In the development of tire tread compounds, one often encounters two competing scenarios. First, fillers must be dispersed to the scale of a few hundreds of nanometers to obtain the largest window of mechanical reinforcement. Although at times, product developers are happy to accept much coarser scale of dispersion, e.g., mean particle size of around a micrometer. At this latter scale, the particles serve as defects. Second, particle networks experience periodic agglomeration and break-up due to large strains experienced by the tread, although otherwise, the compound performance is quite acceptable. This latter event gives rise to what is known as hysteresis and is the primary, responsible factor behind high rolling resistance. One may then argue that eliminating the possibility of particle network formation should also automatically reduce the hysteresis loss. However, a degree of particle networking is required for mechanical stiffness of the compound. These scenarios are discussed in this talk by considering several case studies, involving carbon back and silica-filled rubber compounds.