This circuit is a small signal sine wave generator.
R7 and R8 are used to control the amplitude of the output as desired.
After a few simulations, it looks like the gain of the Q3 stage is constant with the collector feedback bias resistance. Also, those R7 and R8 actually end up controlling the amplitude of the oscillation in Q1's base. I've got no idea why.
Please explain this to me. At first I thought those resistances were there to create a collector feedback bias, and such high values would limit the gain significantly, but it remains constant.