur la Lune. Le choix qui comptait avait été fait bien plus tôt, et cela ressemblait à un budget carburant.
Le déclassement était l'élément de campagne utilisé pour faire fonctionner la mission principale. Il n'y a eu aucune réunion à ce sujet, aucune inscription au registre des risques, aucun ingénieur dissident. La mission a été plus difficile que la marge autorisée, la marge est sortie de la fin de vie et le plan de fin de vie a tranquillement cessé d'exister.
C’est la façon la plus courante pour les systèmes logiciels de devenir permanents. La migration est longue, donc le basculement consomme le trimestre, donc la mise hors service de l'ancien système passe au « prochain sprint ». L’ancien système continue de desservir un filet de trafic. Deux ans plus tard, il est porteur et personne dans l'équipe n'a lu son code.
Le mode d’échec n’est pas la négligence. C’est que le démantèlement est la seule partie d’un projet dont rien ne respecte les délais.
SpaceX ne s’est pas simplement éloigné. Julianna Scheiman, directrice des programmes scientifiques et Dragon de la NASA, a déclaré qu'une manœuvre différente avait été effectuée "pour garantir que la deuxième étape elle-même est sûre, conformément aux règles et réglementations appropriées".
Cette manœuvre satisfaisait aux règles. Cela n’a placé l’objet dans aucun endroit durable. Comme le dit Scheiman, "un mélange d'activité solaire et de forces de gravité l'a mis sur la voie de la Lune".
Un contrôle qui réussit l'examen lors de la mise en service est une déclaration concernant la mise en service. Il ne s’agit pas d’une déclaration sur l’état du système dix-huit mois plus tard, et traiter les deux comme équivalents est la manière dont le risque de modèle s’accumule. La signature est un instantané ; les systèmes sont une trajectoire.
La force qui le déplaçait était la lumière du soleil. La pression du rayonnement solaire – les photons frappant la scène de chute – est, selon la description de Gray, « une force extrêmement douce, mais au fil des mois, elle peut réellement s’accumuler ». Ce n’est pas non plus entièrement prévisible, car un objet qui tombe capte et réfléchit la lumière différemment lorsqu’il tourne.
Aucune poussée n’avait d’importance. Dix-huit mois d'entre eux ont déterminé où aboutiraient quatre tonnes et demi de métal.
Il s’agit d’une dérive, exactement telle qu’elle apparaît dans un pipeline de notation : aucun déplacement de population d’une semaine ne mérite une alerte, et le déplacement cumulé sur dix-huit mois mérite une constatation réglementaire. Si votre cadence de validation est plus lente que votre accumulation de dérive, vous ne surveillez pas le modèle. Vous le visitez simplement de temps en temps.
Voici le détail qui nous a semblé le plus inconfortable. Cet objet a été suivi principalement par des chercheurs d'astéroïdes et des astronomes amateurs, car les systèmes construits pour les débris spatiaux ne peuvent pas voir aussi loin. Le radar excelle en orbite terrestre basse – Gray note qu'il a suivi les gants et les sacs à outils perdus par les astronautes – mais à la distance lunaire, le signal de retour est environ 25,6 milliards de fois plus faible. Gray calcule les orbites des objets à haute altitude spécifiquement parce que ce sont ceux que « l’armée ne suit pas ».
Les enquêtes qui l’ont constaté n’essayaient pas de le faire. Leur travail est la défense planétaire et, comme le dit Gray, « le temps passé à observer les débris est du temps non consacré à la recherche de roches ».
Chaque organisation d'ingénierie a cet arrangement et l'appelle autrement. Le service obsolète n'a pas de tableau de bord ni de rotation d'astreinte, et la seule raison pour laquelle quiconque sait qu'il est dégradant est qu'un ingénieur continue de le remarquer dans des journaux sans rapport. Ce n’est pas de l’observabilité. C'est un volontaire, et les volontaires partent.
La récupération ici est vraiment impressionnante et elle a été faite à partir de données. Fin février, les archives contenaient 1 053 observations, fournies par des observatoires du Mississippi, de l'Utah, de Pékin et d'Angleterre. Le logiciel de Gray a signalé l'impact lunaire en septembre 2025, soit onze mois à l'avance.
Les prévisions se sont ensuite resserrées à mesure que les observations s'accumulaient. Une révision de juin a déplacé l'impact prévu neuf minutes plus tôt et de 170 km vers le nord-ouest. Une révision de juillet l'a décalé de quelques secondes et d'une dizaine de kilomètres. Une dernière mise à jour est arrivée le 1er août.
Cette précision a porté ses fruits : Danuri a commencé à prendre des images environ trente minutes avant la collision et a effectué huit passages sur le site, capturant une véritable couverture avant et après. Vous ne pointez un instrument sur la bonne parcelle de terrain au bon moment que si quelqu'un vous a donné une prévision. Et vous obtenez uniquement une prévision pour un système que quelqu’un est encore en train de mesurer.
Il convient de noter, sans vouloir trop revendiquer : au moment d’écrire ces lignes, aucune mesure publiée du nouveau cratère n’a été publiée. Gray a prédit un seul cratère d’environ 17 m de diamètre. Une grande partie de la couverture médiatique cite jusqu'à 27 m. Les deux sont des calculs préalables à l’impact et ne concordent pas.
Qu’il s’agisse d’un cratère ou de deux est également incertain. L'étage supérieur du Chang'e 5-T1 2022 a surpris tout le monde en réalisant un double cratère. Gray en attend un seul ici, car cette étape ne transportait pas de masse comparable au sommet – une inférence raisonnable, pas un résultat.
Les chiffres en circulation ne sont pas des mesures. Cette distinction constitue tout le travail de validation des modèles, et elle ne cesse d’être vraie car un nombre est répété avec confiance par tout le monde.
