As one who has built both auto-bias and fixed (pot-adjustable)bias amps (EA230 etc.)over the years, I can tell you that the only advantage to fixed bias is low cost. This assumes that the auto bias is implemented properly. For auto bias, I have used both digital control (my EA2-150 circa 1976) and analog control with high-speed sample and hold techniques (ZH270). Either works well and holds the proper tube operating points with far better precision than the fixed bias approach. In the case of the ZH270, I even use the bias information developed from the auto bias circuit to tell the power supply shut down if tube parameters are out of spec.
But here is the sonic reason to control bias in typical push-pull amps. Non-gap output transformers used in P-P amps experience core saturation with as little as 1/4 mA current imbalance between the two halves of the push-pull. Toroid transformers, because of their tight coupling, can saturate at half this current. There is no way that hot output tubes will hold their bias to this tolerance without a good automatic controller. The sonic effects of mismatch over 1/4 mA are loss of transparency in the highs and increased distortion in the bass.
But here is the sonic reason to control bias in typical push-pull amps. Non-gap output transformers used in P-P amps experience core saturation with as little as 1/4 mA current imbalance between the two halves of the push-pull. Toroid transformers, because of their tight coupling, can saturate at half this current. There is no way that hot output tubes will hold their bias to this tolerance without a good automatic controller. The sonic effects of mismatch over 1/4 mA are loss of transparency in the highs and increased distortion in the bass.