English is not my native language and I'm left wondering. Does he talk about the why he thinks the laws could change ? It sounded to me like a vague hypothesis but I'd love to understand better.
If I follow my own musing after watching this I would say time could be the path of least information in an isotropic adimensional space. The axis or path along which compression is the easiest. This would explain nicely why how our brain handles it in such a particular way. If Lee Smolin is wrong it would be the path along which energy cannot be absorbed and entropy can't increase as state relationships are static. Something like a visible effect of the strongest possible force. This could be a simple definition of time. Sadly I'm a layman and most probably completely off the mark.
What if the changes to the constants are random?
I think the idea is that, were we to have the state of everything at a moment, the laws we would use to predict the state at the next moment would be embedded within the state we had captured initially and so the prediction for the next state would have to include the state-change for the laws, for which we would need meta-laws to describe that change, which themselves would need to be embedded in the initial state, ad infinitum.
I think the only interesting idea to draw from this is that if laws change over space and time, all predictions are, at best, only probabilistic if the changes are continuous over space/time.
I can't see how this makes the future any less knowable[1]. It just adds more state.
___
[1] Obviously, we can't predict the future as we haven't the resources or the current state of everything. Naturalism just states that were we to have them, we could predict the future.