Traduisez les modes de défaillance et ils deviennent inconfortablement familiers.
| Qu'est-il arrivé au 2025-010D |
L'équivalent dans une pile de prêts |
| Élimination du carburant dépensé pour la mission principale |
Le déclassement a été supprimé pour protéger une date de lancement |
| Manœuvre conforme, puis pas de propriétaire |
Contrôle signé lors de la mise en service, jamais revérifié |
| Pression de rayonnement solaire sur 18 mois |
Dérive de la population entre les cycles de validation |
| Invisible au radar à distance lunaire |
Pas de télémétrie sur le service obsolète |
| Suivi par des bénévoles avec d'autres emplois |
Un ingénieur qui sait que ça existe |
Concrètement, les choses qui méritent d'être auditées ce trimestre :
- Des tableaux de bord retirés qui marquent toujours. Une version de modèle remplacée est restée connectée à un chemin de secours, décidant tranquillement d'une tranche d'applications.
- Magasins de documents orphelins. Des seaux de relevés bancaires de clients provenant d'un pilote qui s'est terminé. Personne n’en est propriétaire, la politique de rétention n’a jamais été appliquée et il s’agit d’une pure responsabilité.
- Pipelines d’extraction avec un appelant restant. L'ancien analyseur était conservé en vie pour une intégration de prêteur unique, non entretenu et non surveillé.
- Les informations d’identification survivent à leur objectif. Clés API pour un fournisseur de données que vous avez arrêté de payer, toujours valides, toujours dans une configuration quelque part.
- Connecteurs pour les formats qui ont changé. Une banque a repensé la présentation de ses relevés ; le connecteur tombe désormais en panne silencieusement sur un petit pourcentage et personne n'a déclenché d'alarme.
Aucun de ces éléments ne vous bipera. C'est précisément le problème : ils échouent d'une manière qui ne produit aucun signal jusqu'à ce que l'échec soit découvert par quelqu'un d'autre.
La solution, en mécanique orbitale, est bon marché – si vous décidez tôt. Placez l'étage supérieur sur une trajectoire qui quitte complètement le système Terre-Lune et finit par tourner autour du Soleil. L'étage Falcon 9 du lancement d'EscaPADE en novembre 2025 a été éliminé de cette manière. Les lancements lunaires chinois ont évolué dans la même direction.
Mais cela ne fonctionne que comme une décision au moment de la conception. Une fois le propulseur parti, aucune option n’existe à aucun prix. Gray est également honnête quant aux limites : l’astuce « ne fonctionne que pour une petite fraction des déchets », à savoir les missions déjà dirigées au-delà de l’orbite terrestre. Certaines classes du système ne peuvent véritablement pas être retirées proprement, et il est temps de savoir dans quelle classe vous vous trouvez avant de construire.
La question soulevée par Scheiman est la bonne à retenir : « Quelle est la meilleure voie d'élimination future pour les missions à haute énergie qui se déroulent dans le système Soleil-Terre-Lune ? La version du logiciel est plus petite et plus ennuyeuse et tout aussi sans réponse. Quelle est la voie de retrait de ce modèle, de ce pipeline, de ce magasin de données ? À qui appartient-il après la soirée de lancement ? Quel signal nous indique qu’il fonctionne toujours, et qui reçoit ce signal ?
Les systèmes que vous ne pouvez pas voir sont des systèmes que vous ne pouvez pas mettre hors service. Lev Document Intelligence de Levmetrics est conçu de telle sorte que chaque extraction, score et décision laisse une trace vérifiable, y compris celles que vous vouliez désactiver. Essayez-le sur app.levmetrics.com.
Un étage de fusée de quatre tonnes est un professeur utile car il est impossible de perdre poliment le fil. Il a gardé une orbite, un numéro de catalogue, et dix-huit mois après que tout le monde ait déclaré la mission un succès, il est arrivé quelque part que personne n'avait prévu, à une vitesse de 2,43 kilomètres par seconde. La plupart des systèmes abandonnés sont moins dramatiques et plus coûteux. Le déploiement est une étape importante dans la vie d'un système. Ce n’est pas près de la fin.
- Mike Wall, "A SpaceX rocket will crash into the moon this week. Why is it on a collision course?", Space.com, August 3, 2026.
- Chelsea Gohd, "A SpaceX rocket just crashed into the moon. Now, the race is on to get a look at the impact site", Space.com, August 5, 2026.
- Mike Wall, "SpaceX rocket's impact crater on the moon spied by Korean lunar orbiter (photos)", Space.com, August 6, 2026.
- Bill Gray, "Upper stage impacting the moon on 2026 August 5", Project Pluto.
- Adam Kovac, "A SpaceX rocket booster is on track to hit the moon at several times the speed of sound", Scientific American, May 1, 2026.
- Leonard David, "Moon Impact of Rocket Stage Observed", Inside Outer Space, August 5, 2026.
- Benjamin Fernando et al., "preprint on the predicted lunar impact of the 2025-010D upper stage", arXiv.
A four-tonne piece of a SpaceX rocket spent eighteen months on a collision course with the Moon, starting at roughly the moment everyone stopped thinking about it. On August 5 it arrived.
The interesting part is not the collision. It is that the collision was scheduled eighteen months in advance by a decision nobody remembers making.
Yes. A spent SpaceX Falcon 9 upper stage crashed into the Moon on August 5, 2026, at about 06:34 UTC — 2:34 a.m. EDT — near the crater Einstein on the western limb. South Korea's Danuri orbiter confirmed the lunar impact, imaging the site across eight passes that began roughly thirty minutes before contact. The Korea Aerospace Research Institute reported that "changes in the terrain around the impact site and traces of ejecta spread were confirmed." At Lowell Observatory in Arizona, Carl Schmidt's team measured a sodium and lithium gas plume tens of kilometres across that lasted five to ten minutes.
What follows is not a space story. It is what the crash says about every system that outlives the attention paid to it.
The object has no name, only a catalogue designation: 2025-010D. It was the fourth tracked piece of hardware from the tenth orbital launch of 2025, on January 15, which carried Firefly Aerospace's Blue Ghost lander and ispace's Resilience toward the Moon. Blue Ghost landed successfully that March. Resilience lost contact about ninety seconds before touchdown and crashed in June.
