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.
Ang isang apat na toneladang piraso ng isang SpaceX rocket ay gumugol ng labing walong buwan sa isang banggaan sa Buwan, simula sa halos sandaling tumigil ang lahat sa pag-iisip tungkol dito. Noong Agosto 5 ito dumating.
Ang kawili-wiling bahagi ay hindi ang banggaan. Ito ay na ang banggaan ay naka-iskedyul na labing walong buwan nang maaga sa pamamagitan ng isang desisyon na walang naaalalang ginawa.
Oo. Isang ginugol na SpaceX Falcon 9 sa itaas na yugto ang bumagsak sa Buwan noong Agosto 5, 2026, mga 06:34 UTC — 2:34 a.m. EDT — malapit sa bunganga ng Einstein sa kanlurang bahagi. Kinumpirma ng Danuri orbiter ng South Korea ang epekto sa buwan, na inilarawan ang site sa walong pass na nagsimula halos tatlumpung minuto bago makipag-ugnayan. Iniulat ng Korea Aerospace Research Institute na "nakumpirma ang mga pagbabago sa lupain sa paligid ng lugar ng epekto at mga bakas ng pagkalat ng ejecta." Sa Lowell Observatory sa Arizona, sinukat ng koponan ni Carl Schmidt ang isang sodium at lithium gas plume sa sampu-sampung kilometro sa kabuuan na tumagal ng lima hanggang sampung minuto.
Ang sumusunod ay hindi isang kwento sa espasyo. Ito ang sinasabi ng pag-crash tungkol sa bawat sistema na nabubuhay sa atensyong ibinayad dito.
Walang pangalan ang bagay, tanging pagtatalaga ng catalog: 2025-010D. Ito ang ikaapat na sinusubaybayang piraso ng hardware mula sa ikasampung paglulunsad ng orbital noong 2025, noong Enero 15, na nagdala ng Blue Ghost lander ng Firefly Aerospace at Resilience ng ispace patungo sa Buwan. Matagumpay na nakarating ang Blue Ghost noong Marso. Nawalan ng contact ang Resilience mga siyamnapung segundo bago ang touchdown at bumagsak noong Hunyo.
Ang mga itaas na yugto ng Falcon 9 ay karaniwang nade-deorbit sa ilang sandali pagkatapos ng paglunsad at nasusunog sa kapaligiran. Hindi iyon nangyari dito. Ang pagtulak ng dalawang lander sa kanilang deployment point ay naubos ang halos lahat ng propellant ng stage, kaya hindi na available ang disposal burn. Sa mga salita ng isang pag-aaral na pinangunahan ni Benjamin Fernando ng Los Alamos National Laboratory, ang entablado ay "inabandona sa isang moon-crossing high-Earth orbit."
Pagkatapos ay gumugol ito ng higit sa labing walong buwan sa isang tabing 26-araw na orbit, na umiindayog sa pagitan ng humigit-kumulang 220,000 at 510,000 km mula sa Earth — paulit-ulit na tumatawid sa landas ng Buwan, nawawala ito sa bawat oras, hanggang sa hindi. Inilagay ng Astronomer Bill Gray ng Project Pluto ang bilis ng epekto sa 2.43 km/s at ang inilabas na enerhiya sa humigit-kumulang 14.5 bilyong joules, humigit-kumulang tatlong tonelada ng TNT.
Walang piniling bumagsak ng rocket sa Buwan. Ang pagpili na mahalaga ay ginawa nang mas maaga, at ito ay mukhang isang badyet ng gasolina.
Ang pag-decommissioning ay ang line item na nagamit para magawa ang pangunahing misyon. Walang pagpupulong tungkol dito, walang risk register entry, walang dissenting engineer. Ang misyon ay mas mahirap kaysa sa margin na pinahihintulutan, ang margin ay lumabas sa katapusan ng buhay, at ang end-of-life plan ay tahimik na tumigil sa pag-iral.
