Once the swirling is gone, one still sees the speckle pattern move in a kind of slow twinkling pattern. The scattered light is indeed concentrated around 10 degrees from the beam. But the point of the demo is to account for the speed of the twinkling. One notices that it is slower for the scattered light that is closer to the beam.
The speckle pattern is coming from the random phases of the scattered light coming from the different blood cells. When the blood cells move, these phases change. How much movement is significant? The cell has to move about a scattered wavelength ~\lambda/\theta in order to affect the phase significantly. How long does this take? The cell is moving by diffusion. Its displacement x is given by x^2 \aboutequal \zeta t, where \zeta is the diffusion constant. Since x has to be about \lambda/\theta, this means that t ~ 1/\theta^2.
Now, that is something you can see in the scattered light. look at some speckle at one position on the wall. maybe it is coming and going about 4 times a second. Then according to the above, if you look at half of that scattering angle--- half-way toward the beam spot---the twinkling should be 4 times slower: 1 per second. If you look, this sort of works. And with no math it is clear that the twinkling gets a lot slower as one approaches the beam spot.