Falcon 9 upper stages are normally deorbited shortly after launch and burned up in the atmosphere. That did not happen here. Pushing two landers to their deployment point consumed nearly all of the stage's propellant, so the disposal burn was no longer available. In the words of a study led by Benjamin Fernando of Los Alamos National Laboratory, the stage "was abandoned in a moon-crossing high-Earth orbit."
It then spent more than eighteen months in a lopsided 26-day orbit, swinging between roughly 220,000 and 510,000 km from Earth — repeatedly crossing the Moon's path, missing it every time, until it didn't. Astronomer Bill Gray of Project Pluto put the impact speed at 2.43 km/s and the released energy at about 14.5 billion joules, roughly three tons of TNT.
Nobody chose to crash a rocket into the Moon. The choice that mattered was made much earlier, and it looked like a fuel budget.
Decommissioning was the line item that got consumed to make the primary mission work. There was no meeting about it, no risk register entry, no dissenting engineer. The mission was harder than the margin allowed, the margin came out of end-of-life, and the end-of-life plan quietly ceased to exist.
This is the most common way software systems become permanent. The migration runs long, so the cutover eats the quarter, so the decommissioning of the old system slips to "next sprint." The old system keeps serving a trickle of traffic. Two years later it is load-bearing and nobody on the team has read its code.
The failure mode is not negligence. It is that decommissioning is the only part of a project whose deadline nothing enforces.
SpaceX did not simply walk away. Julianna Scheiman, the company's director of NASA science and Dragon programs, said a different maneuver was performed "to ensure that the second stage itself is safe per the appropriate rules and regulations."
That maneuver satisfied the rules. It did not put the object anywhere durable. As Scheiman put it, "a mixture of solar activity and gravity forces have put it on a path towards the moon."
A control that passes review at go-live is a statement about go-live. It is not a statement about the system's state eighteen months later, and treating the two as equivalent is how model risk accumulates. Sign-off is a snapshot; systems are a trajectory.
The force that moved it was sunlight. Solar radiation pressure — photons striking the tumbling stage — is, in Gray's description, "an extremely gentle force, but over months, it can really build up." It is also not fully predictable, because a tumbling object catches and reflects light differently as it turns.
No single push mattered. Eighteen months of them decided where four-and-a-bit tonnes of metal ended up.
That is drift, exactly as it appears in a scoring pipeline: no individual week's population shift is worth an alert, and the cumulative eighteen-month displacement is worth a regulatory finding. If your validation cadence is slower than your drift accumulation, you are not monitoring the model. You are just visiting it occasionally.
Here is the detail we found most uncomfortable. This object was tracked mainly by asteroid surveys and amateur astronomers, because the systems built for space debris cannot see that far out. Radar excels in low Earth orbit — Gray notes it has tracked gloves and tool bags lost by astronauts — but at lunar distance the return signal is about 25.6 billion times fainter. Gray computes orbits for high-altitude objects specifically because they are the ones "the military doesn't track."
The surveys that did see it were not trying to. Their job is planetary defence, and as Gray puts it, "time spent observing junk is time not spent finding rocks."
Every engineering organisation has this arrangement and calls it something else. The deprecated service has no dashboard and no on-call rotation, and the only reason anyone knows it is degrading is that one engineer keeps noticing it in unrelated logs. That is not observability. That is a volunteer, and volunteers leave.
The recovery here is genuinely impressive, and it was made of data. By late February the archive held 1,053 observations, contributed by observatories in Mississippi, Utah, Beijing and England. Gray's software flagged the lunar impact in September 2025 — eleven months ahead.
The forecast then tightened as observations accumulated. A June revision moved the predicted impact nine minutes earlier and 170 km to the northwest. A July revision moved it by a couple of seconds and about ten kilometres. A final update landed on August 1.
That precision had a payoff: Danuri began imaging roughly thirty minutes before the collision and made eight passes over the site, capturing genuine before-and-after coverage. You only point an instrument at the right patch of ground at the right minute if someone gave you a forecast. And you only get a forecast for a system somebody is still measuring.
Worth noting, in the spirit of not overclaiming: as of this writing there is no published measurement of the new crater. Gray predicted a single crater around 17 m across. Much of the news coverage cites up to 27 m. Both are pre-impact calculations, and they do not agree with each other.
Whether it is one crater or two is also unsettled. The 2022 Chang'e 5-T1 upper stage surprised everyone by making a double crater. Gray expects a single one here because this stage carried no comparable mass at the top — a reasonable inference, not a result.
Numbers in circulation are not measurements. That distinction is the whole job in model validation, and it does not stop being true because a number is repeated confidently by everyone.
Translate the failure modes and they get uncomfortably familiar.
| What happened to 2025-010D |
The equivalent in a lending stack |
| Disposal fuel spent on the primary mission |
Decommissioning descoped to protect a launch date |
| Compliant maneuver, then no owner |
Control signed off at go-live, never re-verified |
| Solar radiation pressure over 18 months |
Population drift between validation cycles |
| Invisible to radar at lunar distance |
No telemetry on the deprecated service |
| Tracked by volunteers with other jobs |
One engineer who happens to know it exists |
Concretely, the things worth auditing this quarter:
- Retired scorecards that still score. A superseded model version left wired into a fallback path, quietly deciding a slice of applications.
- Orphaned document stores. Buckets of customer bank statements from a pilot that ended. Nobody owns them, retention policy never applied, and they are pure liability.
- Extraction pipelines with one remaining caller. The old parser kept alive for a single lender integration, unmaintained and unmonitored.
- Credentials outliving their purpose. API keys for a data provider you stopped paying, still valid, still in a config somewhere.
- Connectors for formats that changed. A bank redesigned its statement layout; the connector now fails silently on a small percentage and nobody set an alarm.
None of these will page you. That is precisely the problem: they fail in ways that produce no signal until the failure is someone else's finding.
The fix, in orbital mechanics, is cheap — if you decide early. Put the upper stage on a trajectory that leaves the Earth-Moon system entirely and ends up orbiting the Sun. The Falcon 9 stage from the EscaPADE launch in November 2025 was disposed of this way. Chinese lunar launches have moved in the same direction.
