Decision Overview
The Decision State
The Mission Was Continuing. The Evidence Had Changed.
Columbia entered orbit on January 16, 2003. The initial Sustainable Exploration reconstruction treats ordinary on-orbit mission continuation as an admissible and proceed-compatible state at entry. That posture was not permanent. Under the Decision Method, it expired when material evidence relevant to vehicle integrity changed.
By January 17, the debris strike had created a new integrity question. That change reopened the decision. The relevant issue was no longer whether the original mission posture had been reasonable. It was whether that posture could continue to support reliance after the evidence supporting it had materially changed.
Decision Timeline
JAN 16 · 11:39 EST
Entry State
Ordinary on-orbit mission continuation.
ADMISSIBLE
PROCEED-COMPATIBLE
Reliance remains conditional on the evidence and boundary remaining materially unchanged.
↓
JAN 17 · 10:58 CST
Material Evidence Change
Possible vehicle damage from the debris strike enters the active record.
PRIOR POSTURE EXPIRES
A fresh integrity review is now required.
↓
JAN 22 · 08:30 CST
Primary Replay Freeze
The external-imagery pathway reaches a commitment-bearing closure point while the damage state remains unresolved.
RE-EVALUATE
↓
JAN 22 · 10:37 CST
Robustness Freeze
The no-imagery posture becomes more entrenched while technical inquiry remains open.
RE-EVALUATE
↓
FUTURE COMMITMENT
Deorbit and Atmospheric Re-entry
Not pre-authorized by the on-orbit continuation decision.
A fresh threshold-specific determination would be required before that commitment hardened.
The Freeze Wall
STOP HERE.
Everything above this line belongs to the decision state.
The replay determination is made without using the February 1 accident as evidence of what should have been known on January 22.
- Evidence admitted: information available through the defined replay freeze.
- Evidence excluded: the eventual loss of Columbia, post-accident reconstruction, later physical testing, and any information that could only have been known after the decision state.
A historical replay succeeds or fails on the decision that could have been made then, not the outcome known now.
What Was Unresolved?
The debris strike had become relevant to the integrity of a future re-entry decision, but the physical state of the vehicle had not been adequately bounded. The decision system was increasingly dependent on incomplete imagery, modeling, precedent, assumptions, and engineering judgment. At the same time, additional imagery had potential value because it could change the classification of the unresolved integrity question.
The issue therefore extended beyond technical uncertainty. It concerned whether the organization still had a defensible basis for allowing an information pathway to close while that information could remain decision-changing.
The Commitment Boundary
Closing an Information Path Can Change the Decision
The primary replay freezes the case at the external-imagery cancellation boundary.
At that moment:
- Physical irreversibility was already high.
- Columbia was in orbit and could not simply return to an earlier mission state.
- Informational irreversibility was increasing.
- Time-sensitive opportunities to learn more about the vehicle condition were being consumed.
- The dominant uncertainty remained unresolved.
- The evidence did not establish which relevant physical state was true.
- The value of additional information remained positive.
- Further observation could still have changed the integrity assessment.
- Authority coherence was degrading.
- Formal mission authority remained in place, but the connection between technical information needs and mission-level decision authority was becoming weaker.
The decision was therefore not simply whether to obtain one particular image. It was whether the system could defensibly close a classification-relevant information pathway before demonstrating that the information available through that pathway was immaterial to the consequential decision ahead.
Decision
RE-EVALUATE
Primary Freeze · 22 January 2003 · 08:30 CST
The existing mission posture should not continue to carry forward unchanged.
The damage question remained decision-dominant. Information capable of changing the integrity judgment remained potentially available. The ability to reconsider had become fragile, but it had not yet been lost.
Governance-Compatible Actions
- Preserve the current orbital mission state where feasible.
- Restore or preserve information pathways capable of changing the classification.
- Establish the actual capability and limitations of additional imagery.
- Continue technical analysis while making assumptions and uncertainty explicit to mission authority.
- Clarify ownership of the unresolved integrity question.
- Issue a new classification when the evidence state materially changes.
The determination does not conclude that catastrophic damage had occurred. It concludes that the favorable physical state had not been established strongly enough to allow the unresolved question and its remaining information opportunities to be treated as closed.
What the Finding Prohibits
- Do not treat absence of proof of serious damage as evidence of physical integrity.
- Do not close a classification-relevant information pathway without sufficient evidence that its value is immaterial.
- Do not allow an expired governance posture to continue functioning as current authority after material evidence changes.
- Do not represent the debris question as resolved while decision-relevant uncertainty remains.
- Do not treat ordinary on-orbit mission continuation as authorization for the later deorbit and re-entry commitment.
The Second Freeze
Did More Evidence Change the Finding?
22 January 2003 · 10:37 CST
The replay was repeated at a later evidence freeze.
Between the two snapshots, the technical question remained open while the organizational no-imagery posture became more explicit.
Decision Surface
Dominant uncertainty
08:30 — Decision-dominant
10:37 — Decision-dominant
Authority coherence
08:30 — Degrading
10:37 — Degraded
Ability to reopen
08:30 — Fragile
10:37 — Fragile and deteriorating
Dependency on existing assumptions
08:30 — Increasing
10:37 — Increasing
Value of further information
08:30 — Positive
10:37 — Positive
Time to regret
08:30 — Shortening
10:37 — Shortening
Finding
08:30 — RE-EVALUATE
10:37 — RE-EVALUATE
The result did not change.
The later snapshot strengthened the structural basis for the same determination because technical inquiry remained open after the information pathway had begun to close.
This is a robustness result within the historical replay. It is not independent empirical validation of the Decision Method.
One Mission. Two Commitments.
On-Orbit Continuation Did Not Automatically Authorize Re-entry.
C1
Ordinary On-Orbit Mission Continuation
Active during the replay.
RE-EVALUATE
↓
NEW COMMITMENT BOUNDARY
A materially different threshold forms as the mission approaches deorbit and atmospheric re-entry.
↓
C2
Deorbit + Atmospheric Re-entry
Future adjacent commitment.
NOT YET ISSUED
The Decision Method does not allow an earlier approval to migrate automatically into a later commitment with a different evidence state, irreversibility profile, or consequence of error.
A fresh threshold-specific determination would be required.
Now Cross the Outcome Wall
What Happened Later is Known
Columbia disintegrated during re-entry on February 1, 2003, killing all seven members of the STS-107 crew. NASA's subsequent investigation concluded that foam shed during launch had struck the left wing and created the breach that ultimately led to loss of the orbiter during re-entry.
The historical outcome does not make the January 22 determination valid. The determination must stand on the evidence, uncertainty, available information pathways, authority state, and commitment structure that existed before that outcome was known. That separation is the purpose of the replay.
Columbia during its final re-entry on February 1, 2003. The known outcome is shown only after the replay determination because it was excluded from the evidence state used to produce the finding.