Tommaso, thank you for your comment. It is good to see you in the contest again.
You have made a very astute point which is not lost on me although space constraints meant I did not say much about it in the essay. If you look at the paper by Seth Lloyd "The universal path integral" which I have cited you will see that he concentrates on algorithms. This may be very close to the kind of approach you are thinking of.
There are a couple of reasons why I did not want to focus on algorithms myself even though the universaltiy of computing is certainly very relevant here. One thing is that computers tend to run calculations forward in one direction. A normal sequential program has a time ordering in its calculations. This may be relaxed in a parallel programming architecture but there is still a partial ordering which people will inevitably try to link to temporal causality. As I said in my essay I dont think this kind of causality is important. I aknowledge that you could work with algroithms without making this connection.
The other thing is that path integrals are not just sums over configurations. When you have fermions the sum is replaced with an algebraic integral over grassman varuables. I suspect that the universality we see in maths and physics is also more general in this way. It is an algebraic principle that we do not understand and the kind of universality you get in statistical ensembles is just a good metaphor for that. The universality in computing is a little different again and it fits in somewhere, but here I tried to convey the idea of some other kind of self-referntial, self-organising universality that we understand very little about.
I look forward to reading your essay a little later