But it only works as a design-time decision. Once the propellant is gone, no option exists at any price. Gray is also honest about the limit: the trick "only works for a small fraction of junk," namely missions already headed beyond Earth orbit. Some classes of system genuinely cannot be cleanly retired, and the time to find out which class you are in is before you build.
The question Scheiman raised is the right one to steal: "What is the best future disposal path for high-energy missions that are in the sun-Earth-moon system?" The software version is smaller and duller and just as unanswered. What is the retirement path for this model, this pipeline, this data store? Who owns it after the launch party? What signal tells us it is still running, and who receives that signal?
Systems you cannot see are systems you cannot retire. Levmetrics' Lev Document Intelligence is built so that every extraction, score and decision leaves an auditable trace — including the ones you meant to turn off. Try it at app.levmetrics.com.
A four-tonne rocket stage is a useful teacher because it is impossible to lose track of politely. It kept an orbit, it kept a catalogue number, and eighteen months after everyone declared the mission a success it arrived somewhere nobody intended at 2.43 kilometres per second. Most abandoned systems are less dramatic and more expensive. Deployment is a milestone in a system's life. It is nowhere near the end of it.
- Mike Wall, "A SpaceX rocket will crash into the moon this week. Why is it on a collision course?", Space.com, August 3, 2026.
- Chelsea Gohd, "A SpaceX rocket just crashed into the moon. Now, the race is on to get a look at the impact site", Space.com, August 5, 2026.
- Mike Wall, "SpaceX rocket's impact crater on the moon spied by Korean lunar orbiter (photos)", Space.com, August 6, 2026.
- Bill Gray, "Upper stage impacting the moon on 2026 August 5", Project Pluto.
- Adam Kovac, "A SpaceX rocket booster is on track to hit the moon at several times the speed of sound", Scientific American, May 1, 2026.
- Leonard David, "Moon Impact of Rocket Stage Observed", Inside Outer Space, August 5, 2026.
- Benjamin Fernando et al., "preprint on the predicted lunar impact of the 2025-010D upper stage", arXiv.
A four-tonne piece of a SpaceX rocket spent eighteen months on a collision course with the Moon, starting at roughly the moment everyone stopped thinking about it. On August 5 it arrived.
The interesting part is not the collision. It is that the collision was scheduled eighteen months in advance by a decision nobody remembers making.
Yes. A spent SpaceX Falcon 9 upper stage crashed into the Moon on August 5, 2026, at about 06:34 UTC — 2:34 a.m. EDT — near the crater Einstein on the western limb. South Korea's Danuri orbiter confirmed the lunar impact, imaging the site across eight passes that began roughly thirty minutes before contact. The Korea Aerospace Research Institute reported that "changes in the terrain around the impact site and traces of ejecta spread were confirmed." At Lowell Observatory in Arizona, Carl Schmidt's team measured a sodium and lithium gas plume tens of kilometres across that lasted five to ten minutes.
What follows is not a space story. It is what the crash says about every system that outlives the attention paid to it.
The object has no name, only a catalogue designation: 2025-010D. It was the fourth tracked piece of hardware from the tenth orbital launch of 2025, on January 15, which carried Firefly Aerospace's Blue Ghost lander and ispace's Resilience toward the Moon. Blue Ghost landed successfully that March. Resilience lost contact about ninety seconds before touchdown and crashed in June.
Falcon 9 upper stages are normally deorbited shortly after launch and burned up in the atmosphere. That did not happen here. Pushing two landers to their deployment point consumed nearly all of the stage's propellant, so the disposal burn was no longer available. In the words of a study led by Benjamin Fernando of Los Alamos National Laboratory, the stage "was abandoned in a moon-crossing high-Earth orbit."
It then spent more than eighteen months in a lopsided 26-day orbit, swinging between roughly 220,000 and 510,000 km from Earth — repeatedly crossing the Moon's path, missing it every time, until it didn't. Astronomer Bill Gray of Project Pluto put the impact speed at 2.43 km/s and the released energy at about 14.5 billion joules, roughly three tons of TNT.
Nobody chose to crash a rocket into the Moon. The choice that mattered was made much earlier, and it looked like a fuel budget.
Decommissioning was the line item that got consumed to make the primary mission work. There was no meeting about it, no risk register entry, no dissenting engineer. The mission was harder than the margin allowed, the margin came out of end-of-life, and the end-of-life plan quietly ceased to exist.
This is the most common way software systems become permanent. The migration runs long, so the cutover eats the quarter, so the decommissioning of the old system slips to "next sprint." The old system keeps serving a trickle of traffic. Two years later it is load-bearing and nobody on the team has read its code.
The failure mode is not negligence. It is that decommissioning is the only part of a project whose deadline nothing enforces.
SpaceX did not simply walk away. Julianna Scheiman, the company's director of NASA science and Dragon programs, said a different maneuver was performed "to ensure that the second stage itself is safe per the appropriate rules and regulations."
That maneuver satisfied the rules. It did not put the object anywhere durable. As Scheiman put it, "a mixture of solar activity and gravity forces have put it on a path towards the moon."
A control that passes review at go-live is a statement about go-live. It is not a statement about the system's state eighteen months later, and treating the two as equivalent is how model risk accumulates. Sign-off is a snapshot; systems are a trajectory.
The force that moved it was sunlight. Solar radiation pressure — photons striking the tumbling stage — is, in Gray's description, "an extremely gentle force, but over months, it can really build up." It is also not fully predictable, because a tumbling object catches and reflects light differently as it turns.
No single push mattered. Eighteen months of them decided where four-and-a-bit tonnes of metal ended up.
That is drift, exactly as it appears in a scoring pipeline: no individual week's population shift is worth an alert, and the cumulative eighteen-month displacement is worth a regulatory finding. If your validation cadence is slower than your drift accumulation, you are not monitoring the model. You are just visiting it occasionally.
Here is the detail we found most uncomfortable. This object was tracked mainly by asteroid surveys and amateur astronomers, because the systems built for space debris cannot see that far out. Radar excels in low Earth orbit — Gray notes it has tracked gloves and tool bags lost by astronauts — but at lunar distance the return signal is about 25.6 billion times fainter. Gray computes orbits for high-altitude objects specifically because they are the ones "the military doesn't track."
