If we want to implement a second order low pass filter, why we use
complex conjugate poles rather than two real poles? For instance, cascade
two RC filters with different time constants.
The complex poles will produce a flatter pass band followed by a sharper increase in attenuation in the stopband. At much higher frequency both types will have the same asymptotic roll off of 12 dB/octave.
Read up on topics such as Butterworth, Chebychev, and Bessel filter polynomials. All of these are the optimum method for a different given set of conditions.
Much of filter theory goes back to the days of copper conductor telephone circuits around 1920 - 1935.
"Modern" filter theory was invented by Campbell and Darlington of the Bell Telephone Laboratories in this time period.