# One input stream, three shapes of the past — is branching undo a freestanding concept?

**URL:** <https://forum.essenceofsoftware.com/t/one-input-stream-three-shapes-of-the-past-is-branching-undo-a-freestanding-concept/308>\
**Category:** Concept catalog\
**Created:** [July 10, 2026, 3:21pm UTC](https://forum.essenceofsoftware.com/t/one-input-stream-three-shapes-of-the-past-is-branching-undo-a-freestanding-concept/308 "2026-07-10T15:21:59Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Scott](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.essenceofsoftware.com/scott/32/131_2.png) [@Scott](https://forum.essenceofsoftware.com/u/Scott)\
**Post date:** [July 10, 2026, 3:21pm UTC](https://forum.essenceofsoftware.com/t/one-input-stream-three-shapes-of-the-past-is-branching-undo-a-freestanding-concept/308/1 "2026-07-10T15:21:59Z")

</div>

## Following up on the concept-boundary conversations: we built a small interactive exhibit arguing that “undo” is three concepts, not one concept with options — **[Undoing Undos](https://gracecommons.dev/three-undos.html)**.

Same input stream, three engines, each keeping a different shape of the past and therefore making a different promise:

- **Linear undo** rewinds a timeline (replay-skip).
- **Compensating undo** reverses _effects_ on a timeline, using recorded compensating actions.
- **Branching undo** keeps every timeline — undo-then-diverge preserves the abandoned branch, so the past is a tree and the promise is _no state is ever lost_.

That difference in shape and promise is why we think they don’t merge.

In our catalog, two of the three are grounded patterns with machine-checked models:

- [Undo History](https://gracecommons.dev/compositions/undo-history.html)
- [Compensable Workflow](https://gracecommons.dev/compositions/compensable-workflow.html)

The third is only a registered candidate, and its routing is genuinely unresolved — so we’d like to run the freestanding test in public. Is branching undo:

**(a)** a freestanding concept,  
**(b)** Undo History composed with a branch-point index, or  
**(c)** not a record-level concept at all?

Our current hypothesis is **(c)**: record-space wants a single authoritative timeline — that’s the event log’s total-order invariant, and why the catalog never needed branching — while the tree lives in _draft-space_ (scenarios, explorations, agent backtracking) and collapses to lineage metadata when a branch **promotes** through the ordinary commit seam. Witnesses pointing that way: editor undo-trees, design-tool branching, sandbox environments, and AI agents forking working state.

We’re honestly unsure, and counterarguments would be more useful than agreement. Happy to be told the hypothesis is wrong.

---

<div class="post-metadata">

**Author:** ![Scott](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.essenceofsoftware.com/scott/32/131_2.png) [@Scott](https://forum.essenceofsoftware.com/u/Scott)\
**Post date:** [November 11, 2026, 11:27pm UTC](https://forum.essenceofsoftware.com/t/one-input-stream-three-shapes-of-the-past-is-branching-undo-a-freestanding-concept/308/2 "2026-11-11T23:27:45Z")

</div>

> An update on our own question, since nobody bit. We’ve since answered it ourselves, mostly as (c).
> 
> We retired Undo History. Our execution contract now requires the store to keep every committed version of a record readable, a temporal store in effect. Linear and branching undo of state both become reads over that history, so neither needs a concept of its own: a branch is just a past version you start from.
> 
> What a temporal store can’t do is undo effects that left the system: a sent email, a charged card. That’s why compensating undo stays a concept (Compensable Workflow), since it has to record how to reverse what can’t be rewound.
> 
> So in our catalog the three undos became one storage rule plus one concept. The exhibit is still up, now labelled an exploration. Counterarguments still welcome.
