Basing security on keys, not on secrecy about the mechanism
Open design holds that a system's security should rest on the strength of its keys and the soundness of its mechanisms, verifiable even by someone who fully understands how the whole system works — not on attackers remaining ignorant of the design itself. A design that can be published, reviewed, and still hold up is one whose security claim actually means something; a design that only works as long as it stays secret hasn't been tested against a knowledgeable adversary at all.
This has a practical benefit beyond philosophical tidiness: a design open to review gets its flaws found by defenders doing that review, rather than exclusively by attackers who happen to reverse-engineer or leak it first. Secrecy about a flawed design doesn't fix the flaw; it just delays who discovers it, and offers no guarantee that the discoverer is on the defending side.
Where obscurity still has a minor, honest role
This doesn't mean every implementation detail needs to be public, or that obscurity provides literally zero value — keeping some specifics private can add a modest layer of friction for an opportunistic attacker. The mistake open design warns against is treating that friction as the actual security, rather than as a small addition on top of protection that would hold up even if every detail became known tomorrow.