Comment by StavrosK

I see, thank you. Can't NSQ do that, or at least get probabilistically close to it with multiple nodes (i.e. doesn't having more nodes reduce the probability of data loss)?