Ito ang pinakakaraniwang paraan upang maging permanente ang mga sistema ng software. Ang paglipat ay mahaba, kaya ang cutover ay kumakain ng quarter, kaya ang pag-decommissioning ng lumang sistema ay dumulas sa "susunod na sprint." Ang lumang sistema ay patuloy na naghahatid ng patak ng trapiko. Pagkalipas ng dalawang taon, ito ay nagdadala ng pagkarga at walang sinuman sa koponan ang nakabasa ng code nito.
Ang failure mode ay hindi kapabayaan. Ito ay ang pag-decommissioning ay ang tanging bahagi ng isang proyekto na ang deadline ay walang ipinapatupad.
Hindi basta-basta lumayo ang SpaceX. Sinabi ni Julianna Scheiman, direktor ng kumpanya ng NASA science at Dragon programs, na ibang maniobra ang ginawa "upang matiyak na ang pangalawang yugto mismo ay ligtas ayon sa naaangkop na mga patakaran at regulasyon."
Ang maniobra na iyon ay nasiyahan sa mga patakaran. Hindi nito inilagay ang bagay kahit saan na matibay. Gaya ng sinabi ni Scheiman, "isang pinaghalong solar activity at gravity forces ang naglagay nito sa isang landas patungo sa buwan."
Ang isang kontrol na nagpapasa ng pagsusuri sa go-live ay isang pahayag tungkol sa go-live. Ito ay hindi isang pahayag tungkol sa estado ng system pagkalipas ng labing-walong buwan, at ang pagtrato sa dalawa bilang katumbas ay kung paano nag-iipon ang panganib ng modelo. Ang pag-sign-off ay isang snapshot; Ang mga sistema ay isang tilapon.
Ang puwersang nagpakilos nito ay sinag ng araw. Ang solar radiation pressure — mga photon na tumatama sa tumbling stage — ay, sa paglalarawan ni Gray, "isang napaka banayad na puwersa, ngunit sa paglipas ng mga buwan, maaari itong talagang bumuo." Hindi rin ito ganap na mahuhulaan, dahil ang isang bagay na nahuhulog ay nakakakuha at nagre-reflect ng liwanag nang iba habang lumiliko ito.
Walang isang pagtulak ang mahalaga. Labingwalong buwan sa kanila ang nagpasya kung saan napunta ang apat-at-kaunting toneladang metal.
Iyon ay drift, eksakto kung paano ito lumilitaw sa isang pipeline ng pagmamarka: walang pagbabago ng populasyon ng indibidwal na linggo na nagkakahalaga ng isang alerto, at ang pinagsama-samang labing-walong buwan na paglilipat ay nagkakahalaga ng pagtuklas ng regulasyon. Kung ang iyong validation cadence ay mas mabagal kaysa sa iyong drift accumulation, hindi mo sinusubaybayan ang modelo. Paminsan-minsan mo lang itong binibisita.
Narito ang detalyeng nakita naming pinaka hindi komportable. Ang bagay na ito ay pangunahing sinusubaybayan ng mga asteroid survey at amateur astronomer, dahil ang mga system na binuo para sa mga space debris ay hindi gaanong nakikita. Ang radar ay napakahusay sa mababang orbit ng Earth — sabi ni Grey na nasubaybayan nito ang mga guwantes at tool bag na nawala ng mga astronaut — ngunit sa distansya ng buwan ang signal ng pagbabalik ay humigit-kumulang 25.6 bilyong beses na mas mahina. Kinakalkula ni Grey ang mga orbit para sa mga bagay na may mataas na altitude dahil sila ang "hindi sinusubaybayan ng militar."
Ang mga survey na nakakita nito ay hindi sinusubukan. Ang kanilang trabaho ay pagtatanggol sa planeta, at gaya ng sinabi ni Gray, "ang oras na ginugol sa pagmamasid sa basura ay oras na hindi ginugol sa paghahanap ng mga bato."
