How to Choose Guardians, and the Failure Mode Everyone Picks

The mechanism asks for a group of people, a threshold, and nothing else. Almost all of the risk is in who you pick, and the common mistake is not carelessness. It is choosing the people who are easiest to ask.

The failure mode

The people who are easiest to add are the people most likely to be in your disaster.

Your partner, who was in the same flood. Your colleague, whose company laptop was in the same theft. Two family members who both keep their recovery details in the same house you do.

A threshold protects against a guardian being unavailable independently. It does nothing against guardians being unavailable together, and the convenience that made them easy to nominate is usually the same proximity that correlates their failure.

So the question to ask about each name is not "do I trust them" but "what would have to happen for this person to be unreachable, and does that thing also happen to me?"

The two hard constraints, measured rather than assumed

At least three. The chain rejects a policy naming fewer, and the interface surfaces the floor so you do not learn it as a failed transaction.

Only confirmed, identifier-backed guardians count. An invitation that has not been accepted contributes nothing to your threshold. CONTROL that shows this follows from the data rather than from a rule somebody chose: an email-only invite has no identifier attached until it is accepted, so there is nothing for the policy to name.

The consequence people miss: a policy of five with two unaccepted invitations is a policy of three, and if your threshold was set against the five, you may be one absence from being unable to recover.

What actually makes a good guardian

Independence. Different household, different device, different way of being contacted. Geographic spread matters more than closeness.

Durability. Somebody who will still be reachable in three years and whose relationship to you is not contingent on a job, a lease or a romance.

Competence with the step. They will have to do something at a moment when you are stressed and they are confused. A guardian who cannot follow the instruction is a guardian who is not there.

Not all from one context. Three colleagues is one context. Two relatives and a friend from a different city is three.

Who not to choose

Anybody who is already a single point of failure in your life. If losing your house loses your phone and your paperwork, the person living in that house is not diversification.

Anybody you would not want to know you use this. Being a guardian reveals a relationship, and the list is a small social graph.

Anybody who would find a coordinated request hard to refuse. A threshold assumes guardians decide independently. Somebody who would go along with whatever the others say reduces your real threshold below the number you configured.

And nobody has done this for real here

No guardian has completed a recovery on this wallet. Not one, for anybody.

The mechanism is written and on-chain and the policy is enforced, and the part that has never been exercised is the human part, which is the part these guides are about. Treat your setup as untested until you test it, and testing it is deliberately its own guide.