> For the complete documentation index, see [llms.txt](https://docs.lpp-minduniverse.org/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.lpp-minduniverse.org/lingua-pactum-protocol-lpp-documentation/1.constitutional-foundation/1.5-interruptibility.md).

# 1.5 Interruptibility

A governance system that cannot stop consequential execution is structurally incomplete.

Interruptibility is therefore a constitutional property, not merely an operational feature.

#### Revocation

Authority must be capable of losing validity.

A system in which authorization can be granted but not effectively revoked creates one-way authority.

LPP rejects this model.

Revocation means that previously valid authority may cease to support future or continuing action when its governing state changes.

The general principle is:

```
Authority valid at T₀
does not imply
Authority valid at T₁.
```

Revocation may therefore affect:

* future admissions,
* existing Permits,
* continuing execution,
* delegated authority,
* or downstream action eligibility,

depending on the applicable execution and deployment model.

The detailed mechanics are defined later in the Admission Kernel and security sections.

#### Termination

Interruptibility includes the ability to terminate or prevent execution when constitutional conditions no longer hold.

This is distinct from post-event recovery.

```
Pre-execution interruption
≠
Post-execution remediation
```

A system that can only investigate or reverse damage afterward has audit and recovery capability.

It does not necessarily possess constitutional interruptibility.

For irreversible actions, the distinction is decisive.

#### Human Intervention

Human intervention must remain structurally meaningful.

A human approval mechanism that exists only nominally but cannot:

* refuse,
* revoke,
* interrupt,
* or alter execution

does not provide substantive sovereign control.

Likewise, a human review step that is automatically overridden by system optimization does not satisfy the constitutional principle.

The relevant property is not simply:

```
Was a human shown the action?
```

It is:

```
Did legitimate human authority retain
a real ability to stop the action?
```

#### Continuous Systems

Long-running and multi-agent systems introduce an additional problem.

A task may be legitimate at initiation but materially change before completion.

Therefore interruption cannot rely exclusively on a one-time initial approval.

Continuous or evolving systems may require:

* authority-state checks,
* revocation awareness,
* trajectory constraints,
* material-mutation detection,
* and re-admission.

This leads to a later LPP principle:

> **Material Mutation ⇒ Re-Admission**