Collaborative Working Sessions - NIST, Governments, and standards
NIST, Governments, and standards Notes
Jarl Gullberg
standards from what has been done for a long time vs industry standard goal: make it easy to implement, interoperate, verifiability gov: e.g. vuln-reporting ; usually abstract good: NIST SSDF, actionable, not overly specific: describes outcomes, model,…
what outcome? being able to reproduce binaries independently why? CIA + non-repudiation + traceability correctness, integrity, availability correctness: avoid build-time race-conditions integrity: make it tamper-proof traceability: because we are imperfect and need to improve after a failure availability: because r-b ensures artifacts can be produced later/elsewhere non-repudiation: because we need to know who did what
need to reduce effort to implement via anti-bikeshedding consuming 3rd party binaries, need trust. Attestation vs audit (effort/price) reference-implementations do help.
scope: Transformation from any source material through a set of tools into output artifacts, in a deterministic way
it is easier to add users to ?
=> need to find existing standards with users especially standards with consideration for security
- outcome
- guidelines (e.g. use hermetic environment)
- example implementations
ToDo: create an inventory of national standards aligning with r-b outcomes so that we can influence standards process + docs to create demand for r-b, so that investment into r-b happens as a result
formalizing conformance testing to r-b in ecosystems
standards provide a shortcut / framework for reasoning about outcomes. e.g. governments know they want resilient secure software for critical infrastructure, but they don’t know that they need r-b for that.
convergence towards common approaches
watch for and engage with emerging standards, become part of the working group (e.g. CRA(cyber-resilience), SLSA)
Add website docs for standards bodies / workgroup engangement / vendor partnering
Create our WG
previous summit: https://reproducible-builds.org/events/hamburg2024/ReproducibleSummit2024EventDocumentation.pdf page 44
Sketch page
Developers of software System designers Operators Security experts Auditors Researchers
Hermetic Determinism
Tamper proof (integrity)
Availability
Non-repudiation
Traceability (providence)
Interoperability
Correctness
Spec -> publish, advocate, engage Lobby -> NIST -> educate
Getting standardization bodies to care about reproducible builds
We care about standards because
- Standards are a framework for reasoning about outcomes
- They act as shortcuts to solution design
- They encourage convergence on common approaches
Reproducibility is a means to achieving these outcomes
- Tamper-resistance (integrity)
- Availability
- Non-repudiation
- Traceability
- Correctness
These are well-known terms of art within existing security and safety standards, coalesced around security and resilience.
The road to standardization
- We want to attach and map to existing standards, working top-down to document and align on de jure directives
- We want to normalize solutions in the ecosystem, working bottom-up to form de facto implementations
- We want to watch for and engage with emerging standards, such as the Cyber Resiliency Act
Next steps
- Make an inventory of national, international, industrial, and ecosystem standards to identify applicable areas of work
- Mapping reproducibility outcomes to requirements in applicable standards
-
Picking one and doing it!
-
SLSA is an existing standards body with a known focus on reproducibility
-
The CRA is a set of emerging standards with opportunities for advocay and adaptation
-
- Create a working group to drive the work