 I think another good way of thinking that is some sort of timeline of timescales. Going all the way from femtoseconds up to hours in this case, down in the femtosecond range we have physics. You can even go down to octosecond laser experiments. This is the realm of quantum mechanics and everything. In theory you should be able to integrate the time-dependent relativistic Schrodinger equation at some point. But the price for this is that you're stuck with very small, simple systems. You don't have any solvent around them. You can study maybe the vibration of the angle or possibly around picoseconds, the rotation around the torsion, the bond. Now the advantage of this is that in simulations you might have an extreme detail about these phenomena and they agree perfectly with the ab initio physical equations. The problem here is that I can't extrapolate this by 15 orders of magnitude even if I had the computer power. Because again the equations might not be so exact. The models might not be so exact that they will be valid if I extrapolate them by 10 to the power of 15. And in particularly if I have some sort of simplified model there are also issues about the parameters. Will they be valid up here? Now if I move to this other point up here or down there rather I have a ribosome. It's a beautiful piece of machinery in the lab. I can't look at the ribosome but I can certainly use simplified experiment to check that it's producing proteins. And I'm having dozens, hundreds, thousands, millions of these molecules which creates beautiful emerging but at the cost of not having the detail. I can no longer put a flag on the atoms. For a very long time these were separate worlds not talking to each other. But as computers and models got better and better we gradually moved up the ladder. So you can think of this as moving bottom up. So you're using physics but maybe simplified models. And somewhere here in the middle we might accept that we're having classical models not quantum models but you can study a simulation such as an iron flowing through an iron channel that would take roughly 10 nanoseconds or so. One of the fastest semi-biological processes. If you can simulate a microsecond you can see a small protein folding and larger proteins somewhere around the milliseconds. And this is a concept originally borrowed from Vijay Pan then. I think both Vijay and I used to say that at the time we were kind of at microseconds we would like to be at milliseconds and we would love to be at seconds. I don't think we need to be at seconds because what's happened at the same time is that the experiments have gotten better and better at resolving very fast phenomena. So today we can measure currents through iron channels with electrophysiology in the lab with the time resolution that's occasionally down to at least a hundred microseconds or something. So something down here. And that means that the top down approaches are now overlapping with the bottom up approaches. And that means that if our models overlap we can use both of them to study any concept you're interested in. And if they agree that means that we can explain these room temperature and room length scale processes all the way from the molecular detail down in individual atoms which is going to prove remarkably powerful.