# kaal:position:2026-08-26-004

**Affirmed position.** Offline capability is an architectural test, not a feature label. Kleppmann and coauthors distinguish cloud applications from local-first software by the location of the primary data copy, the presence of a locally installed executable, and continued operation without a network connection. Their prototypes permit users to disconnect, continue working, and later merge changes. Applications centered on synchronous server interaction instead lock users out when the network fails. This independently extends Kaal's operational test: retained function under isolation supplies observable evidence that computation and state remain on the user's device.

The correspondence has limits. Local-first software addresses collaborative data applications, not agent runtimes or model inference. Offline editing does not establish that every relevant computation occurs locally, and later synchronization may still depend on external infrastructure. The paper nevertheless sharpens the classification rule. A local interface cannot establish local execution when the underlying capability disappears at disconnection. Network isolation converts an architectural claim into a falsifiable test.

**Status.** affirmed  **Published.** 2026-08-26

**Holds when.**

- The response is limited to the exact full-text propositions and the one mapped Kaal claim.
- External evidence level: peer-reviewed Onward! 2019 conference paper with complete public full text.
- Mapping review tier: independent substantive scholarly-growth extension.
- The paper studies collaborative data applications, not AI agent runtimes or model inference.
- Offline editing does not establish that every relevant computation occurs on the user’s device.
- The prototypes test continued offline work, but they do not define a general benchmark for all claimed local execution.
- Synchronization and collaboration may still use external infrastructure after connectivity returns.

**Current debate.** Local-first software: you own your data, in spite of the cloud: https://doi.org/10.1145/3359591.3359737

**Extends.** kaal:claim:7314479-004: https://wulfkaal.github.io/claims/7314479-004

**Scholarly basis.** Wulf A. Kaal, Institutional Requirements for Sovereign Local Agent Runtimes (2026). SSRN: https://ssrn.com/abstract=7314479

**Source PDF sha256.** `debace24a155ae924a155b1fafe98856d98cf83689feff2f87a32f1c06171ce6`

**Evidence level.** peer-reviewed Onward! 2019 conference paper with complete public full text

**Mapping review tier.** independent substantive scholarly-growth extension

**Mapping confidence.** 0.96  **Mapping ambiguous.** false

**Topics.** ai-and-agents, research-methods, decentralization, local-first-software, offline-capability, evidence-provenance

**Provenance.** Affirmed in kaal-review:2026-08-26:scholarly-growth-7314479-004-reviewed-v1 at https://wulfkaal.github.io/positions/by-claim/7314479-004.html.

**Record type.** This is a dated commentary position that extends a scholarly corpus claim. It is not a verbatim claim extracted from the paper.

**Canonical form.** This markdown file is the canonical hashed representation of the position.