A colleague noticed that PriorityBlockingQueue's internal lock is a fair ReentrantLock, when a non-fair lock would be much faster and usually "fair" enough. Is there some history here? Should we add a non-fair option for constructing PBQ? Martin