The surveys that did see it were not trying to. Their job is planetary defence, and as Gray puts it, "time spent observing junk is time not spent finding rocks."
Every engineering organisation has this arrangement and calls it something else. The deprecated service has no dashboard and no on-call rotation, and the only reason anyone knows it is degrading is that one engineer keeps noticing it in unrelated logs. That is not observability. That is a volunteer, and volunteers leave.
The recovery here is genuinely impressive, and it was made of data. By late February the archive held 1,053 observations, contributed by observatories in Mississippi, Utah, Beijing and England. Gray's software flagged the lunar impact in September 2025 — eleven months ahead.
The forecast then tightened as observations accumulated. A June revision moved the predicted impact nine minutes earlier and 170 km to the northwest. A July revision moved it by a couple of seconds and about ten kilometres. A final update landed on August 1.
That precision had a payoff: Danuri began imaging roughly thirty minutes before the collision and made eight passes over the site, capturing genuine before-and-after coverage. You only point an instrument at the right patch of ground at the right minute if someone gave you a forecast. And you only get a forecast for a system somebody is still measuring.
Worth noting, in the spirit of not overclaiming: as of this writing there is no published measurement of the new crater. Gray predicted a single crater around 17 m across. Much of the news coverage cites up to 27 m. Both are pre-impact calculations, and they do not agree with each other.
Whether it is one crater or two is also unsettled. The 2022 Chang'e 5-T1 upper stage surprised everyone by making a double crater. Gray expects a single one here because this stage carried no comparable mass at the top — a reasonable inference, not a result.
Numbers in circulation are not measurements. That distinction is the whole job in model validation, and it does not stop being true because a number is repeated confidently by everyone.
Translate the failure modes and they get uncomfortably familiar.
| What happened to 2025-010D |
The equivalent in a lending stack |
| Disposal fuel spent on the primary mission |
Decommissioning descoped to protect a launch date |
| Compliant maneuver, then no owner |
Control signed off at go-live, never re-verified |
| Solar radiation pressure over 18 months |
Population drift between validation cycles |
| Invisible to radar at lunar distance |
No telemetry on the deprecated service |
| Tracked by volunteers with other jobs |
One engineer who happens to know it exists |
Concretely, the things worth auditing this quarter:
- Retired scorecards that still score. A superseded model version left wired into a fallback path, quietly deciding a slice of applications.
- Orphaned document stores. Buckets of customer bank statements from a pilot that ended. Nobody owns them, retention policy never applied, and they are pure liability.
- Extraction pipelines with one remaining caller. The old parser kept alive for a single lender integration, unmaintained and unmonitored.
- Credentials outliving their purpose. API keys for a data provider you stopped paying, still valid, still in a config somewhere.
- Connectors for formats that changed. A bank redesigned its statement layout; the connector now fails silently on a small percentage and nobody set an alarm.
None of these will page you. That is precisely the problem: they fail in ways that produce no signal until the failure is someone else's finding.
The fix, in orbital mechanics, is cheap — if you decide early. Put the upper stage on a trajectory that leaves the Earth-Moon system entirely and ends up orbiting the Sun. The Falcon 9 stage from the EscaPADE launch in November 2025 was disposed of this way. Chinese lunar launches have moved in the same direction.
But it only works as a design-time decision. Once the propellant is gone, no option exists at any price. Gray is also honest about the limit: the trick "only works for a small fraction of junk," namely missions already headed beyond Earth orbit. Some classes of system genuinely cannot be cleanly retired, and the time to find out which class you are in is before you build.
The question Scheiman raised is the right one to steal: "What is the best future disposal path for high-energy missions that are in the sun-Earth-moon system?" The software version is smaller and duller and just as unanswered. What is the retirement path for this model, this pipeline, this data store? Who owns it after the launch party? What signal tells us it is still running, and who receives that signal?
Systems you cannot see are systems you cannot retire. Levmetrics' Lev Document Intelligence is built so that every extraction, score and decision leaves an auditable trace — including the ones you meant to turn off. Try it at app.levmetrics.com.
A four-tonne rocket stage is a useful teacher because it is impossible to lose track of politely. It kept an orbit, it kept a catalogue number, and eighteen months after everyone declared the mission a success it arrived somewhere nobody intended at 2.43 kilometres per second. Most abandoned systems are less dramatic and more expensive. Deployment is a milestone in a system's life. It is nowhere near the end of it.
- Mike Wall, "A SpaceX rocket will crash into the moon this week. Why is it on a collision course?", Space.com, August 3, 2026.
- Chelsea Gohd, "A SpaceX rocket just crashed into the moon. Now, the race is on to get a look at the impact site", Space.com, August 5, 2026.
- Mike Wall, "SpaceX rocket's impact crater on the moon spied by Korean lunar orbiter (photos)", Space.com, August 6, 2026.
- Bill Gray, "Upper stage impacting the moon on 2026 August 5", Project Pluto.
- Adam Kovac, "A SpaceX rocket booster is on track to hit the moon at several times the speed of sound", Scientific American, May 1, 2026.
- Leonard David, "Moon Impact of Rocket Stage Observed", Inside Outer Space, August 5, 2026.
- Benjamin Fernando et al., "preprint on the predicted lunar impact of the 2025-010D upper stage", arXiv.
A four-tonne piece of a SpaceX rocket spent eighteen months on a collision course with the Moon, starting at roughly the moment everyone stopped thinking about it. On August 5 it arrived.
The interesting part is not the collision. It is that the collision was scheduled eighteen months in advance by a decision nobody remembers making.
Yes. A spent SpaceX Falcon 9 upper stage crashed into the Moon on August 5, 2026, at about 06:34 UTC — 2:34 a.m. EDT — near the crater Einstein on the western limb. South Korea's Danuri orbiter confirmed the lunar impact, imaging the site across eight passes that began roughly thirty minutes before contact. The Korea Aerospace Research Institute reported that "changes in the terrain around the impact site and traces of ejecta spread were confirmed." At Lowell Observatory in Arizona, Carl Schmidt's team measured a sodium and lithium gas plume tens of kilometres across that lasted five to ten minutes.
