What kept sticking with me in SIGN was this:
two people meet, one of them submits “we met,” the other never confirms, and the protocol refuses to create the attestation.
That is where this gets serious.
Some claims are not like “I sent a payment.” They are mutual facts. We met. We agreed. We both took part. If one side can write that alone, the system is not preserving truth. It is preserving whoever got there first.
So the chain has to stay strict. Shared event happens. One side submits. The second side does not confirm. The attestation never hardens. Access stays locked because the second confirmation never arrived. No downstream app gets to act like the claim is settled evidence.
That is the Sign example that stayed with me. The record does not become usable just because one side was faster to write it down. It stays blocked until both sides confirm the same reality.
That design choice made the project feel more serious to me. A lot of systems can store claims. The harder thing is refusing to upgrade a one-sided claim into something other apps can trust.
That is where $SIGN feels useful to me. If apps build on these rails, they need both confirmations before a mutual claim can unlock action.
My question is whether apps built on top keep that discipline, or start relaxing it because one-sided attestations move faster.
Some facts should stay locked until both sides sign that they happened.