Actually I am rather flattered by the fact that an experimental design, which I did a couple of years ago, still is under discussion.
The good ole question about how much phase shift you need to run Polyakov mixers for I/Q software defined radio (SDR).
Have a look: http://draaggolf.blogspot.com/2010/01/30m-subharmonic-iq-sdr-receiver.html
Yep, this is a very simple concept design, it worked, it suppressed the lower side-band. And yes, it uses 45 degrees of phase shift at the oscillator frequency, which is half the beat-frequency (for reasons of using frequency doubling Polyakov mixers).
Please see the comments on why other designs, resulting in 180 degrees of beat-frequency shift, still seem to work just fine.
My design is not perfect at all, using a potentiometer to adjust a radio-frequency phase-shift is neither elegant, nor stable. Keep in mind, that this quick and dirty design was more a proof of concept rather than a production ready machine.