The presence of forks in permissionless blockchains is a recurrent issue. So far this has been handled either a posteriori, through local arbitration rules (e.g., " keep the branch which has required the most computational power ") which are applied once a fork is present in the blockchain, or a priori, via a Byzantine resilient agreement protocol periodically invoked by a committee of well identified and online nodes. In the former case, local arbitration rules guarantee that if they are correctly applied by a majority of the users of the system, then with high probability forks are progressively resolved, while in latter case, the sequence of Byzantine resilient agreements decide on the unique sequence of blocks to be appended to the blockchain. The question we may legitimately ask is the following one: To prevent the period of uncertainty inherent to optimistic-based solutions, are we doomed to rely on the decisions made by a unique committee whose members are already actively involved in the creation of blocs ? We negatively answer this question by presenting a solution that combines the best features of optimistic and pessimistic approaches: we leverage the presence of users and the " public-key as identities " principle to make users self-organize in small Byzantine resilient committees " around " each new object (i.e., blocks and transactions) to decide on their validity. Once validated, objects can be pushed in the network, appended to the blockchain without fearing any fork nor double spending attacks: we guarantee a " 0 "-confirmation delay. Additionally, our solution mitigates selfish attacks. We are not aware of any solutions enjoying such features.