Ljgj, I had planned to stay out of this discussion so as not to color the responses, but since you ask, I'll be brief. Since I don't design drivers, I have primarily done various designs with stock/modified drivers. I started out with multi-way systems with simple and complex crossovers. I studied diligently, believing that if I could just get the right drivers and the right..... So, then I tried 2 way designs, same story. Now don't get me wrong, these were all good sounding speakers. But, all the tradeoffs. Slopes, cascading phase shifts with varying slopes, overlap problems, beaming frequencies, signal losses, driver matching, impedance fluctuations at varying frequencies with the corresponding response fluctuations, different driver phase shift angles added to the capacitor/coil phase shifts, corrective circuits, zobel networks, information loss, etc,etc. At some point, I became aware that if the whole ball of wax was simplified, more information could get through to the driver, with more coherent structure, due to the lack of the above. I am not promoting that any system is perfect. However, I am standing on the premise that more musical information, in a more coherent form will yield a more natural presentation, even if some aspects such as frequency extremes and even some dynamics are somewhat limited by the nature of a single driver. The idea being that a usable frequency range of extremely well resolved information is better than a wider range that has lost some of the information, and muddled it. Add point source imaging and my choice became clear. This information retrieval/retention theory has been standard in analog source design for years.You cannot retrieve musical information that has been lost earlier in the signal chain. I think it also applies to all other aspects of the system including speakers. I can see that I am in the small minority in this belief, but that does not deter me, nor does it mean that I am not correct. Perhaps the perfect driver has not come along yet, but if/when it does, the theory will prove itself. A driver connected directly to the amp, with no intervening components other than the necessary wire is the least offending structure to the signal we are trying to listen to.