What follows is not a space story. It is what the crash says about every system that outlives the attention paid to it.
The object has no name, only a catalogue designation: 2025-010D. It was the fourth tracked piece of hardware from the tenth orbital launch of 2025, on January 15, which carried Firefly Aerospace's Blue Ghost lander and ispace's Resilience toward the Moon. Blue Ghost landed successfully that March. Resilience lost contact about ninety seconds before touchdown and crashed in June.
Falcon 9 upper stages are normally deorbited shortly after launch and burned up in the atmosphere. That did not happen here. Pushing two landers to their deployment point consumed nearly all of the stage's propellant, so the disposal burn was no longer available. In the words of a study led by Benjamin Fernando of Los Alamos National Laboratory, the stage "was abandoned in a moon-crossing high-Earth orbit."
It then spent more than eighteen months in a lopsided 26-day orbit, swinging between roughly 220,000 and 510,000 km from Earth — repeatedly crossing the Moon's path, missing it every time, until it didn't. Astronomer Bill Gray of Project Pluto put the impact speed at 2.43 km/s and the released energy at about 14.5 billion joules, roughly three tons of TNT.
Nobody chose to crash a rocket into the Moon. The choice that mattered was made much earlier, and it looked like a fuel budget.
Decommissioning was the line item that got consumed to make the primary mission work. There was no meeting about it, no risk register entry, no dissenting engineer. The mission was harder than the margin allowed, the margin came out of end-of-life, and the end-of-life plan quietly ceased to exist.
This is the most common way software systems become permanent. The migration runs long, so the cutover eats the quarter, so the decommissioning of the old system slips to "next sprint." The old system keeps serving a trickle of traffic. Two years later it is load-bearing and nobody on the team has read its code.
The failure mode is not negligence. It is that decommissioning is the only part of a project whose deadline nothing enforces.
SpaceX did not simply walk away. Julianna Scheiman, the company's director of NASA science and Dragon programs, said a different maneuver was performed "to ensure that the second stage itself is safe per the appropriate rules and regulations."
That maneuver satisfied the rules. It did not put the object anywhere durable. As Scheiman put it, "a mixture of solar activity and gravity forces have put it on a path towards the moon."
A control that passes review at go-live is a statement about go-live. It is not a statement about the system's state eighteen months later, and treating the two as equivalent is how model risk accumulates. Sign-off is a snapshot; systems are a trajectory.
The force that moved it was sunlight. Solar radiation pressure — photons striking the tumbling stage — is, in Gray's description, "an extremely gentle force, but over months, it can really build up." It is also not fully predictable, because a tumbling object catches and reflects light differently as it turns.
No single push mattered. Eighteen months of them decided where four-and-a-bit tonnes of metal ended up.
That is drift, exactly as it appears in a scoring pipeline: no individual week's population shift is worth an alert, and the cumulative eighteen-month displacement is worth a regulatory finding. If your validation cadence is slower than your drift accumulation, you are not monitoring the model. You are just visiting it occasionally.
Here is the detail we found most uncomfortable. This object was tracked mainly by asteroid surveys and amateur astronomers, because the systems built for space debris cannot see that far out. Radar excels in low Earth orbit — Gray notes it has tracked gloves and tool bags lost by astronauts — but at lunar distance the return signal is about 25.6 billion times fainter. Gray computes orbits for high-altitude objects specifically because they are the ones "the military doesn't track."
The surveys that did see it were not trying to. Their job is planetary defence, and as Gray puts it, "time spent observing junk is time not spent finding rocks."
Every engineering organisation has this arrangement and calls it something else. The deprecated service has no dashboard and no on-call rotation, and the only reason anyone knows it is degrading is that one engineer keeps noticing it in unrelated logs. That is not observability. That is a volunteer, and volunteers leave.
The recovery here is genuinely impressive, and it was made of data. By late February the archive held 1,053 observations, contributed by observatories in Mississippi, Utah, Beijing and England. Gray's software flagged the lunar impact in September 2025 — eleven months ahead.
The forecast then tightened as observations accumulated. A June revision moved the predicted impact nine minutes earlier and 170 km to the northwest. A July revision moved it by a couple of seconds and about ten kilometres. A final update landed on August 1.
That precision had a payoff: Danuri began imaging roughly thirty minutes before the collision and made eight passes over the site, capturing genuine before-and-after coverage. You only point an instrument at the right patch of ground at the right minute if someone gave you a forecast. And you only get a forecast for a system somebody is still measuring.
Worth noting, in the spirit of not overclaiming: as of this writing there is no published measurement of the new crater. Gray predicted a single crater around 17 m across. Much of the news coverage cites up to 27 m. Both are pre-impact calculations, and they do not agree with each other.
Whether it is one crater or two is also unsettled. The 2022 Chang'e 5-T1 upper stage surprised everyone by making a double crater. Gray expects a single one here because this stage carried no comparable mass at the top — a reasonable inference, not a result.
Numbers in circulation are not measurements. That distinction is the whole job in model validation, and it does not stop being true because a number is repeated confidently by everyone.
Translate the failure modes and they get uncomfortably familiar.
| What happened to 2025-010D |
The equivalent in a lending stack |
| Disposal fuel spent on the primary mission |
Decommissioning descoped to protect a launch date |
| Compliant maneuver, then no owner |
Control signed off at go-live, never re-verified |
| Solar radiation pressure over 18 months |
Population drift between validation cycles |
| Invisible to radar at lunar distance |
No telemetry on the deprecated service |
| Tracked by volunteers with other jobs |
One engineer who happens to know it exists |
Concretely, the things worth auditing this quarter:
- Retired scorecards that still score. A superseded model version left wired into a fallback path, quietly deciding a slice of applications.
- Orphaned document stores. Buckets of customer bank statements from a pilot that ended. Nobody owns them, retention policy never applied, and they are pure liability.
- Extraction pipelines with one remaining caller. The old parser kept alive for a single lender integration, unmaintained and unmonitored.
