OK I'll go with that. I'm not doubting that shaking with a large head space brings the wort to saturation quickly. It's just that once the yeast start picking up that initial 8ppm or whatever saturation is, with a stagnant wort, the only way it's going to get any more is from diffusion across the interface. That's not going to be very much once diffusion starts and the concentration gradient drops off drastically. With even a gently stirred wort, the concentration of O2 at the interface will be the same as it is everywhere else in the wort, dropping as the yeast consume the O2. This speeds diffusion as it increases the gradient. If it's stagnant, the interface remains near saturation with respect to the headspace, and diffusion is slowed. See Fick's 2nd law. Solutions are reasonably simple with simple, well-defined boundary conditions.
It's generally the O2 availability that limits the cell density (for a starter) in most cases, right? If you raise the saturation by injecting pure O2, say to 12 or 15ppm, even with a stagnant wort, don't you get more cells because of this? I thought that this was the case, and not just homebrewer heresy. If it was the case, then I was thinking that adding more during the drop from 8 to 0 would also create more cells. Not the case?
On another note, 8ppm in a litre is 8mg. 3L of 23wt% O2 air, with air at 1.23g/L, means there is about 850mg of O2 in the head space. I bring this up because it seems to me it puts to rest the idea of needing to bring any air into the head space from outside the container. You could re-saturate 100x with what is already present in the head space.