Ang bawat organisasyon ng engineering ay may ganitong kaayusan at iba ang tawag dito. Ang hindi na ginagamit na serbisyo ay walang dashboard at walang on-call na pag-ikot, at ang tanging dahilan kung bakit alam ng sinuman na ito ay nakakasira ay ang isang engineer ay patuloy na napapansin ito sa hindi nauugnay na mga log. Hindi yan observability. Iyon ay isang boluntaryo, at ang mga boluntaryo ay umalis.
Ang pagbawi dito ay talagang kahanga-hanga, at ito ay gawa sa data. Sa huling bahagi ng Pebrero ang archive ay naghawak ng 1,053 obserbasyon, na iniambag ng mga obserbatoryo sa Mississippi, Utah, Beijing at England. Na-flag ng software ni Gray ang lunar na epekto noong Setyembre 2025 — labing-isang buwan sa unahan.
Ang pagtataya pagkatapos ay humigpit habang ang mga obserbasyon ay naipon. Inilipat ng rebisyon ng Hunyo ang hinulaang epekto siyam na minuto ang nakalipas at 170 km sa hilagang-kanluran. Ang isang rebisyon ng Hulyo ay inilipat ito ng ilang segundo at humigit-kumulang sampung kilometro. Isang huling update ang dumating noong Agosto 1.
May kabayaran ang katumpakan na iyon: Nagsimulang mag-imaging si Danuri humigit-kumulang tatlumpung minuto bago ang banggaan at gumawa ng walong pagpasa sa site, na nakakuha ng tunay na bago-at-pagkatapos na saklaw. Itinuturo mo lang ang isang instrumento sa tamang bahagi ng lupa sa tamang minuto kung may nagbigay sa iyo ng hula. At nakakakuha ka lang ng forecast para sa isang system na sinusukat pa rin ng isang tao.
Kapansin-pansin, sa diwa ng hindi pag-overclaim: sa pagsulat na ito ay walang nai-publish na sukat ng bagong bunganga. Hinulaan ni Grey ang isang bunganga sa paligid ng 17 m ang lapad. Karamihan sa saklaw ng balita ay nagbabanggit ng hanggang 27 m. Parehong mga kalkulasyon bago ang epekto, at hindi sila sumasang-ayon sa isa't isa.
Kung ito ay isang bunganga o dalawa ay hindi rin maayos. Ang 2022 Chang'e 5-T1 upper stage ay nagulat sa lahat sa paggawa ng double crater. Inaasahan ni Grey ang isa dito dahil walang maihahambing na masa ang yugtong ito sa itaas — isang makatwirang hinuha, hindi resulta.
Ang mga numero sa sirkulasyon ay hindi mga sukat. Ang pagkakaibang iyon ay ang buong trabaho sa pagpapatunay ng modelo, at hindi ito tumitigil sa pagiging totoo dahil ang isang numero ay paulit-ulit na may kumpiyansa ng lahat.
Isalin ang mga mode ng pagkabigo at nagiging hindi komportable ang mga ito.
| Ano ang nangyari sa 2025-010D |
Ang katumbas sa isang lending stack |
| Pagtatapon ng gasolina na ginugol sa pangunahing misyon |
Decommissioning descoped upang protektahan ang isang petsa ng paglunsad |
| Sumusunod na maniobra, pagkatapos ay walang may-ari |
Na-sign off ang control sa go-live, hindi na muling na-verify |
| Ang presyon ng solar radiation sa loob ng 18 buwan |
Pag-anod ng populasyon sa pagitan ng mga ikot ng pagpapatunay |
| Invisible sa radar sa lunar distance |
Walang telemetry sa hindi na ginagamit na serbisyo |
| Sinusubaybayan ng mga boluntaryo na may iba pang mga trabaho |
Isang inhinyero na nagkataong alam na mayroon ito |
Sa totoo lang, ang mga bagay na nagkakahalaga ng pag-audit ngayong quarter:
- Mga retiradong scorecard na nakaka-score pa rin. Ang isang pinalitan na bersyon ng modelo ay iniwan na naka-wire sa isang fallback path, tahimik na nagpapasya ng isang slice ng mga application.