- Credentials outliving their purpose. API keys for a data provider you stopped paying, still valid, still in a config somewhere.
- Connectors for formats that changed. A bank redesigned its statement layout; the connector now fails silently on a small percentage and nobody set an alarm.
None of these will page you. That is precisely the problem: they fail in ways that produce no signal until the failure is someone else's finding.
The fix, in orbital mechanics, is cheap — if you decide early. Put the upper stage on a trajectory that leaves the Earth-Moon system entirely and ends up orbiting the Sun. The Falcon 9 stage from the EscaPADE launch in November 2025 was disposed of this way. Chinese lunar launches have moved in the same direction.
But it only works as a design-time decision. Once the propellant is gone, no option exists at any price. Gray is also honest about the limit: the trick "only works for a small fraction of junk," namely missions already headed beyond Earth orbit. Some classes of system genuinely cannot be cleanly retired, and the time to find out which class you are in is before you build.
The question Scheiman raised is the right one to steal: "What is the best future disposal path for high-energy missions that are in the sun-Earth-moon system?" The software version is smaller and duller and just as unanswered. What is the retirement path for this model, this pipeline, this data store? Who owns it after the launch party? What signal tells us it is still running, and who receives that signal?
Systems you cannot see are systems you cannot retire. Levmetrics' Lev Document Intelligence is built so that every extraction, score and decision leaves an auditable trace — including the ones you meant to turn off. Try it at app.levmetrics.com.
A four-tonne rocket stage is a useful teacher because it is impossible to lose track of politely. It kept an orbit, it kept a catalogue number, and eighteen months after everyone declared the mission a success it arrived somewhere nobody intended at 2.43 kilometres per second. Most abandoned systems are less dramatic and more expensive. Deployment is a milestone in a system's life. It is nowhere near the end of it.
- Mike Wall, "A SpaceX rocket will crash into the moon this week. Why is it on a collision course?", Space.com, August 3, 2026.
- Chelsea Gohd, "A SpaceX rocket just crashed into the moon. Now, the race is on to get a look at the impact site", Space.com, August 5, 2026.
- Mike Wall, "SpaceX rocket's impact crater on the moon spied by Korean lunar orbiter (photos)", Space.com, August 6, 2026.
- Bill Gray, "Upper stage impacting the moon on 2026 August 5", Project Pluto.
- Adam Kovac, "A SpaceX rocket booster is on track to hit the moon at several times the speed of sound", Scientific American, May 1, 2026.
- Leonard David, "Moon Impact of Rocket Stage Observed", Inside Outer Space, August 5, 2026.
- Benjamin Fernando et al., "preprint on the predicted lunar impact of the 2025-010D upper stage", arXiv.
A four-tonne piece of a SpaceX rocket spent eighteen months on a collision course with the Moon, starting at roughly the moment everyone stopped thinking about it. On August 5 it arrived.
The interesting part is not the collision. It is that the collision was scheduled eighteen months in advance by a decision nobody remembers making.
Yes. A spent SpaceX Falcon 9 upper stage crashed into the Moon on August 5, 2026, at about 06:34 UTC — 2:34 a.m. EDT — near the crater Einstein on the western limb. South Korea's Danuri orbiter confirmed the lunar impact, imaging the site across eight passes that began roughly thirty minutes before contact. The Korea Aerospace Research Institute reported that "changes in the terrain around the impact site and traces of ejecta spread were confirmed." At Lowell Observatory in Arizona, Carl Schmidt's team measured a sodium and lithium gas plume tens of kilometres across that lasted five to ten minutes.
What follows is not a space story. It is what the crash says about every system that outlives the attention paid to it.
The object has no name, only a catalogue designation: 2025-010D. It was the fourth tracked piece of hardware from the tenth orbital launch of 2025, on January 15, which carried Firefly Aerospace's Blue Ghost lander and ispace's Resilience toward the Moon. Blue Ghost landed successfully that March. Resilience lost contact about ninety seconds before touchdown and crashed in June.
Falcon 9 upper stages are normally deorbited shortly after launch and burned up in the atmosphere. That did not happen here. Pushing two landers to their deployment point consumed nearly all of the stage's propellant, so the disposal burn was no longer available. In the words of a study led by Benjamin Fernando of Los Alamos National Laboratory, the stage "was abandoned in a moon-crossing high-Earth orbit."
It then spent more than eighteen months in a lopsided 26-day orbit, swinging between roughly 220,000 and 510,000 km from Earth — repeatedly crossing the Moon's path, missing it every time, until it didn't. Astronomer Bill Gray of Project Pluto put the impact speed at 2.43 km/s and the released energy at about 14.5 billion joules, roughly three tons of TNT.
Nobody chose to crash a rocket into the Moon. The choice that mattered was made much earlier, and it looked like a fuel budget.
Decommissioning was the line item that got consumed to make the primary mission work. There was no meeting about it, no risk register entry, no dissenting engineer. The mission was harder than the margin allowed, the margin came out of end-of-life, and the end-of-life plan quietly ceased to exist.
This is the most common way software systems become permanent. The migration runs long, so the cutover eats the quarter, so the decommissioning of the old system slips to "next sprint." The old system keeps serving a trickle of traffic. Two years later it is load-bearing and nobody on the team has read its code.
The failure mode is not negligence. It is that decommissioning is the only part of a project whose deadline nothing enforces.
SpaceX did not simply walk away. Julianna Scheiman, the company's director of NASA science and Dragon programs, said a different maneuver was performed "to ensure that the second stage itself is safe per the appropriate rules and regulations."
That maneuver satisfied the rules. It did not put the object anywhere durable. As Scheiman put it, "a mixture of solar activity and gravity forces have put it on a path towards the moon."
A control that passes review at go-live is a statement about go-live. It is not a statement about the system's state eighteen months later, and treating the two as equivalent is how model risk accumulates. Sign-off is a snapshot; systems are a trajectory.
The force that moved it was sunlight. Solar radiation pressure — photons striking the tumbling stage — is, in Gray's description, "an extremely gentle force, but over months, it can really build up." It is also not fully predictable, because a tumbling object catches and reflects light differently as it turns.
