Minimum intervention#
SLOP8374 · Week 11
A conservator stabilises a painting.
They do not repaint it.
The conservation principle#
Change only what is necessary to make it work.
Document every change, in order, each with a justification.
Stop as soon as it runs.
What “minimum” means in practice#
- Updating a dependency, not rewriting the code that calls it
- A shim, not a migration
- Restoring compatibility, not adding features
- Leaving the formatting alone, however much it hurts
Three levels of intervention#
From least to most intrusive:
| Level | Method | Original code |
|---|---|---|
| Shim | A compatibility layer around it | Untouched |
| Migration | Update the calls to a living API | Changed mechanically |
| Rewrite | Replace the implementation | Gone |
The conservator’s preference is always the least intrusive option that achieves stability.
The restoration log#
A complete, ordered list of every change and why you made it.
The log is as important as the code. Without it, your restoration is indistinguishable from a rewrite — and unreviewable.
Case study: HTMX#
HTMX revives ideas that were standard in 2005: HTML over the wire, progressive enhancement, the server owning state.
Restoration or reinvention? It kept the model and replaced the implementation — which is exactly the Ship of Theseus question from last week, shipped as a library.
What you deliberately do not change#
The temptation is constant: refactor, modernise, improve.
Every improvement is a departure from the original. That may be correct — but it must be a decision you made, written down, not a habit you did not notice.
When restoration is wrong#
Some projects should stay dead:
- Superseded, with a maintained successor
- Insecure by design
- Archived with an explicit request
- Finished
Choosing not to restore is a legitimate finding.
Required reading#
Brandi, C. (2005 [1963]) Theory of Restoration, trans. C. Rockwell. Istituto Centrale per il Restauro / Nardini.
ICOMOS (1964) International Charter for the Conservation and Restoration of Monuments and Sites (The Venice Charter).
Before next week#
Restore one piece of your dig site. One — not the project.
Which part you choose is itself a finding: which part was worth saving?