To protect our assets, we design systems to require multiple keys for access. Having multiple keys eliminates the single point of failure of a single key (one compromised key does not provide access). Having the keys distributed eliminates the single point of failure in a single target that holds keys (a compromised device holding some but not all keys does not provide access). We interact with multiple keys and distributed keys every day. Logging into an account with our username and password, hosts may prompt us for a code sent to our phones or approval through authentication applications. Multi-factor authentication (MFA) is a form of distributed key system, with factors including: “(i) something you know (e.g., password/personal identification number); (ii) something you have (e.g., cryptographic identification device, token); and (iii) something you are (e.g., biometric).”
Distributing the keys by factor-type mitigates the risk if any single factor is compromised, but MFA still has single target vulnerability because one person holds all of the keys. Distributing the keys among different individuals improves the security. This shows up in national security (nuclear keys), legal governance (multiple individuals sign to bind a company), software (multi-signature wallets), and elsewhere. Distributing keys eliminates the risk if one individual is compromised, though systems typically issue and distribute new keys following any compromise to ensure system integrity.
Software encryption methods can replicate distributed key systems using secret sharing algorithms (such as those developed by Shamir or Blakley) to divide data into a number (n) of fragments, which are then distributed to different parties. The algorithm allows any combination of fragments above some threshold number (m) to reconstruct the original data—requiring m of n keys. Like its physical key counterpart, this eliminates many single points of failure and is a secure method for protecting digital assets.
However, like its physical key counterpart, every system that relies on distributed keys is vulnerable at the endpoint when the necessary keys have unlocked access. Here, there is a single point of failure, and additional security must be considered to protect the endpoint, such as verification, authentication, and access timeouts.