No single push mattered. Eighteen months of them decided where four-and-a-bit tonnes of metal ended up.
That is drift, exactly as it appears in a scoring pipeline: no individual week's population shift is worth an alert, and the cumulative eighteen-month displacement is worth a regulatory finding. If your validation cadence is slower than your drift accumulation, you are not monitoring the model. You are just visiting it occasionally.
Here is the detail we found most uncomfortable. This object was tracked mainly by asteroid surveys and amateur astronomers, because the systems built for space debris cannot see that far out. Radar excels in low Earth orbit — Gray notes it has tracked gloves and tool bags lost by astronauts — but at lunar distance the return signal is about 25.6 billion times fainter. Gray computes orbits for high-altitude objects specifically because they are the ones "the military doesn't track."
The surveys that did see it were not trying to. Their job is planetary defence, and as Gray puts it, "time spent observing junk is time not spent finding rocks."
Every engineering organisation has this arrangement and calls it something else. The deprecated service has no dashboard and no on-call rotation, and the only reason anyone knows it is degrading is that one engineer keeps noticing it in unrelated logs. That is not observability. That is a volunteer, and volunteers leave.
The recovery here is genuinely impressive, and it was made of data. By late February the archive held 1,053 observations, contributed by observatories in Mississippi, Utah, Beijing and England. Gray's software flagged the lunar impact in September 2025 — eleven months ahead.
The forecast then tightened as observations accumulated. A June revision moved the predicted impact nine minutes earlier and 170 km to the northwest. A July revision moved it by a couple of seconds and about ten kilometres. A final update landed on August 1.
That precision had a payoff: Danuri began imaging roughly thirty minutes before the collision and made eight passes over the site, capturing genuine before-and-after coverage. You only point an instrument at the right patch of ground at the right minute if someone gave you a forecast. And you only get a forecast for a system somebody is still measuring.
Worth noting, in the spirit of not overclaiming: as of this writing there is no published measurement of the new crater. Gray predicted a single crater around 17 m across. Much of the news coverage cites up to 27 m. Both are pre-impact calculations, and they do not agree with each other.
Whether it is one crater or two is also unsettled. The 2022 Chang'e 5-T1 upper stage surprised everyone by making a double crater. Gray expects a single one here because this stage carried no comparable mass at the top — a reasonable inference, not a result.
Numbers in circulation are not measurements. That distinction is the whole job in model validation, and it does not stop being true because a number is repeated confidently by everyone.
Translate the failure modes and they get uncomfortably familiar.
| What happened to 2025-010D |
The equivalent in a lending stack |
| Disposal fuel spent on the primary mission |
Decommissioning descoped to protect a launch date |
| Compliant maneuver, then no owner |
Control signed off at go-live, never re-verified |
| Solar radiation pressure over 18 months |
Population drift between validation cycles |
| Invisible to radar at lunar distance |
No telemetry on the deprecated service |
| Tracked by volunteers with other jobs |
One engineer who happens to know it exists |
Concretely, the things worth auditing this quarter:
- Retired scorecards that still score. A superseded model version left wired into a fallback path, quietly deciding a slice of applications.
- Orphaned document stores. Buckets of customer bank statements from a pilot that ended. Nobody owns them, retention policy never applied, and they are pure liability.
- Extraction pipelines with one remaining caller. The old parser kept alive for a single lender integration, unmaintained and unmonitored.
- Credentials outliving their purpose. API keys for a data provider you stopped paying, still valid, still in a config somewhere.
- Connectors for formats that changed. A bank redesigned its statement layout; the connector now fails silently on a small percentage and nobody set an alarm.
None of these will page you. That is precisely the problem: they fail in ways that produce no signal until the failure is someone else's finding.
The fix, in orbital mechanics, is cheap — if you decide early. Put the upper stage on a trajectory that leaves the Earth-Moon system entirely and ends up orbiting the Sun. The Falcon 9 stage from the EscaPADE launch in November 2025 was disposed of this way. Chinese lunar launches have moved in the same direction.
But it only works as a design-time decision. Once the propellant is gone, no option exists at any price. Gray is also honest about the limit: the trick "only works for a small fraction of junk," namely missions already headed beyond Earth orbit. Some classes of system genuinely cannot be cleanly retired, and the time to find out which class you are in is before you build.
The question Scheiman raised is the right one to steal: "What is the best future disposal path for high-energy missions that are in the sun-Earth-moon system?" The software version is smaller and duller and just as unanswered. What is the retirement path for this model, this pipeline, this data store? Who owns it after the launch party? What signal tells us it is still running, and who receives that signal?
Systems you cannot see are systems you cannot retire. Levmetrics' Lev Document Intelligence is built so that every extraction, score and decision leaves an auditable trace — including the ones you meant to turn off. Try it at app.levmetrics.com.
A four-tonne rocket stage is a useful teacher because it is impossible to lose track of politely. It kept an orbit, it kept a catalogue number, and eighteen months after everyone declared the mission a success it arrived somewhere nobody intended at 2.43 kilometres per second. Most abandoned systems are less dramatic and more expensive. Deployment is a milestone in a system's life. It is nowhere near the end of it.
- Mike Wall, "A SpaceX rocket will crash into the moon this week. Why is it on a collision course?", Space.com, August 3, 2026.
- Chelsea Gohd, "A SpaceX rocket just crashed into the moon. Now, the race is on to get a look at the impact site", Space.com, August 5, 2026.
- Mike Wall, "SpaceX rocket's impact crater on the moon spied by Korean lunar orbiter (photos)", Space.com, August 6, 2026.
- Bill Gray, "Upper stage impacting the moon on 2026 August 5", Project Pluto.
- Adam Kovac, "A SpaceX rocket booster is on track to hit the moon at several times the speed of sound", Scientific American, May 1, 2026.
- Leonard David, "Moon Impact of Rocket Stage Observed", Inside Outer Space, August 5, 2026.
- Benjamin Fernando et al., "preprint on the predicted lunar impact of the 2025-010D upper stage", arXiv.
Cycle de vie
Notation de crédit
Fraude