 The first such model I'm going to propose is called diffusion collision. Diffusion collision. And it's surprisingly simple and plain. The idea with diffusion collision is that I start out with some sort of chain, completely random. And in this chain I first form the secondary structure elements. Maybe a small beta sheet there. Once those elements have formed, this is going to be a reasonably fast process. Once those elements have formed, it's just going to be a matter of packing them into the exact protein structure. That might be a slower process, but since I've now already formed the secondary structure elements, it's not going to take forever. I don't have to test every single torsion in the protein backbone. And you know what? This is true. It's on cases. Not all cases, but there's one special case I'll show you. So I think I already showed you this simulation of the BB A5 protein that we folded many years ago in folding at home. Let's hit the button and see what happens when it folds. So with diffusion collision, the idea is that the secondary structure elements should form rapidly and independently. And here you see the helix forming, and within a few more nanoseconds here the beta sheet had also formed. It looked to be independent, right? Then they're going to flop around a bit and keep searching. And eventually the beta sheet will have formed a nice, stable, beautiful thing somewhere around 40 nanoseconds. And then we're happy to have the folded state somewhere there. That looked like diffusion collision, and it was diffusion collision. The way we could prove that is that checking if the beta sheet formation is dependent or independent of the alpha helix formation. That's actually very easy. We know that from mathematical statistics. The definition of two processes if they are independent is simply that the probability of having both the alpha sheet and alpha helix and the beta sheet compare that to the probability of having the alpha helix multiplied by the probability of having the beta sheet. In this particular case, it turns out that they are identical to the statistical accuracy we have, and that means that there is no interdependence here. The reason why this happens for this simple protein though, it's a super small protein that has been engineered to fold fast, and in general we know this is not the case. Just imagine if you had two beta sheets, they would be dependent on each other. So as beautiful as this is, and as excited as we were when we were able to show this some 20 years ago, it's not generally true. So we're going to need to look into other models.