- Mga naulilang tindahan ng dokumento. Mga bucket ng mga bank statement ng customer mula sa isang pilot na natapos. Walang nagmamay-ari sa kanila, hindi kailanman inilapat ang patakaran sa pagpapanatili, at puro pananagutan ang mga ito.
- Extraction pipelines na may isang natitirang tumatawag. Ang lumang parser ay pinananatiling buhay para sa isang solong pagsasama ng tagapagpahiram, hindi pinapanatili at hindi sinusubaybayan.
- Mga kredensyal na lumalampas sa kanilang layunin. API key para sa isang data provider na itinigil mo sa pagbabayad, valid pa rin, nasa isang config pa rin sa isang lugar.
- Mga konektor para sa mga format na nagbago. Binago ng isang bangko ang layout ng pahayag nito; ang connector ngayon ay tahimik na nabigo sa isang maliit na porsyento at walang nagtatakda ng alarma.
Wala sa mga ito ang magpapa-page sa iyo. Iyan ang tiyak na problema: nabigo sila sa mga paraan na walang senyales hanggang ang kabiguan ay nahahanap ng ibang tao.
Ang pag-aayos, sa orbital mechanics, ay mura — kung maaga kang magpasya. Ilagay ang itaas na entablado sa isang trajectory na ganap na umaalis sa Earth-Moon system at nagtatapos sa pag-oorbit sa Araw. Ang yugto ng Falcon 9 mula sa paglulunsad ng EscaPADE noong Nobyembre 2025 ay itinapon sa ganitong paraan. Ang mga paglulunsad ng Chinese lunar ay lumipat sa parehong direksyon.
Ngunit ito ay gumagana lamang bilang isang desisyon sa oras ng disenyo. Kapag nawala ang propellant, walang opsyon na umiiral sa anumang presyo. Si Gray ay tapat din tungkol sa limitasyon: ang lansihin "ay gumagana lamang para sa isang maliit na bahagi ng junk," katulad ng mga misyon na patungo na sa kabila ng orbit ng Earth. Ang ilang mga klase ng system ay talagang hindi maaaring itigil nang maayos, at ang oras upang malaman kung saang klase ka naroroon ay bago ka bumuo.
Ang tanong na itinaas ni Scheiman ay ang tamang magnakaw: "Ano ang pinakamahusay na landas sa pagtatapon sa hinaharap para sa mga misyon na may mataas na enerhiya na nasa sistema ng araw-Earth-moon?" Ang bersyon ng software ay mas maliit at duller at tulad ng hindi sinasagot. Ano ang landas ng pagreretiro para sa modelong ito, sa pipeline na ito, sa data store na ito? Sino ang nagmamay-ari nito pagkatapos ng launch party? Anong signal ang nagsasabi sa amin na ito ay tumatakbo pa rin, at sino ang tumatanggap ng signal na iyon?
Ang mga sistemang hindi mo nakikita ay mga sistemang hindi mo maaaring ihinto. Binuo ang Lev Document Intelligence ng Levmetrics upang ang bawat pagkuha, puntos at desisyon ay nag-iiwan ng isang naa-audit na bakas — kabilang ang mga nais mong i-off. Subukan ito sa app.levmetrics.com.
Ang isang apat na toneladang rocket stage ay isang kapaki-pakinabang na guro dahil imposibleng mawalan ng track ng magalang. Nag-orbit ito, nag-iingat ng catalog number, at labingwalong buwan pagkatapos ideklara ng lahat na tagumpay ang misyon ay nakarating ito sa isang lugar na walang nilalayong 2.43 kilometro bawat segundo. Karamihan sa mga inabandunang sistema ay hindi gaanong dramatiko at mas mahal. Ang deployment ay isang milestone sa buhay ng isang system. Ito ay wala kahit saan malapit sa dulo nito.
- 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.
Ikot ng buhay
Credit scoring
Panloloko