Is Rolling Your Own Entropy The New Standard? (Coldcard Q&A P1)

— The Coldcard random number generation (RNG) failure exposed a fundamental security flaw where the device relied on pseudo-random generation instead of true random generation, causing weak key derivation for years undetected.

— Simplicity and single-path execution are critical in security-critical code; the RNG implementation was overcomplicated with fallback logic and unnecessary handoffs between firmware components.

— Source-available code (publicly viewable) differs materially from open-source code (reusable and auditable by developers with incentive); Coldcard's source-available model lacks incentive for third-party security review.

— Users who rolled their own entropy using 100+ dice rolls generated cryptographically secure keys independent of Coldcard's firmware and face no immediate risk.

— Strong passphrases provide secondary protection (analogous to "poor man's multisig") but should not be treated as primary seed security; funds with weak passphrases warrant relocation.

— Coldcard can still be used safely as a signing-only device for imported seeds or for rolling dice entropy, rather than discarding the hardware entirely.

— If you generated a Coldcard seed using dice rolls (100+ rolls), no action is required; your entropy is independent and secure. If you used only a passphrase to protect weak Coldcard-generated entropy, migrate funds to a newly generated seed within a reasonable timeframe.

— Moving forward, treat hardware wallet entropy generation as a critical attack surface: roll your own dice (256 bits per ~100 rolls) or import pre-generated seeds rather than trusting device RNG, even on reputable devices.

— Prioritize truly open-source hardware wallets (where builders can reuse and audit code with commercial incentive) over source-available devices; conduct regular red-team testing with tools like Kammy to verify key generation pathways.